Previously, when double-clicking on a table cell to edit it, there was a possibility of a race condition. A refresh of the display was requested, then the run loop was passed through once. But it was possible for the display refresh to not be queued when the run loop was passed through, in which case the refresh would cancel the editing. Or at least I think that's what was happening. ;-)
This commit (hopefully) eliminates the race condition by synchronously refreshing the layout and display.
Made some miscellaneous formatting fixes as well.
The changes in CPBox were messing with the subviews encoding. This patch restores the previous behavior of CPBox.
The original bug might be reintroduced, but this was definitly not the good solution
Previously a CPPredicate created with format `value == nil` was converted back to `value == <CPNull @ xxxx>`
This patch ensures `predicateFormat` returns `value == nil`
Test added in CPPredicateTest.j
The horizontal margin was larger than the vertical margin, and larger than the margin guideline for Cocoa.
Also one of the hint text lines was much closer to its textbox than the other two, and not well aligned within its textfield (should be right aligned).
Issue#2406 reported a behavior different from Cocoa Framework. This test validates that Cappuccino CPSplitView is working well when adding
and removing a subview. A CPSplitView that has only one subview should resize it to take the whole space available.
This PR adds a new test to Tests/Manual
Previously, adding a "Custom Formatter" (direct subclass of NSFormatter) to a cell in Xcode would cause a failure in nib2cib, because it didn't know how to deal with it.
This commit adds support for custom formatters in nib2cib (and thus in xibs), and updates the CPFormatter test app to demonstrate that this works.
Previously, when having a CPTableView in a CPBox, nib2cib failed due to a superview not defined yet. This was raised because in nib2cib we used the method setFrame on the CPBox to modify the frame of the CPBox (we need to modify this frame because the sizing difference between cappuccino and cocoa). We now directly modify the attribute _frame from the object and update the bounds in the same time.
This PR fixed another bug as well. Previously the contentView of the CPBox was not encoded, we now encode it. This allows us to get a full CPView object for the contentView, previously we got a weird (I have no idea how this object was created though...) object CPView with some missing attributes. This raised a crash because _trackingAreas or _themeState were not defined in this object.
Fixed#2400
When the predicate part of a subquery expression was containing
variables, the substitution was ignored.
Now, the subpredicate can contain variables that will be substituted
when calling evaluateWithObject:substitutionVariables:.
Added expression test and predicate parsing test.
For some (yet unknown) reason, when declaring a view with tracking area
in a CIB, there’s some weird things happening that try to use the
_trackingAreas array when it’s not yet initialized… So, replacing
_trackingAreas.length by [_trackingAreas count] resolves the problem.
Previously, the method CPBundle loadCibNamed:(CPString)aName owner:(id)anOwner was broken, it loaded the cib asynchronous instead of synchronous. This was due to bad refactoring of a previous commit refs#c5b250236fa662e3da04e442f4414f95d64d9308
This PR refactor this piece of code and make works the method synchronous as it should.
More information here : https://github.com/cappuccino/cappuccino/commit/c5b250236fa662e3da04e442f4414f95d64d9308#commitcomment-14109453
Previously, we could only give a selector and a target for perform in the runLoop. Now we can give a block.
This feature is used in the CPTextField class. Previously, when we wanted to call a function at the end of the stack we used window.setTimeout, however due to HTML5 specifications this wasn't called just at the end of the stack but at least 4ms (more information here https://developer.mozilla.org/en-US/docs/Web/API/WindowTimers/setTimeout#Minimum_delay_and_timeout_nesting). Now we give a block to perform, and this block will be performed in the next runloop.
Unittest has been added in Tests/Foundation/CPRunLoopTest.j
Removed the ’release-inline’ option when building as it is not practical to have two different options. The release build now always has inlined msgSend functions. If a release build is needed without inlined msgSend functions an edit of the ”-O2” option to ”-O” in the Jakefile is needed.
Previously, CPImage -size returned a reference to the internal _size object, which allowed the caller to directly change _size. In Cocoa a copy of the size is returned.
This commit returns a copy of _size to ensure no unwanted side effects.
This PR adds the feature of CPNotificationQueue.
Cappuccino provides a framework for sending messages between objects within
a process called notifications. CPNotificationQueue objects (or simply notification queues)
act as buffers for notification centers (instances of CPNotificationCenter).
Whereas a notification center distributes notifications when posted,
notifications placed into the queue can be delayed until the end of the current pass through the run loop
or until the run loop is idle. Duplicate notifications can also be coalesced so that only one notification
is sent although multiple notifications are posted. A notification queue maintains notifications
(instances of C¨Notification) generally in a first in first out (FIFO) order.
When a notification rises to the front of the queue, the queue posts it to the notification center,
which in turn dispatches the notification to all objects registered as observers.
More informations here :https://developer.apple.com/library/prerelease/ios/documentation/Cocoa/Reference/Foundation/Classes/NSNotificationQueue_Class/index.html#//apple_ref/occ/instm/NSNotificationQueue/enqueueNotification:postingStyle:coalesceMask:forModes:
Unit-Tests in Tests/Foundation/CPNotificationQueueTest.j
Previously, when opening a CPPlatformWindow a crash happened when this new window was blocked by the browser (addblock, or browser feature).
Now when opening an external window, we check if the DOMWindow has been created.
This PR has modified the method isVisible of CPPlatformWindow. We now check if the _DOMWindow is not NULL and undefined.
In some places the objj_msgSend function is called directly. Most of the times the old slower version is called. This commit will use the never faster version instead.
NEW: CPURLConnection -operation method, CPError CPURLErrorDomain constant.
CPURLConnection -operation gives access to the operation generated by the new CPURLConnection creator.
This allows to create dependencies between operations and setup priorities early.
When the connection fails with a status code 404 or is cancelled, the
operation is also cancelled and
the next operation in the queue is started.
For example the method ’initWithName:’ will return different abbreviation depending on the order CPDictionary returns keys from the method ’keyEnumerator’. As this is undefined the answer can vary. Test cases now handle all cases.
Previously, the CPTabView did not react with the first mouse event (for instance when a CPTabView was in a not focused window). Now it does as in cocoa.
To do that, the CPTavView override the method acceptsFirstMouse. Previously, the tabView had a CPBox, it now has a _CPTabViewBox. This new box reacts differently than the CPBox for the method hitTests. It now returns nil when the click was performed on the segmentedControl of the CPTabView, previously this same click would be handle byt the CPBox.
Previously, when a view was about to layout, only layoutSubviews was called. Now the methods viewWillLayout and viewDidLayout are called as well. This is like in Cocoa.
This PR fixed the issue with the appearance and the view based tableView. We now handle everything related with appearance when making the layout of a view.
This PR removed a small optimization in CPView. Now when making setNeedsLayout and setNeedsDisplay, the methods drawRect: and layoutSubviews are always called even if there are not override. (As in cocoa)
Previously, when using a viewBased CPTableView Cappuccino just crashed.
This was due to the new appearance feature. Indeed, at the end of the method initWithCoder: of CPView, we set the appearance of the current view and the themeStates of the view and its subviews. For an unknown reason some of these subviews were not completely initialized (see comment line 3742), this made crashed cappuccino because _themeState was null.
To fix this issue, we send the method setAppearance when the current stack was performed. (Magic setTimeout...).
This is for sure not the ultimate best fix, but it's prevent to have a broken master. We will need in the future to fix this issue in a better way.
The purpose of this feature is to create theme for every object of the AppKit. Previously, we could not theme object like CPColor, however CPColor could be very interesting to theme, value like alternateSelectedControlColor were hard coded in the framework. With this PR, we can now theme a CPColor and take some values from the current theme of the application. Then, this PR offers the possibility to refactor the themeDescriptors, some classes contains theme attributes for other class (CPTabView and CPTabViewItem for instance).
How doe it work ? A new category _CPObject+Theme.j has been added to the AppKit. This category contains every theme methods needed to theme an object (previously this category was in CPView). We add utils method like _encodeThemeObjectsWithCoder: and _decodeThemeObjectsWithCoder: in this category to be able to code and decode easily a coder for every object (this is generic).
This PR did not refactor the themeDescriptors, it only add this new mechanism for the methods alternateSelectedControlColor and secondarySelectedControlColor of CPColor. We will need to refactor that gradually
Previously, when making a right click and when a cappuccino menu was not displayed, the system menu was displayed. Now we don't display this menu anymore as in cocoa. We only display when making a right click on a CPTextField.
Previously, when opening a popover, the animation was blinking on Chrome and Opera.
To resolve (temporary I hope) this issue, we launch the last part of the animation at the beginning of the next runloop. We do that by wrapping the desired code in window.setTimeout(function(){},0).
A google chrome issue has been opened here : https://code.google.com/p/chromium/issues/detail?id=523044&thanks=523044&ts=1440095724
Each default (and only supported) CPAppearance now have an associated
theme state. As the appearance is correctly propagated in CPViews
hierarchy, it is now possible to have dark controls when the an
ancestor view has a dark appearance.
Added automated tests
This patch contains a very naive implementation of the
NSVisualEffectView. This will only work on very recent unreleased
version of Safari, but should be supported by all at some point. This
allow to use the Yosemite/iOS blurry effect.
Not all options are supported (especially the
`CPVisualEffectBlendingModeBehindWindow` mode…). But hey! it’s a start
:)
Tests in Manual/CPVisualEffectViewTest
This patch adds the `CPAppearance` class and uses it where needed.
There is nearly no impact, but in `_CPPopoverWindowView` where the
`CPPopoverAppearance` needs to be converted to a `CPAppearance` object
nib2cib support has also been added.
*This patch brings no new functionality.*
This PR adds the possibility to set the timeout and the cache policy of a CPURLRequest.
For that, there are new things in the framework :
- CFHTTPRequest has now the functions setTimeout(), getTimeout() and isTimeoutRequest
- CPURLRequest has now the methods +requestWithURL:cachePolicy:timeoutInterval: and -initWithURL:cachePolicy:timeoutInterval:
- CPURLConnection does not call the delegate didReceiveData: when a request has timed out. I will call the delegate didFailWithError:
- Cache is now possible for a CPURLRequest, only CPURLRequestReturnCacheDataElseLoad, CPURLRequestReturnCacheDataDontLoad and CPURLRequestReloadIgnoringLocalCacheData are supported. By default CPURLRequestReloadIgnoringLocalCacheData.
- Default timeout has been set to 60sec as in cocoa.
- A request is considered as timeout when there isn't any response (text/xml/type) and when the status of the response is 0.
When parsing a predicate format that contains a constant boolean expression
(e.g., 'key = YES') the value of the returned constant value expression has
the wrong type. Instead of using native types as values, the parser uses
[CPNumber +numberWithBool:].
This commit changes the format parser (CPPredicateScanner) to use native
booleans for constant value expression.
There's a test case that checks the parse result and an additional test
that ensures boolean expressions evaluate as expected with different kinds
of objects. The latter test passed before the fix but was added to ensure
compliance with Cocoa.
XcodeCapp 4.0 is a major release of our beloved tool. In a nutshell it:
- allows to manage multiple projects simultaneously
- allows to follow operations and cancel them
- has a per project Error and Warning reporting
- supports capp_env as you can define additional paths and objj include path per project
- much more things
Previously, CPTableView was listening to all end editing notifications in cell-based mode.
This caused a crash if the textfield is an actually subview of a dataview.
This patch ensures to listen only editing did end notification for direct subview of tableView.
This patch adds support for the new way of defining the default font from Xcode 7 beta. The patch also supports the old explicit 'Lucida Grande' of previous version.
Previously, the selection of a textField did not work as expected when the user did a right click or a drag even if the label was set to none selectable.
Now, with the css style user-select, a textField can only be selected if it is selectable and enable.
Credit to @primalmotion
Previously, every AppKit tests used the same sharedApplication. Due to this implementation, a could not pass because a previous test made failed the current test. For instance, a test could fail because the window of the previous test resigned (just imagine a new window is the key window in the current test), and this resign could raise an error. The error was displayed for the current test thought this test was perfect !
We now instead of using sharedApplication create a new CPApplication per unit-test file in the class method setUp.
Previously, the unittests for the CPDatePicker broke the tests suite of cappuccino. A CPDatePicker uses a CALayer, and the CALayer display methods wasn't wrap in a #if PLATFOrM(DOM). Now it does, so during the tests we do not try to access do the element document anymore.
This PR fixes also some warnings when compiling.
This PR adds the delegate methods applicationShouldTerminate and applicationShouldTerminateMessage in CPApplicationDelegateProtocol.
The delegate applicationShouldTerminate does not exactly work as in Cocoa. In Cocoa, this method is called in CPApp -terminate, but in cappuccino it is called in the method onbeforeunload of the window. In JS, this is the only time where we can prevent to reload the HTML page. If the developer cancel to reload the page, the browser will ask the user if he wants to reload or not the page thought (natural behavior of js).
The method applicationShouldTerminateMessage allows you to define what will be the text displayed in the confirmation alert.
Test app in Tests/Manual/CPPlatformWindow/
Added the method closeAllPlatformWindows in CPPlatform. This platform close all platform windows of the application (except the main one). This method is now used when reloading or leaving a cappuccino application, the application will now close every external window openend by the application.
This PR adds the features CPUserNotification and CPUserNotificationCenter in Foundation.
The CPUserNotificationCenter allows you to send user notification to the system.
Right now, we only propose what the W3C proposes. We can set for a notification the title, informativeText and the icon.
We only support local notification yet.
The protocol CPUserNotificationCenterDelegate has been added as well.
Test app in Tests/Manual/CPUserNotificationTest
Previously, when making objj2objcskeleton on a category of a cappuccino class, the cocoa category was not a cocoa class but still a cappuccino class. For instance, a CPView category was still a CPView category instead of being a NSView category.
Fixed#2344
Previously, when having a CPTableView cell based, the themeState of the textField were wrong. The text color was black as every textField was considered as editable. The CPTableView handled differently the textField, as long a textField is not editing, the textField is mark as non editable. Previously, the textFields were considered editable all the time.
More informations here : https://groups.google.com/forum/?fromgroups#!topic/objectivej/zImy4sj0Xz4
Previously, it was not possible to specify the desired output format of the command objj.
Now we can have either the default format or a xml format if --xml-output-format is added.
Previously, the method documentVisibleRect did not return the expected result. An issue happened when using this method on a scaled view. The converted rect was converted in the wrong context.
Unit test in Tests/AppKit/CPScrollViewTest.j
Fixed#2356
The command objj has now the option -c or --objc to generate objective-c files. The first arg is the file and the second the destination folder.
The objective-c class generated are basic and only contains IBOutlet and IBAction. To do that, the objective-c-parser is used. This parser is also used in xCodeCapp.
This PR adds the support for baseWritingDirection in CPControl.
Support of nib2cib has been added as well.
Test app in Tests/Manual/CPTextFieldEditingStyleTest/
Previously, when popen was called, the 3 streams (stdin, stderr and stdourt) where never closed. This was the reason of the having an impossible number of open files.
This patch ensure all streams are closed after using them. This means that the ulimit trick is no more necessary
Previously, the nextValidKeyView was wrong when the views were in a scrollView. Cappuccino did not take in account the possibility of scrolling, the lowest views were considered as outside of the platformWindow.
Now, it works as cocoa! The nextValidKeyView is the good one in a scrollView.
This pull request fixes another issue as well. A CPTextField can now become firstResponder even if the textField is not visible (as in cocoa). Previously a jump of the (html)window occurred to the textField. Now, cappuccino will internally scroll if needed to the element, focus it and then go back to previous scrolling position.
Previously, when clicking somewhere else in the system, cappuccino did no raise the notifications willResignActive and didResignActive of the application. As well cappuccino did not raise the notifications willBecomeActive and didBecomeActive when the user was back on the application.
Now it does !
Test app in Test/Manual/CPPlatform
Previously, when jumping from a platformWindow to another platformWindow where popover where opened, the wrong windows became key and main. This things occurred weird behavior of the platformWindows.
Now, it works as in cocoa, canBecomKeyWindow, becomeKeyWindow, becomeMainWindow, resignKeyWindow and resignMainWindow are called in the good order.
Previously, when having several platformWindows, jumping from another platform to another platform did not update the good key window of the application, specially when the expected windows was a panel.
Now it does. The CPPlatformWindow keeps a reference to the previous keyWindow when the window browser is about to blur. The we use this reference to update the keyWindow of the application.
The -tile method now leaves the frame height unchanged instead of
sizingToFit the theme attribute « min-size ».
The « min-size » is still in use when we layout the ephemeral subviews.
This is how other controls with a min-size work.
Also added -minimumFrameSize. This CPControl subclass will return the
frame for sizeToFit, i.e. when you want to sync the frame attribute
with the minimum/visible height.
Tests : CPSegmentedControlTest & Nib2CibAlignment manual tests.
Fixes: #2341
Previously, when switching from a platformWindow to another platformWindow or to another application, Cappuccino did not behave as Cocoa, Cappuccino still considered that the window was key and main.
Now it works as in cocoa, the platform window lost its focus and is not the keyWindow and mainWindow of the cappuccino application.
Test app in Tests/Manual/CPPlatformWindow/
Before this commit, the tabViewItems and the segments were 2 different
collections. Any indexed change in the tabViewItems asked to re-sync
the segments making it difficult to maintain any persistent state. Now
the tabView items are the segments content.
Also fixed the selection update when items are removed.
See CPTabViewTest.j and manual test Manual/CPTabViewNib/
changes for fix#2332. The test-cased failed, because now a
a CPDecimalNumber has a different UID compared to a CPNumber with the same
value. That means isEqual: will return false. In the test-cases the formatter now
generates CPNumbers to make the two Numbers compareable via isEqual.
Previously, CPNumber and CPDecimalNumber shared the same method UID and CPNumberUIDs-Dictionary.
This leads to some unexpected errors for example a [CPDecimalNumber zero] which refers to itself after
decoding. CPDecimalNumber now overwrites the method UID and has its own UID-Dictionary.
Fixes#2332
CPSearchField were not properly aligned in Aristo2 theme.
This fix uses the nib2cib-adjustment-frame information to align it.
It also improve the Nib2CibAlignement test.
CPSegmentedControl
FIXED: widthForSegment: and frameForSegment: now return the correct values when the segments sizeToFit
Added indexed accessors for the segment object
Improved tile performance
With manual and ojtest
Previously, when reloading a CPTableView the run loop was explicitly call to layout the tableView. This is not the case in Cocoa.
You can call several times the method reloadData and this will only lay out the tableView one time.
If the segment count is set to 0, the [self selectedSegment] now
correctly return -1 (like cocoa).
If the segment count is set to n from 0, the first segment is selected.
See manual test.
Previously, the var clickedRow and clickedColumn were only updated with a doubleClick.
Now clickedRow and clickedColumn are updated in the scope of a trackMouse as in cocoa. Once the method stopTracking:at:mouseIsUp is called, clickedRow and clickedColumn are set to -1 again. This var can't be used outside a user event.
Previously, when closing a platform window using the browser close
button, the represented `CPWindow` was not correctly closed.
This patch ensure `-(void)close` is called correctly by using the DOM
`unload` event. Also, delegate method `- (void)windowWillClose:` is
correctly called.
Added @ignore directive for private methods.
Changed a parameter naming
Moved some layout code from indexed accessors to public method.
Accessors should only deal with the model, not the layout.
Only populate the tooltips handler functions when necessary (when there is a tooltip and when the view is in a window) or clear them otherwise.
Tests in /Tests/AppKit/CPViewTest.j
Previously, the default template generated by capp gen for a new themeDescriptor project was not similar as the themeDescriptor of Aristo1 or 2.
This could confuse new developer who would like to create a new theme.
Added manual test showing fix for setLabel:forSegment for flexible
segments.
Added ojtest showing that the selection is preserved when a selected
segment is removed.
These methods are not useful with the current public API but necessary
if we want to implement CPTabView insertion/deletion methods and the
CPContentBinding.
Now tileWithChangedSegment: invalidates the frame of changed segments
and the following segments on the right. Then we just ask for the frame
of the last segment which will recompute all the invalidated frames and
give the total width of the container.
Added -intrinsicContentSize : the container size based on the sizes of
its segments.
Previously, widthForSegment was returning the current width of a
segment.
Now we separate the 2 concepts: the segment width can only be set
explicitly with setWith:forSgment: or in IB. If the width is 0, it
means the actual frame sizeTofit. In this case, we compute lazily the
frame and frameForSegment: return the actual frame.
The frame is cached and can also be invalidate by setting it to a zero
frame, for example when the content of a segment changes and we need to
recompute it.
Use frameForSegment: in the code when we mean to get the real width as
opposed to the declared width.
Previously, the CPApplication dispatched the current event when having the auto complete menu of a CPTokenField opened.
This occurs weird behavior. For instance when hitting enter on the menu and having a default button, the action of the button was triggered.
Now, the CPTokenField handles the key enter when the autocomplete menu is opened.
Before this fix, the CPTableColumn binder was not correctly reloading
the table when the number of rows changed. Now we reload fully the
dataviews when the number of rows changes. If no rows are
inserted/deleted, there is an optimization: we just need to reload the
objectValues and leave the dataviews untouched.
With Test in Tests/AppKitCPTableViewTest.j
Fixes#2317
This PR adds the possibility to layout or not a CPView.
Previously, once setNeedsLayout was called on a CPView, it wasn't possible to cancel the layout of the view.
Now we can as in cocoa. The method setNeedsLayout will still work (it calls the method setNeedsLayout: with YES).
UnitTests in Tests/AppKit/CPViewTest.j
This PR adds the method viewAtColumn:row:makeIfNecessary: in CPTableView.
This method first attempts to return an available view, which is generally in the visible area. If there is no available view, and makeIfNecessary is YES, a prepared temporary view is returned. If makeIfNecessary is NO, and the view is not available, nil will be returned.
An exception will be thrown if row is an invalid row index and if column is an invalid column index.
The returned result should generally not be held onto for longer than the current run loop cycle. Instead they should re-query the table view for the row view.
UnitTests in Tests/AppKit/CPTableViewTests.j
This PR adds the KVC operators unionOfObjects, distinctUnionOfObjects, unionOfArrays, distinctUnionOfArrays, distinctUnionOfSets for array and set.
Unit-Tests Tests/Foundation/CPKVCArrayTest.j
Unit-Tests Tests/Foundation/CPSetTest.j
Note: It seems that CPThemeStateKeyWindow cannot be tested in the
console. A manual test with multiple windows and table views exists in
Manual/TableTest/OldTest/
Previously, when the frame of a superview of the targetedView would update, the popover did not update its position.
Now it does by observing each frame of each superviews of the targetedView.
This patch provides a new option to capp. You can do
```objj
capp gen [-f] [-l] -T CustomTheme1 -T CustomTheme2 MyProject
```
This will copy/symlink custom theme(s) from your CAPP_BUILD into the
current project’s `Resources` folder
The DOM_OPTIMIZATION flag was not used by any of the current compilers. This resulted in unreachable code in CPDOMDisplayServer.h. This commit removes this code.
Previously the method `CPDOMDisplayServerSetStyleBackgroundSize` was named `CPDomDisplayServerSetStyleBackgroundSize`, which was inconsistent with the names of all other similar methods in the CPDOMDisplayServer.h file. This commit fixes this and updates all instances of its use in CPView.j.
Previously the showcase of the themes were broken. This was due to a change of the CPCollectionView, previously we used to use the delegate collectionViewDidChange. This delegate method has been removed. We now use an observer on the selectionIndexes.
Previously the CPStepper class returned the `CPStepper` type instead of the `id` type. This caused warnings to be raised when initializing subclasses of CPStepper.
This commit fixes this issue by specifying the `id` type in the return values.
Fixes#2293
Previously, when xCodeCapp opened the errors & warnings panel, xcc was always the new active application of the system.
Now an option is available in the preferences of xcc to disable this possibility. If disable the errors & warnings panel will open itself, but your favorite text editor will still be the active application of your system (you will keep the focus).
Previously, the button + and - of a CPPredicateEditor were blurred for Aristo2. Now they are nice and not blurred anymore.
This PR also add a second state for the button, the button will be darker when clicking on it.
Fixes#2183
Previously, the default size of a CPCheckBox created with the constructor method checkBoxWithTitle did not take in account the size of the checkBox view.
Now when calculating the size of the control, it will firstly calculate the size of the title and then add the size of the checkBox view and the offset.
Fixes#2281
This commit adds the `docset` command to jake, to create a Docset from a doxygen build. It adds a shell script to the "support" directory that manages the build, and includes an Info.plist file that gets copied to the Docset when built.
FIXED : After an operation started, either from a queue or explicitely, its finished property was never set to true.
This was causing dependant operation to never be executed.
After this commit, cancelled operations are started (but not executed) and correctly marked as finished.
Tests: CPOperationQueueTest and CPOperationTest for an operation managed by a queue or run explicitely.
Previously, the CPDatePicker took the key code CPReturnKeyCode all the time. It means, the key event wasn't redistributed to the other component, for example a defaultButton in a window.
Now instead of returning YES, it returns [super performKeyEquivalent:anEvent]
This commit adds a manual testing application for checking Cross-origin behaviour in different browsers. A README file included in this application provides details on how to run the tests.
Lastly, we removed the usage of max-size to enable some controls like CPTextField to have
a specific height. This may caused a CPTextField to have a wrong height.
As all controls in a CPPredicateEditor have a CPControlSize to small, we now force their height to the minimum size.
Previously the option of the observer in _CPPopoverWindow were set to 0, now it's set to CPKeyValueObservingOptionNew.
This PR fix also another issue, the popoverWindow registered the observer in the wrong order. It firstly observed the frame of the sender and then removed it directly. Now it register when ordering front the popover and remove it when closing the popover
Previously it was not possible to deep-copy a CPURLRequest, which meant that a CPURLConnection could not keep track of both the original and modified versions of a request.
This commit adds the copy method to CPURLRequest.
Previously, CPURLConnection did not implement the originalRequest and currentRequest methods (added to Cocoa in 10.8).
This commit adds this functionality.
NB: this is dependent on CPURLRequest having deep-copy capabilities to make a copy of the original request, which has been added in another commit on this pull request.
Previously, the withCredentials parameter was accessed primarily through the CPURLConnection, which gave limited access to changing the request before it was sent off.
This commit removes withCredentials from CPURLConnection and places it on CPURLRequest so that it may be more easily modified prior to sending the request. This also simplifies the logic in CPURLConnection for creating and establishing connections with credentials.
In this commit, all CFHTTPRequests are assumed to not use withCredentials unless they are explicitly set. Setting [aURLRequest withCredentials] will set the withCredentials property on the underlying XMLHTTPRequest prior to the connection being opened.
Previously, when assigning a listDelegate to a CPComboBox, we never deleted the observers added by this method. Now we add and delete these observers in the method addObservers and removeObservers.
This PR fix another issue. Previously, the behavior of the panel of the comboBox wasn't the same as the one in Cocoa. Now, when closing the panel, the hit view won't be the first responder as it was. Cocoa works like this as well.
Previously, a z-index value of -1000 caused some browsers to not display certain UI elements correctly. Setting this value to 0 instead seems to fix this.
This commit changes the default value in the templates generated by `capp gen`.
Fixes#2232, Refs #2194
Previously, the CPClipView registered the documentView to the notificationCenter as an observer to be notified when the frame/bounds of the documentView change.
Now, we don't use the notificationCenter anymore. The class CPView send a message to its superview (when it's a clipView) when the frame or bounds change. This fix the memory leak of the CPClipView
Fixed#2264
Previously, the CPDatePicker didn't have a focus ring when an user was editing it.
Now it does. It works for Aristo1 and Aristo2.
Test App in Tests/Manual/CPDatePickerTest
Fixed#2263
Previously, the CPDatePicker didn't have a focus ring when an user was editing it.
Now it does. It works for Aristo1 and Aristo2.
Test App in Tests/Manual/CPDatePickerTest
Prevously, when removing a CPTableView or a CPScrollView, the CPNotificationCenter kept a reference of these observers in the notification center.
Now, the CPNotificationCenter does only have a observer when necessary.
Previously, the method addObserverForName:object::usingBlock: didn't exist in Cappuccino.
Now it does.
To observe, you need to use the method - (id <CPObject>)addObserverForName:(CPString)aNotificationName object:(id)anObject usingBlock:(Function)block
To unregister observations, you pass the object returned by this method to removeObserver:. You must invoke removeObserver: or removeObserver:name:object:.
Test /Tests/Foundation/CPNotificationCenterTest.j
Fixed#2259
When setting controlSize, CPTextfield height was forced to min-size height.
Now we check that the frame size is between the min and the max size. It allows to constrain
a CPControl to have a fixed height by specifying a min-size equivalent to the maxsize.
In CPTextField case, we only have a min-size which enable us to have a specific size.
Changes has been made in Aristo2 only.
This patch adds support for CPTextField with multiline content. It uses
a DOM textarea when needed, instead of DOM input.
It also disable the spellchecking, fixes a bug where the input element was not
correctly resized when the frame changed, and refactor some code
Previously, when moving from platformWindow to another platformWindow, the targeted window didn't get the new event because the keyWindow of the application was still set to the other platformWindow.
This PR fixes another issue about opening panel in a platformWindow. When opening a panel in a platformWindow, cappuccino set the new key window to the wrong platformWindow.
Previously the Jakefile template for the capp utility would emit a template where the "configuration" build environment parameter could have multiple values. At the head of the file, the configuration parameter could be an environment variable composed of ENV['CONFIG'], ENV['CONFIGURATION'] or ENV['c']. It would default to "Debug".
However, other locations in this file did not use the detected configuration, and only looked at ENV['CONFIGURATION'] for the environment variable.
This commit fixes this by using the detected configuration environment variable as the build environment.
Previously, ObjJ was ignoring unknow ivar type. This patch adds some check to ensure the type is either a known class, the current class
itself, a global, a basic JS type or a declared custom type.
In order to declare custom types, this patch introduces the @typedef keyword.
For instance, this will throw a warning:
```objj
@import <Foundation/Foundation.j>
@implementation MyClass: CPObject
{
NUSuppaType mode;
}
@end
```
This will not:
```objj
@import <Foundation/Foundation.j>
@typedef NUSuppaType
@implementation NUMyClass: CPObject
{
NUSuppaType mode;
}
@end
```
Declared types are shared accross all application, one type can only be declared once.
Previously, the notification xib converted didn't show up. Now it does again.
This PR fixes another issue. Now when checking the the compilation issues, we only pass objj file to the command objj.
The Doxygen configuration file caused warnings to appear about obsolete tags. (The version of doxygen that the file was created for was 1.7.1, which was released in 2010.)
This commit uses the automatic doxygen updater (doxygen -u) to update it to a more recent version (1.8.7).
Additionally, this commit changes the .doxygen file to build a docset by default. The effect of this is to simply emit a `Makefile` in the output directory that can be further used to compile a docset. Parameters for the project publisher and name were added for this docset.
Previously, the text cursor of the CPTextField didn't take in account the content-inset. This raised issue with class as the CPComboBox, when the mouse was over the popUpButton, there was still the text cursor instead of the default one.
Now it works as in Cocoa.
This pull requests adds a nice feature to xCodeCapp, now each time a fill is processing, xCodeCapp will launch the command objj to check the compilation issues. If it finds something, it will show them up on the classic error panel. This basically show import warnings, warnings you don't catch with your browser but only during the build of the application.
Objj is only launched if no errors was found after the first parsing (the parser which translate a file to objective-c).
This feature can be disabled on the preferences panel.
The menu has a new menu item "Check Compilation Issues". This action will launch objj on each files of the project and shows warnings and errors found.
The command objj will take automatically the OBJJ_INCLUDE_PATH in the index.html or index-debug.html. This options can be disabled in the preferences panel as well.
This PR only works with the PR #2248
Previously, when setting font, alignement, intercellSpacing or itemHeight, CPComboBox was assuming the listDelegate was already created.
This patch stores the values in CPComboBox if the listDelegate is not ready, and apply them once it is.
This pull requests adds the following feature for the command objj:
- Option -h or --help to get the help of the command
- Option -I or --objj-include-paths to specify the frameworks to be used
- Possibility to pass several files to the command, for example objj AppController.j Test.j
This patch allows user to disable the commonJS build phase by setting the system environement variable IGNORE_ENV_COMMONJS.
For instance `export IGNORE_ENV_COMMONJS=1`
The constructor initWithWindow: initializes a new CPPlatformWindow and set the given CPWindow as a fullPlatformWindow and bridgless window.
When using this constructor, you don't need to work with the platformWindow anymore, you can use the method orderFront: and orderOut: of the CPWindow to open or close the CPPlatformWindow.
You can also use the method setFrame: of the CPWindow to automatically change the contentRect of the CPPlatformWindow. In one word we assume that the given CPWindow and the CPPlatformWindow will have the same behavior. When not using the constructor initWithWindow:, the method setFrame: of the given CPWindow won't do anything to the CPPlateformWindow.
Test app in Tests/Manual/CPPlatformWindow/
Previously, when calling the method orderOut: on the full platform window of a CPPlatformWindow, cappuccino didn't close the parent CPPlatformWindow. When he does, this only works if the parent platformWindow is not the primary platform.
Test app in Tests/Manual/CPPlatformWindow/
Previously when chaining sheet on the same window, Cappuccino crashed.
Now chaining a sheet in the delegate method of the method endSheet: will open the next sheet.
Now when opening a sheet, we check if another sheet is currently closing. If yes, we wait till we get the notification of the closing of the previous sheet. And then we open the next sheet.
Fixed#2159
Test app in Tests/Manual/AttachedSheet2/SheetWindowController.j
Previously, the Secure Text Field size was slightly smaller when placed using Interface Builder.
This commit adjusts the size in nib2cib by ensuring CPTextField _adjustNib2CibSize is called on CPSecureTextField.
Previously, when the cursor was hover a CPTextField the cursor was an arrow till we selected the textField.
Now when the cursor is hover a CPTextField, and if this textField is enabled and editable or selectable we show the selecting cursor. This is the default behavior in Cocoa as well.
Updated manual test Tests/Manual/CPTextField/
Previously, when having a CPTableView with a selectionStyle set to CPTableViewSelectionHighlightStyleNone, we needed two clicks to be able to select or click on a control of the cell. Now we only need one click as in Cocoa.
This is fixed by changing the method hitTest and returning the good view which can become the firstResponder.
Previously, when making a right click on a non-selectable CPTextField which is contained in a dataView of a CPTableView set the bezel attribute to yes.
Now, when making this things, Cappuccino will check if the CPTextField if editable or not.
Previously, when closing a CPWindow, the window kept a reference of the firstResponder. This things raised an issue when opening a window again and asking the window to make the first responder to the same element as the one we had.
Now when closing a window, the window will update its firstResponder to nil.
Previously, when making a CPDate with some constructors, we could pass a string or nil to create a date. For example it was allowed to do var date = [CPDate dateWithTimeIntervalSince1970:"10"];
Now this is not possible anymore and it works like in Cocoa, only number are authorized.
Previously, when changing the controlSize of a CPSegmentedControl, Cappuccino didn't have the theme to do that.
Now, Cappuccino supports the controlSize small and mini. Aristo and Aristo2 have new attributes for that.
The attribute theme "default-height" has been replaced by "min-size" and "max-size" (used everywhere else in Cappuccino).
The CPSegmentedControl has now the theme attribute "nib2cib-adjustment-frame".
This pull request fixed alignment issue when adding a CPSegmentedControl from xCode. In the method initWithCode, we didn't care about the divider thickness, now we do. This same method will also add leftovers pixel to have exactly the same size as in xCode. Leftovers pixels are here because FLOOR operation and size of the font.
Manual test in Tests/Manual/CPSegmentedControlTest
Manual test in Tests/Manual/Nib2CibAlignment
Previously, handlers of events that could make the popover to close were not calling CPPopover's performClose:. This patch ensure performClose: is called, and so delegates are consulted if the popover can be closed or not
Previously, when encoding and decoding a CPTableView, the tableView had several _CPTableDrawView (in the subviews).
Now when encoding, we make sure to remove _CPTableDrawView from its superview.
Previously, when adding a view (which is the firstResponder of the window) to the same window, the firstResponder was set to nil.
Now it keeps the same firstResponder.
objj.vim cannot highlight much syntax correctly, and it should be updated. I've modified objc.vim to objj.vim which reads javascript.vim in the beginning, and now it looks good. Because javascript.vim & c.vim must be loaded into objj.vim, so they should also be committed.
Fixes#2211
When running the "jake" commands in the repository directory, it would print:
$ jake --help
narwhal/packages/narwhal-jsc/bin/narwhal-jsc: line 33: narwhal/packages/narwhal-jsc/bin/narwhal-jscore: No such file or directory
Now, it will print:
$ pwd
/home/kevin/workspace/Cappuccino
$ jake --help
Usage: jake [OPTIONS] targets...
-f --jakefile FILE: Use FILE as the jakefile.
-T --tasks: Display the tasks with descriptions, then exit.
-D --describe: Describe the tasks then exit.
-P --prereqs: Display the tasks and dependencies, then exit.
-v --verbose: Log message to standard output.
-h --help: displays usage information (final option)
Fixes#2181
Previously, CPImageView would sometimes attach a notification to CPImageDidLoadNotification when the image was set, but would never remove this notification. The view could then leak memory as the notification center would be keeping a reference even if the view was no longer being used.
This commit fixes that issue by removing the observation once the image has loaded.
Fixes#2198.
-setDataView: was not working because the caching system was picking
views cached with the column identifier which is persistent.
Test: CPTableViewTest -testLayout
Previously, Cappuccino didn't handle retina device when drawing for canvas2D. Now it does.
To do that, Cappuccino will firstly calculate the pixel ratio of the current device, then it needs to change the css style of the canvas by multiply it by the current pixel ratio and finally scale the canvas by this pixel ratio.
More information about high DPI drawing here : http://www.html5rocks.com/en/tutorials/canvas/hidpi/
Added the method `setAllowsHighDPIDrawing:` and `allowsHighDPIDrawing` to deactivate or activate this feature.
Fixed#2175
Previously, CPButtonBar was not updating its layout after changing the visibility of its buttons.
This patch adds makes CPButtonBar observersing the buttons and relayout if the visibility of one button is updated.
Exclusive bindings are impossible in IB
Possible in code (tested with segmented control and selectedIndex & selectedTag) but then you get unexpected behavior. A segment can be selected after a click and then
changes to another segment.
After this commit, extra exclusive bindings are just ignored and we warn about it.
Fixed: Added an out of bounds check to _unloadDataViews:...
Revert: revert -reloadData to the previous behavior where views & data were reloaded, not only data. That's what cocoa does for view based tables.
Previously, the CPDatePicker sent an action to the target when the method setObjectValue: and setDateValue were called.
Now it sends the action only when the user makes an action.
Fixed#2206
Controls can implement -isExlusiveBinding: to determine if bindings are exclusive. Defaults to NO.
Exclusive bindings are mainly CPSelectedIndexBinding | CPSelectedTagBinding | CPSelectedValueBinding
When searching by range, the option of diacritic insensitive search was not being contemplated
Also add support to strip variants of 'E' 'I' 'O' 'U' that were not implemented.
Source: http://www.ascii-code.com/
This could cause memory leaks and was unnecessarily verbose.
CPPopUpButton , CPRadio and CPSegmentedControl now use a generic CPBinder method for searching the binder involved when the control is asked to reverse set the binding, typically after a user interaction.
Fixes#2199
TODO: prevent user to bind multiple selection bindings when these bindings are exclusive.
Usage: when a control is known to have exclusive bindings (like selected index/tag/value), use this method when
you don't know the binding name and want to send values to the binded object (via reverseSetValueFor:).
In CPBinder binderMap, picks the first binder matching the receiver class or subclass.
Previously, when clicking on a CPDatePicker in a CPTableView, cappuccino didn't select the tableView row firstly.
Now it does as in cocoa.
The stepper still have the same behavior as a CPButton in a CPTableView.
refs #2201
Previously, calling CPBinder unbindObjectForKey would not actually unbind each of the bindings because it was passing the wrong object to the unbind method which would silently ignore the issue. The commit passes the correct parameter to allow unbinding to actually take place.
Fixes#2197.
Previously, when hitting the backspace key on a none editable/enable field, this triggered the browser's back button.
Now it does not anymore.
All credits for @apajarita
Fixed#2189
The width computing of a string is always a little random, and that led to truncated labels. This patch adds
a pixel in width in order to fix any rounding errors.
Previously, when using a textual or graphical CPDatePicker in a CPTableView Cappuccino crashed.
This PR fix this issue by adding the subviews of a CPDatePicker in the method layoutSubviews instead of in the initialization.
This PR add another feature on the graphical CPDatePicker, the button next and previous month are now in the mode continuous as in Cocoa.
Since PR #2170, CPRadioGroup were truncated because they are specific components translated into a NSMatrix (CPView) object.
This fix sets CPRadio controlSize according to xCode value and uses its frame adjustment to adjust CPRadio inside a CPRadioGroup.
When running ./bootstrap.sh, it will print the error message:
"./bootstrap.sh: line 138: [: too many arguments"
This fix correct it by filtering out the OpenRDK Runtime Environment line.
Fixes#2179
Previously, when reloading the CPOutlineView and changing an item to another parentItem, the CPOutlineView deleted the reference of the item.
Now, when removing old data to avoid memory leaks, the CPOutlineView checks if the pending item to delete doesn't have another parent item.
Previously, when removing a view, Cappuccino didn't clean the notification center. The notification center kept in reference old views.
This PR fix this issue. When a CPView is added to a view, the methods _removeObservers and _addObservers are called. In these both methods we remove and add the observer to the notification center if needed. _removeObservers and _addObservers are called for the view and its subviews. These both methods are called through the method viewWillMoveToSuperview.
When a CPView is removed, we only call the method _removeObservers.
When a CPWindow is closed, we call the method _removeObservers on its contentView.
When a CPWindow is about to be opened, we call the method _removeObservers and _addObservers on its contentView.
Refs #1880
Refs #2024
Test app in Tests/Manual/AttachedSheet2/SheetWindowController.j
Previously, when adding, removing, replacing a CPView, the framework didn't call the methods viewDidMoveToSuperview, viewDidMoveToWindow, viewWillMoveToSuperview and viewWillMoveToWindow as in Cocoa. Now it does.
The main change is that these methods will be called when removing a CPView.
Added unit-test in Tests/AppKit/CPViewTest.j
Previously, when a CPTextField was in a CPPlatformWindow, the checking of if the CPTextField was usable or not (depending on its position) in a CPPlatformWindow didn't take in account the content-inset of the CPTextField. Now it does.
Previously, when reloading a CPOutlineView with different datas than before, the CPOutlineView kept a reference of the previous datas.
Now, when reloading, the CPOutlineView will delete these references when we don't need it anymore. For that, the CPOutlineView will only try to delete the previous reference when we have reloaded the children of a node, otherwise it will keep the other references.
This PR fix some issue with the following methods :
- (BOOL)isExpandable:(id)anItem;
- (BOOL)isItemExpanded:(id)anItem;
- (CPInteger)rowForItem:(id)anItem;
- (CPInteger)levelForItem:(id)anItem;
- (id)itemAtRow:(CPInteger)aRow;
- (CPInteger)levelForItem:(id)anItem;
- (CPInteger)levelForRow:(CPInteger)aRow;
Previously these methods took care about not displayed items. Now they just work with displayed items.
Added unit-tests in Test/AppKit/CPOutlineViewTest.j
Test app in Test/Manual/CPOutlineViewTestCib
Previously, when opening a popover after another one, the CPPopover was blurred.
This occurred due to some CSS Transitions specification :
"Since this specification does not define when a style change event occurs, and thus what changes to computed values are considered simultaneous, authors should be aware that changing any of the transition properties a small amount of time after making a change that might transition can result in behavior that varies between implementations, since the changes might be considered simultaneous in some implementations but not others."
The fix consists to access to a transform value of the style of the _DOMElement to force it to recalculate the values.
More information here : https://code.google.com/p/chromium/issues/detail?id=388082Fixed#2143
Test app in Test/Manual/CPPopover
Previously, the deleteForward:, deleteBackward: and delete: methods in CPTextField called setStringValue, which would consult a formatter, and if the formatter failed all of the text would be deleted.
Now the code has been refactored such that when one of the delete methods is invoked via a key event, the formatter is not used and the DOM event is used, which results in the expected behavior.
A bug in checking the selected range was also fixed in deleteForward:.
The CPFormatter test was updated with buttons to call the delete methods directly to ensure they work correctly when not invoked from a key event.
Closes#1997
Without this change, using a traditional scroll wheel mouse with discrete steps would result in very slow scrolling in Firefox.
With this fix we handle these events like in Cocoa: as "[im]preciseScrollingDeltas", which `CPScrollView` in turn knows to apply the configurable line scroll amount for. For backwards compatibility, [event deltaX] and [event deltaY] are premultiplied with a suitable constant, while [event scrollingDeltaX] and [event scrollingDeltaY] show the true values needed.
Browsers other than Firefox seem to always send pixel scrolling information even for old style scrolling devices, at least on the Mac, so are not affected by this change.
Fixes#2013.
This parameter exists on CPView, so it has been removed here. A note was added on the setter to identify that the parameter was defined in the superclass.
Previously, _textFieldBackgroundColor was used as the internal variable, and getter/setter methods were explicitly declared to handle getting/setting this value.
This commit changes the name of the internal variable to match the Cocoa compliant method (backgroundColor/setBackgroundColor). It also declares an @accessor option to auto-generate the getter. The setter is explicitly declared because it needs to mark the object as needing display.
Additionally, the logic for the setting is moved from the older method, setTextFieldBackgroundColor, to the new one, setBackgroundColor, and a comment was made to clarify that the older one was deprecated with a link to this issue.
CPDateFormatter previously only defined symbols for "en"-locale. Using another locale simple yielded nil-results for most keys. This fix addresses this behavior by always returning the english symbols as default unless other symbols have been explicitly provided or set.
CPDateFormatter only supports the "en" locale by default. All other locales yield empty results for most symbols. The tests added ensure, that the english "default" values are returned if no specific data is available (which is currently the case for every locale except "en").
Previously when hitting escape on a CPPanel, the key chain event was broken.
Now when the user hit escape, cappuccino do the following process
- Check if the firstResponder or the chain of the nextResponder implement cancelOperation:
- If not, performKeyEquivalent on the firstResponder and so on
- If not, call the method cancel: on the firstResponder and the chain (not in the case of the CPPopover)
Previously, a complete: method was called if the user hit escape on a void window. This is not a feature of Cocoa, @apajarita or @aljunberg can you confirm that ?
This commit is related to the commit #2139
When using the press tool, it can happen that one of the theme attribute is not set as not everything is instantiated.
This fix makes sure that the attribute is set before accessing it to avoid any crash of the press tool.
Previously, the delegate popoverWillShow was called even if the popover was already shown. This PR fix this issue.
Also, it was possible to launch several times the method orderFront of the _CPPopoverWindow from the CPPopover.
For example with a double click on a button. Now the _CPPopoverWindow has a var isOpening and nothing else is possible during the opening of the popover (not possible to close or to click somewhere else etc etc). This work as in Cocoa.
Test app in Tests/Manual/CPPopover/
From the original PR:
CPURLConnection was modified to reduce the withCredentials flag to a file-scope member. This has been done to facilitate use of withCredentials when a developer cannot modify the CPURLConnection manually (e.g. https://github.com/wireload/Ratatosk/blob/master/WLRemoteLink.j, line 466).
Refs #1296, #2138
From the original PR:
Currently, CPURLConnection has no way to access the withCredential
property of an XMLHTTPRequest, meaning we can't utilize cross-domain
calls using CORS in our applications.
This commit adds the withCredentials property to CFHTTPRequest.
Refs #1296
Previously, the method setTextColor changed the color of the placeholder as well.
Now it doesn't as in cocoa
Test app in Test/Manual/CPTextFieldEditingStyleTest
Previously, when making a click on the stepper of a textual CPDatePicker, the intern value of the stepper wasn't updated with the correct value of the CPDatePicker. This occurred when the user changed the dateValue in the delegate methods of the textField.
Previously, a panel or a popover didn't close when the user hit the key escape.
Now it does. The panel or the popover will close on the keyDown event as in cocoa.
Test app in Tests/Manual/CPPanelTest/ and Tests/Manual/CPPopoverTests
Fixed#1767
Previously, the unsigned delegate methods _implementedDelegateMethods was erased by the CPTextField, so some delegate methods didn't work any more.
This PR just renames the unsigned delegate methods to _implementedTokenFieldDelegateMethods.
Previously, the update of the style of an editing textField didn't work.
Now when editing, we can update the font, color, alignment and vertical-alignment of the textField
Test app in Tests/Manual/CPTextFieldEditingStyleTest/
Fixed#2057
Previously, when editing a cell, the other cells of the tableView lost the themeState of the firstResponder.
Now the other cells, as the tableView, keep this state when a cell is editing.
Fixed#2133
Previously it wasn't possible to select a set of dates with a shift click, now we can.
This works only when the mode CPRangeDateMode is set.
It has the same behavior as in Cocoa.
Previously, the textual datePicker didn't work because it couldn't become the firstResponder.
This PR fixed another issue as well. Previously it wasn't possible to modify a date when moving only with tab and when the datePicker came firstResponder with tabulation.
Previously, bootstrap didn't handle some OpenJDK Runtime Environment version as: rhel-2.4.7.1.el6_5-x86_64 u55-b13.
Now it did, the regular expression to catch the version of the jdk was wrong.
Fixes#2113
Previously when making a paste with a right click caused a crash because the location of the fake event was set to null.
This crash occurs only if the user didn't hit the keyboard yet.
Now, the location of the fake event takes the location of the last event of the app, who is the right click.
Previously, the _CPPopoverWindow was trying to remove itself as observer while it wasn't defined
as an observer. This would cause a warning. This patch encapsulates addObserver and removeObserver into methods
that actually check if it's already observing/not observing when trying to observe/not observe
If a table view row is selected its text will turn white to offset against the blue selection colour. However, when the table is not the first responder, or the window is in is not the key window, the light grey highlight colour is used instead, but the text remains white.
This fix ensures the text becomes black in this case by utilising the new first responder and key window theme states.
Fixes#2009, fixes#2001.
The nib2cib change which introduced support for the plus and minus template images in nib2cib also broke support for converting radio groups.
This fix restores that support.
Every view in the key window now takes on the 'key window' theme state. This makes it easy to theme everything in inactive windows in more muted colours. Also, combined with the first responder theme state of the previous commit we can have visuals like "first responder but in an inactive window" for text views and such things.
When a CPView becomes the first responder it now takes on the 'first responder' theme state, as does all its subviews. This allows it to naturally change appearance when it has the keyboard focus, such as with focus rings and brighter colour schemes.
This change makes sure there is both a level indicator which goes into warning and critical display mode for low levels and one which goes into the same for high levels in the manual test for level indicator.
Refs #2071.
Previously, an errant semicolon resulted in improperly scoped variables in CPSegmentedControl.
This fix removes the semicolon in favour of a comma.
Fixes#2117
Previously, _CPToolTip was not testing if the view and so parentWindow was still here. During the timer, it might be gone, which would cause crash. This patch ensure the window is still there before showing the tooltip
This PR add the possibility to update cappuccino from xCodeCapp. How does it work :
- Create a temporary file for cappuccino
- Download Cappuccino. The downloaded Cappuccino will be either the last release or the last version of the master branch (option in the preference)
- Make a jake clean
- Make a jake install
There is a small window to follow the progress of the install (3 steps ; download - clean - install).
If there is a issue with the install, the developer will be notify with which step failed (xCodeCapp internally log everything).
The environment var CAPP_NOSUDO is set to 1. It means, if a developer did a bootstrap or a previous jake install in sudo, the update of cappuccino will fail.
Changed the credit of xCodeCapp and pass it to the version 3.1
Previously when calling setContentSize several times, the final frame of the view was wrong.
I removed the animation as well. This doesn't work, and need a lot more of implementation.
Rename _numberOfRowsDidChange -> _dataViewsNeedReloadAfterContentChange.
Subclasses use this method to tell if a full view reloading is needed
when calling -reloadData.
Currently CPOutlineView returns YES - this is the previous behavior.
Added private - (void)_reloadDataForRowIndexes:(CPIndexSet)rowIndexes
columnIndexes:(CPIndexSet)columnIndexes
This is the internal method for reloading objectValues only.
Possible regression from #fe260a8
Regression: -reloadData does not reload views any more even if the table is empty (see CPOutlineViewCibTest).
BUG: -removeTableColumn: error.
Conflicts:
AppKit/CPOutlineView.j
AppKit/CPTableHeaderView.j
AppKit/CPTableView.j
Now capp_lint is fully integrated in xCodeCapp.
You can check the entire project with the menuitem of the app.
There is an option in the preference to make a capp_lint (on the file who was just processed) each time a file is processed by xCodeCapp.
Error of capp_lint are reported in the error tableView. The messages displayed the following informations : path of the file, line and error.
Previously we had to install capp_lint separately from Cappuccino.
Now capp_lint will be installed with Cappuccino.
Before capp_lint was installed in /urs/local/bin/, now it is in /usr/local/narwhal/bin/.
There are some weird race conditions happening in Chrome 34 regarding the transitionend event. Setting the timeout to 10ms is not noticealble for the user, and seems to fix the issue.
If the receiver is an identifier and not an ivar that needs 'self.' infront there is no need to assign it to a temporary variable. Also if the identifier is 'self' there we assume it is not nil.
This patch fixes a warning thrown by recent versions of Chrome about IE ways of doing. It simply uses aDOMEvent.returnValue if aDOMEvent.preventDefault() and aDOMEvent.cancelBubble if aDOMEvent.stopPropagation are not avavailable
Previously, when delegates implemented tableView:selectionIndexesForProposedSelection:, the selection with arrow were broken. This patch ensure to try to select next/previous row if the current row cannot be selected.
It also cleans up the selection indexes when sending selectAll: as Cocoa does.
This color has a standard 'transparency' checkerboard pattern which is especially useful for our scrolling tests where it makes scrolling easy to visualise.
Refs #2078, #2079.
Previously when using a local copy to install Cappuccino, we used to copy the local capp-base at /usr/local/narwhal and then we tried to move /usr/local/narwhal/cappuccino-cappuccino-base*/* to /usr/local/narwhal/. .
Now we move /usr/local/narwhal/cappuccino-base*/* to /usr/local/narwhal/. .
Added protocol CPAnimationDelegate.
Also, now, delegate methods are called as we are used to do it in Cappuccino
Test app in Tests/Manual/CPAnimationTest.
In browsers without support for _DOMElement.removeEventListener CPPopover
throws an exception when closing, and the popover is not removed. This fix
checks for support implicitly using the _browserAnimates variable.
Previously, selecting an empty text field hid the contentView, causing the placeholder string to dissapear. This fix leaves the contentView of the text field visible when the internal _stringValue is nil or empty.
Fixes#2049
Previously, jake install was calling tusk to ensure installed packages are up to date, and was used to install the new build. As tusk and all that narwhal suite is deprecated and causes more and more problems, this patch:
- removes the up-to-date check: it's very likely that no package will never be updated anymore
- uses a local function to install the freshly build
Previously, bootstrap.sh was always trying to fetch cappuccino-base from github. This patch provides a new option "--copy-local [PATH]" that will let the user give a path of a local copy of the cappuccino-base zip. This can be pretty useful in an automated build system that cannot always access the internet.
When having a CPSegmentedControl within a a XIB files, nib2cib crashes with the following error:
```
'undefined' is not an object (evaluating 'self._themeStates[aSegment].without')
```
This pull request fixes the problem by making sure that the drawBezel method is called with a valid segment index when accessing the themeState.
Reduction: https://dl.dropboxusercontent.com/u/14629300/BugCPSegmentedControl.zip
Previously with the CPRoundRectBezelStyle we got a CPNull themeState. Now we get an instance of ThemeState with the theme CPButtonStateBezelStyleRoundRect. CPButtonStateBezelStyleRoundRect is equal to CPThemeState("roundRect").
Without this fix, in Internet Explorer the plist parser would only receive an empty XML document for the main Info.plist (and any other such loaded documents) if the server did not provide the content type "text/xml".
This fix parses the XML correctly even if there is no such content type.
Fixes#2051.
This fix ensures the PLIST_FIRST_CHILD() code gets to run both statements in the XML parsing while loop.
There's no obvious way to trigger an error due to this bug, but better correct than wrong.
Refs #2051.
Depending on whether min < max or max > min CPLevelIndicator changes the order of the normal/warning/critical color. This commit enables this behavior for both cases.
Previously, the day/month/hour of the textual datePicker had always two digits, even when the number was under 10. For instance the day 8 was displayed 08.
Now it's like in cocoa, 8 is displayed 8.
Fixes#2059
Previously, clicking anywhere in an empty CPTableView with variable row height was causing a crash due to the code trying to find the latest cached row height. This patch adds a check to ensure the last cached row height is not nil.
Before this commit, row templates and views belo,ging to them were copied via archiving/unarchiving.
This was causing problems when the view was not completely CPCoding compliant, like for CPDatePicker.
With this commit, template views use CPCopying (-copy method) when available.
Also added CPCopying for CPDatePicker as a category in CPPredicateEditor.
Test: in CPPredicateEditorTest, on the right panel, choose a keypath, operator, and select 'Dates' in the popup for the right expression.
Fixes#2041
This PR fixes several issues about the textual CPDatePicker :
- Fixed the issue about keyView loop when not assigning the nextKeyView
- Changed the behavior of the component when editing or deleting the date. It works as in cocoa now.
- Fixed bug about frameSize of each component when editing
- Fixed bug about NaN or null values
- Added optimization
- Refactoring the code
Commit df2be92 breaks CPComparisonPredicate if one side of the comparison is a property which returns a BOOL.
CPPredicateParser parses literal BOOLs (YES/NO) into 0/1 using CPNumber>>numberWithBool:. CPComparisonPredicate however does not cast a BOOL from a property using numberWithBool:. E.g.
[[YES,NO] filteredArrayUsingPredicate:[CPPredicate predicateWithFormat:'self = YES']]
used to return [true], now it returns [].
This fix addresses the issue by "typecasting" the lhs/rhs values to numbers using CPNumber>>numberWithBool: if they are of type boolean.
Fixes#2028
When opening a sheet, with a double click the frame origin of the sheet changed and the parent disappeared.
This PR fix this issue.
Test app in Tests/Manual/AttachedSheet/AppController.j
Previously, the style of the blueish view that shows up on CPTableView when user is about to drop something on it was hard coded.
This patch adds the following theme attributes in CPTableView
// theme of the drop view for CPDragOperationOn on a non selected row
dropview-on-background-color
dropview-on-border-color
dropview-on-border-width
dropview-on-border-radius
// theme of the drop view for CPDragOperationOn on a selected row
dropview-on-selected-background-color
dropview-on-selected-border-color
dropview-on-selected-border-width
dropview-on-selected-border-radius
// theme of the drop view for CPDragOperationAbove on a non selected row
dropview-above-border-color
dropview-above-border-width
// theme of the drop view for CPDragOperationAbove on a selected row
dropview-above-selected-border-color
dropview-above-selected-border-width
Aristo and Aristo2 have been updated, and use the same old hardcoded values.
Without this fix, -CPView convertPoint:fromView was broken due to the changes made in #1998. In particular, the transform for the fromView would be calculated and then just thrown away if the views were in the same window. This happened to work in some specific cases and the lack of unit tests concealed the error.
This fix properly applies the transforms when two views are in the same window without one of them necessarily being inside the other.
Refs #1998.
With this feature, CFURL and CPURL can now correctly transform a relative scheme URL starting with a double-slash.
For example, + CPURL URLWithString:@"//a.se/a" relativeToURL:@"ftp://b.se/c" would represent an absolute URL of "ftp://a.se/a".
This PR adds a new feature for CPControl. Now when leaving the control, the userInfo of the notification CPTextDidEndEditingNotification contains the last movements of the control as in COCOA
Test app in Tests/Manual/CPTextFieldMovementsTest
Previously shouldSelectTableColumn wasn't called in CPTableView and CPOutlineView.
This PR resolves this issue, shouldSelectTableColumn is now called between selectionShouldChangeInTableView and tableViewSelectionIsChanging as in COCOA.
Test app Tests/Manual/TableTest/DelegateSelectionTest/
The Cocoa signature for this method
+ (NSNumber *)numberWithBool:(BOOL)value
implies that a number is being returned. In Cappuccino a Boolean is returned.
This leads to some problems e.g. in
addObserver:forKeyPath:options:context: with CPKeyValueObservingOptionPrior
and observeValueForKeyPath:ofObject:change:context: with CPKeyValueObservingOptionPrior.
Setting CPKeyValueObservingOptionPrior leads to CPKeyValueChangeNotificationIsPriorKey being set to 1 (which should be equal to [NSNumber numberWithBool:YES] as documented in https://developer.apple.com/library/mac/documentation/Cocoa/Reference/Foundation/Protocols/NSKeyValueObserving_Protocol/Reference/Reference.html. The actual value though is true as [CPNumber numberWithBool:YES] does not return a number but a boolean. This commit fixes this issue.
Previously the delegate method shouldSelectItem didn't work in CPOutlineView.
It didn't work because the checking method _delegateRespondsToSelectionIndexesForProposedSelection and _delegateRespondsToShouldSelectRow didn't check if the real delegate of the outlineView implemented the methods
Previously the method shouldSelectRow was called even it's the method selectionIndexesForProposedSelection was implemented by the delegate. Now it works as in Cocoa.
The method selectionIndexesForProposedSelection is called when it has to be called, like in Cocoa.
This PR adds new methods to check and call the delegate/datasource methods more easily
Test app in Tests/Manual/TableTest/DelegateSelectionTest/
Previously, a decoded CPNumberFormatter was converting the minimum and maximum values to 0 if they were set to nil. This was causing unexpected behavior when creating a CPNumberFormatter from a xib. This patch actually decodes _minimum and _maximum values as objects to preserve nil.
Test added in Foundation/CPNumberFormatter.j
Previously the call of the delegate method of a CPTableView weren't call in the good order, specially the selecting method.
Now it works as in Cocoa.
Fixed typos
Test app in Tests/Manual/TableTest/DelegateSelectionTest
This PR adds the support of scaling in CPView.
There are two new public methods in CPView : -(void)scaleUnitSquareToSize: and -(void)setScaleSize:
-(void)scaleUnitSquareToSize: works exactly as in COCOA, it means if you set a first scale to 0.5 and then 0.5 again, the scaleSize of the view will be 0.25
-(void)setScaleSize: works with the value given. If you set 0.5 after you just seted 0.5 the scaleSize will be 0.5. This method is definitly better in using, specially when using the scaleSize binding with a slider.
Test app in Tests/Manual/ScalingTest
Previously, using a truncating line break mode on a CPTextField and setting the string value to "test test" was displaying "test test". This patch changes the css white-space property from "nowrap" to "pre" which preserves the white spaces.
This caused 'null is not an object' type error when class declaration with 'accessors' declared ivars was in the same file as the protocol it was conforming to.
Earlier the compiler generated the 'if' without the last ';' when a empty statement is used, with sometimes devastating result. The same if an empty statement was after the 'else'.
I have also added a test case for some different empty statement senarios.
Previously, if a window had an autoresizingMask and was marked as not being constrained (for example _CPMenuBarWindow), and the CPWindowConstrainToScreen was set to NO, resizing the platform window would not autoresize the window.
This commit ensures that windows that autoresize but don't care about constraining will resize with the platform window no matter what.
Fixes#1981.
Previously when opening a submenu from a menu, the main menu disappears because a issue with the poolMenuWindow.
This PR fixes this bug.
It also handle more properly the case when an user is making several right clicks on a responder (origin problem with the ghost menu).
Fixed#1887
Copy, cut and delete are now disabled if there's no selection in the active text field. Cut, paste and delete are greyed out if the text field is not editable (a label).
Refs #1964.
When the Edit menu is used to cut or to paste, Cappuccino needs to do all the work of making it happen. But the current code relied on the browser doing part of the work.
This fix adds new state so CPTextField can know if it should expect the browser to do some of the work or not.
Refs #1964.
This made it impossible to select some (non-editable) text, and then to click a Cappuccino menu option like Edit > Copy. Upon the click of Edit, the just selected text would be lost.
This was caused by some very mysterious code focusing and blurring an input every time propagation was stopped. Since it wasn't documented, the snippet was just removed for now. If it turns out it was useful we'll need to add it back and make sure it's not called in this scenario (while documenting it properly).
Refs #1964.
Previously, addChildWindow:ordered: was setting the platform window of the child window.
This patch restore the correct behaviour, and fix a bug with _CPAutocompleteMenu. Now the autocomplete menu is opened as a child of its textfield window using the correct platform window.
Previously it was possible to have a highlighted CPMenuItem for a disabled CPMenuItem. (To get that, select an item, close the menu, disabled the item manually and then open the menu again).
Now it is not possible to get this behavior.
Previously, addChildWindow:ordered: did not set not the child window's platform window. This could result to open them in the wrong browser window. This patch makes sure to set the parent's platform window as the child window's platform window.
The div's id was changed to cappuccino, and it was supposed to be cappuccino-body, which broke code that relied on the DOM remaining stable during startup.
Previously, nib2cib sent the output of plutil to a temp file, then read that file. plutil supports writing to stdout, which avoids the temp file completely, so that is what we do now.
Previously if - CPTextField `deleteBackward:` was invoked without it being a browser backspace key event for an actively edited text field, the first character in the current selection would not be deleted.
This fix makes it so that `deleteBackward:` works properly from any caller. The fix also undoes a mistake in a recent previous commit which prevented the text field from visually updating after deletion.
Previously, attempting to use an unsupported built in image such as NSActionTemplate would generate an error message that said, "The image “NSActionTemplate” cannot be found." This would lead the user to think that there was something wrong with nib2cib, instead of understanding that only certain built in images are supported (currently NSAddTemplate and NSRemoveTemplate).
With this commit, an attempt to use an unsupported built in image (where built in is assumed to be any name that matches /^NS[A-Z][A-Za-z]+$/) results in the error message "The built in image “<name>” is not supported."
This matches Cocoa behaviour. Cocoa also supports double click to select the current word but we currently don't know what word is being clicked. In some browsers (e.g. Safari) this will work natively. Chrome on the other hand selects something random after double click on a non editable div.
Like in Cocoa, a non-editable but selectable text can now be clicked and all text can be selected with the Edit > Select All option or the associated keyboard equivalent. Just like in Cocoa this works on selectable text labels as well (text fields without a bezel).
A non-editable text field does not use an input field, and so does not track what is selected within it. We will use a workaround for now but in a CoreText based text field in the future we'd be likely to always have a correct selectedRange and be able to remove this hack.
Fixes#1972.
Without this change, text fields with setEditable:NO but setSelectable:YES could not become the first responder, and so would not listen for copy: events.
This change allows selectable fields to become the first responder, taking advantage of the distinction between first responder and first key responder maintained internally in CPTextField - a merely selectable field does not become the first key responder.
Refs #1972.
Previously it wasn't possible to create project with symlink.
Now you can set in the preferences of xCodeCapp if you want to create your project with symlink or not with a radioButton.
When launching xCodeCapp, the system will check if $CAPP_BUILD is defined or not. If not, the button in the preference will be disabled and a toolTip will advice how to create a $CAPP_BUILD variable.
Previously, CGContextAddArc would not work with an empty path. In Cocoa, CGContextAddArc implicitly starts a path, so it does not require a non-empty path.
This commit allows CGContextAddArc to be used with an empty path.
The Objective-J parser and compiler now handles protocol syntax. The Objective-J runtime has new functions to handle protocols. The method 'conformsToProtocol:' has been added to CPObject as an instance and a class method.
Previously a fixed margin of 40px was maintained at the left/right of a window when resizing it, which was excessive.
Now the minimum size allowed during a resize is the same as the minimum margin allowed when moving a window, which is determined based on the type of window.
Previously, when the platform window was resized, windows were moved/resized to stay within the virtual screen, but if the window had a minimum size, it was not respected.
This commit ensures that if a window has a minimum size, it will always be respected.
Previously window heights were constrained to the usable screen when the platform window (virtual screen) resized, but widths were not. The code was there to constrain the width, but it wasn't being used.
This commit ensures the width constraining code is executed.
The platform window is the virtual screen for Cappuccino windows. Per Cocoa, when the screen (platform window) resizes, CPApplicationDidChangeScreenParametersNotification is sent and the CPApplication delegate method applicationDidChangeScreenParameters: is called.
Previously, window moves were constrained such that a minimum margin of the window was visible on screen, but window resizes were not constrained, allowing the user to effectively move the window off screen.
Now resizes are constrained such that a minimum margin of the window is always visible on screen so that the window does not get "lost".
Previously, whenever the platform window (virtual screen) resized, windows were moved or resized to keep them on screen. This could not be easily overridden by the user.
Two class methods have been added:
CPWindow +setConstrainWindowsToUsableScreen:
CPWindow +constrainWindowsToUsableScreen
If +setConstrainWindowsToUsableScreen:NO is called, then resizing the platform window will not move or resize user windows.
Previously, _CPAutocompleteMenu was trying to relayout itself using its related textfield's window frame. This was
causing a crash if one of the CPTokenField's superview has been removed. This fix ensures to stop layouting if the related text field has no window anymore.
Previously it wasn't possible to create a capp project from xCodeCapp.
Now you can create one from the menu item Create Project, xCodeCapp will load the project as the same time is no error occurs.
When choosing an existing project, xCodeCapp removes the old one (this is asked to the user, would you like to replace the files...).
In case of errors, the errors panel appears.
Previously, the index[-debug].html in the default capp templates did not include a loading progress function. We just had the plain spinner, which gave no visible sign of progress.
Now the capp templates include a progress function and a nice iOS7-ish progress thermometer is displayed during loading (except when run directly from source). 'jake debug' and 'jake release' will now generate the application/image sizes needed to display loading progress. The old spinner.gif has been removed.
In addition, the message displayed when JavaScript is unavailable has been simplified and shows a link to www.enable-javascript.com instead of the missing link to our old site.
Previously, Cappuccino only called OBJJ_PROGRESS_CALLBACK when an executable or image sprite file was finished loading. Only a slow connection, this would result in long periods with no visible progress.
Now OBJJ_PROGRESS_CALLBACK is called continuously during download via the XHR progress event. All supported browsers except IE 9 support this event.
Previously, in IE you could scroll the entire browser window to the right a very long way, leaving white space to the right of the Cappuccino app. This was caused by a 100,000px wide measurement span that was supposed to be positioned at left:-100,000 but was mistakenly at left:-10,000, which left 90,000px on screen.
This commit fixes the left position of that span, eliminating the ability to scroll the app.
Fixes#1725
Previously, unless you were using a sheet alert, there was no way to set a custom dismissal handler for an alert. In Cocoa alerts are synchronous, in Cappuccino they are asynchronous, which meant that a single dismissal handler had to handle all possible non-sheet alerts.
This commit adds several methods which add two ways to specify a custom dismissal handler: via delegate/selector, and via a block (function). The new methods are:
setDidEndSelector:
runModalWithDidEndBlock:
beginSheetModalForWindow:didEndBlock:
Note that didEnd blocks are transient and take precedence over didEnd selectors.
The CPAlertTest demo was updated to fix some bugs and added the ability to run the tests using blocks.
Previously the inner element's offsetWidth was tested before and after setting the overflow to scroll. This is unnecessary, we can just set the overflow to scroll and test the outer element's clientWidth vs. offsetWidth. By definition, clientWidth does not include scrollbars, whereas offsetWidth does. So if they are equal then overlay scrollers must be in use.
Note that FireFox (as of version 22) does not support overlay scrollers, so even if the system does, Cappuccino will not on FireFox unless the programmer forces overlay scrollers.
Previously CPTimerMinTimeInterval was used as a constant for the *default* time interval used when a timer was initialized with an interval <= 0. It isn't the minimum, it is possible to create a timer with an interval less than that constant's value.
The name has been changed to CPTimerDefaultTimeInterval to make it clearer what it's purpose is. It has also been changed to a #define, that will result in smaller generated code.
Previously, if the last row's height was increased, empty rows below would not be pushed down. This was due to an off-by-one in -_unboundedRowsInRect (now _exposedRowsInRect) which caused the incorrect rect of the last row to be retrieved.
This commit fixes that bug and also makes the following changes:
- Added FULL_ROW_HEIGHT(), ROW_BOTTOM() and HAS_VARIABLE_ROW_HEIGHTS() macros to make the code clearer.
- Documentation cleanup.
- Eliminated some intermediate variables.
- Rewrote -_rectOfRow to reduce redundant code and fixed some logic errors.
- Renamed some variables to make their intention clearer (to me at least).
- Added an out of bounds check in -rowAtPoint: before doing the relatively expensive binary search.
- In -noteHeightOfRowsWithIndexesChanged:, the first valid index is used instead of the first index (which might be invalid).
- Fixed a bug in -noteHeightOfRowsWithIndexesChanged: where row height was not set correctly if a row was not in the index set.
- Fixed incorrect use of CEIL in -_unboundedRowsInRect, it was wrapped around the divisor instead of the entire expression.
- Fixed drawBackgroundInClipRect: drawing one too many rows.
- Fixed -rowAtPoint not checking for x in bounds.
BREAKING CHANGE:
Previously, CPTableView -rowAtPoint would incorrectly return a valid row index when the x value was outside of the table's bounds. Now an out of bounds x value will return -1, which is consistent with Cocoa (and the intent of the method).
Previously, the auto-enabling logic for CPMenu would not check to see if a menu item had enabled bindings, and would ignore the state determined by those bindings.
Now, if a menu is auto-enabling and an item has an enabled binding, the binding is used to set the enabled state of the item.
Previously setObjectValue checked for a formatted value, and then called [self stringValue], which ended up formatting the value again.
Now all of the cases (no formatter, formatter succeeds, formatter fails) are handled explicitly in setObjectValue, eliminating the extra call to the formatter.
Previously, no attempt was made to become/resign first key responder when a CPTextField was unhidden/hidden. This could lead to a case where a hidden CPTextField was made first responder, then unhidden (and have a focus ring), but would not receive key input.
Cocoa specifies that viewDidHide will be called when an unhidden view is added to a view hierarchy with a hidden ancestor view, and viewDidUnhide is called when an unhidden view is removed from a view hierarchy with a hidden ancestor view. This was not the case in Cappuccino.
This commit adds that functionality, and also removes a spurious viewDidHide/viewDidUnhide call that was generated as a side effect in initWithCoder.
In Objective-C, both `[@34 compare:nil]` and `[@34 compare:[CPNull null]]` throw invalid argument exceptions.
This fix makes the same true in Objective-J.
Refs #1959.
Previously, nib2cib would not correctly calculate the app/resource directories if the xib was in a subdirectory of Resources. Also, image resources in subdirectories did not have the subdirectory encoded with the image name.
Now xibs and images may be anywhere under Resources.
Fixes#1960.
Previously, _createDOMImageElement was not called until after [super initWithCoder] was called. But that call would end up in CPImageView -setObjectValue, which would attempt to access _DOMImageElement before it had been created.
Now _DOMImageElement is checked before being accessed or created.
Previously, a case could arise where a text field was set as the initial first responder but during cib loading was not yet assigned to a window. In this case _isWithinUsablePlatformRect would fail.
Now _isWithinUsablePlatformRect checks to make sure the text field has a window.
Tagging with UIDs can make it much easier to debug encoding/decoding related bugs where it's not clear that the right view ended up in the right spot.
The new data attribute is called "data-cappuccino-uid".
Also collect and simplify appkit_tag_dom_elements code a little.
Without this fix, the `appkit_tag_dom_elements` flag would only affect CPImageViews unarchived from a coder (e.g. in a collection view, from a cib). It would have no effect on a CPImageView created in code.
This change refactors the _DOMImageElement code to a shared method, reducing redundancy while fixing the bug.
Without this fix, code like `[[[CPImage alloc] initWithContentsOfFile:nil] description]` would cause a crash.
This fix makes `[[CPImage alloc] initWithContentsOfFile:nil]` behave like its equivalent in Cocoa: it returns nil without warning or error.
If a cib file had a custom CPWindow subclass, that cib file would not be properly processed by flatten as it'd fail to find the relevant class (the fact that the class can't be found is an issue in itself). This would cause an exception.
This fix makes CPCibWindowTemplate act similar to CPCibCustomView when the referenced class can't be found: it instantiates a regular CPWindow. This is sufficient for flatten's purposes.
The fontinfo and imagesize Xcode projects still explicitly refer to using the Mac OS X 10.5 SDK, while all other Xcode projects simply refer to the latest Mac OS X SDK, with no explicit OS version. This change modifies these two projects to mirror other Xcode projects in the project.
Fixes#1957
Previously, certain situations could easily lead to an infinite loop in - CPView nextValidKeyView. For instance, opening a sheet view with no responder returning YES for canBecomeKeyView would freeze Cappuccino.
This was caused by the cycle detection in nextValidKeyView not being able to pick up on cycles where the nextKeyView of the original receiver did not participate. Since content views often have a next key view inside of them, and the views inside rarely have a next key view back "up" to the content view, the content view would never be seen again even if there was a cycle.
This fix detects cycles at any level by remembering all previous visited responders.
This change brings nearly all copy and paste related into a single, self-contained class. This new class is still somewhat coupled to CPPlatformWindow+DOM.j in the sense that it's highly specific and meant to be called in a very specialised way. Despite this, it's still a significant decoupling and the new code is more readable and easier to follow.
Without this fix, if a menu item was disabled it could only be automatically re-enabled if the menu item had a `validateMenuItem:` or `validateUserInterfaceItem:` enabled target.
This meant that for example if the Edit > Select All menu item was validated when the first responder was something like a collection view, it'd become disabled. If then the first responder was changed to a text field and revalidation occurred, the menu item would not become enabled and Select All would not be possible in the text field neither through the Edit menu nor the Cmd-A/Ctrl-A keyboard equivalent.
This fix ensures that menu item validation does not only take negative action (disabling enabled items which should be disabled), but also positive action (enabling disabled items which should be enabled), even when there's no `validateMenuItem:` or `validateUserInterfaceItem:`.
Image copy and paste to and from the system clipboard is not available in Cappuccino today. However, the image in the collection view can be used to test internal Cappuccino copy and paste (using Edit > Copy, Edit > Paste from the Cappuccino menu).
This change begins to clean up the control flow of copy and paste. Before, copy and paste was dealt with partly in CPPlatformWindow, partly in CPApp and partly in CPTextField.
Now, CPApplication has been removed from the path, simplifying it down to the two more expected actors.
These targets might be rare but could occur in some Cappuccino text editor widget, in which case our standard native paste handling from Safari should work fine.
Without this fix, when text was copied from a CPTextField, the value would be copied into the system clipboard (if possible), but not into the Cappuccino pasteboard. This meant that the text could not be pasted back in Cappuccino using e.g. Edit > Paste from the Cappuccino menu.
This fix ensures the appropriate text is copied both into Cappuccino's pasteboard and the system clipboard (when possible).
This test makes it easy to test copy and paste both inside a Cappuccino app, and externally to and from other Cappuccino instances in other browser, or other applications, when using a text field or a collection view as the source/destination.
This fix switches to full native copy and paste in browsers which support it. This allows copy from anywhere - a collection view, a table view and so on - without the browser greying out the copy menu or beeping.
In the future this can easily be extended to copy and paste images as well.
This change also improves Safari paste support. In recent versions of Safari, pasting from the system clipboard into Cappuccino has been impossible. This remains impossible due to the lack of beforepaste events, but now the code cleanly switches to a Cappuccino-only paste at least.
Without this fix, certain JS contents could cause `- CPDictionary description` and `- CPArray description` to crash.
This change limits how deeply the description code will recurse before returning a default "…" description.
Without this fix, code like `[@{ "a": window } description]` would crash with a "Maximum call stack size exceeded" exception.
Now`CPDescriptionOfObject()` function simply describes the window object as `window` rather than trying to serialise it into a huge description string.
Without this feature, attempts to paste data into a Cappuccino app (outside of a text field) are handled by redirecting the paste into a hidden text field. Then a paste event is generated from whatever is captured in this field after a 0 timeout. This method is prone to timing related bugs where the paste event actually has an empty clipboard.
With this new feature we read the paste data directly from the browser's native paste event when the browser supports it (Chrome 10+, Safari 5+, Firefox 22+). This is much more dependable and may in the future also enable us to read other types of paste data such as images.
This change adds nib2cib support for `NSValue` when the value represents a NSPoint, NSSize, NSRect or an NSRange, which enables user defined runtime attributes of these types to be set in Xcode and then properly decoded in a Cappuccino app.
Without this fix, CGRectFromString would call CGSizeFromString with a value like "{1, 2}}" (not the extra trailing "}"). Although CGSizeFromString didn't complain, it wasn't correct either.
Firefox is not accepting the initial values for these two attributes, making it crash with an ILLEGAL VALUE error in any application using CPBezierPath (especially ones with CPPredicateEditor that is using bezier path to draw the selection arrows.)
Without this fix, trying to build Cappuccino or run a jake script with a too low `ulimit -n` would cause a fatal error.
This change will automatically increase the ulimit when possible and only error out when it's not. The automatic change does not permanently affect the terminal session as it only targets the soft limit.
Previously it was possible to select rows in a CPTableView even if there wasn't columns. It was also possible to make a drag/drop.
This PR fixes theses both problems.
Previously the drag and drop and multiple selection from a void cells caused an exception.
With this fix, you can start a drag/drop and a multiple selection from a empty row.
Fixes#1857
Added a new folder "Resources" in Tools/capp/Resources/Templates/Framework/. As the folder is empty be default for frameworks it contains a .gitignore file to be able to add it to git.
Motivation is to prevent (spelling) mistakes when manually creating a Resource folder for Frameworks (e.g. "Ressources").
Previously the tableView didn't correctly remove a given tableColumn. It was removed in the method load (who is called by the layoutSubviews).
Now the tableColumns is removed before the layoutSubviews (as in cocoa).
It fixed also another problem, before this fixe it wasn't possible to remove all of the columns of a tableView (the last column was always displayed). This is fixed also
Fixes#1913
You can test that with the app of t00f
Previously, if you held down the shift key while making the first selection in a CPCollectionView it would raise an exception claiming {-1, XXX} was out of range. This was because `firstIndex` returns CPNotFound (-1) if nothing is selected.
In Cocoa, holding down the shift key while making the initial selection behaves as if you made an initial selection.
This commit catches a CPNotFound and sets the first index to the index of the item under the mouse pointer.
The program below crashed on the last line with a 'null is not an object' error message. Also added a test case.
@implementation MyObject {
- (int):(int)a {
return a;
}
@end
var a = [[MyObject alloc] init];
var b = [a:3];
Previously, only stderr was checked for response text. In some edge cases it would write an error message to stdout. In addition, the default error message provided when the return status was non-zero and the output was empty was implemented incorrectly.
Now all output from nib2cib is correctly captured.
Also updated help to indicate how to enable debug logging.
- CPImageInBundle() can't be used from a ThemeDescriptor. Replaced with PatternImage().
- DRY! Put clock image sizes in variables so they can be maintained in one place.
- When displayed from the theme showcase, the hand images in HandLayer -setImage: were still _CPCibCustomResources, not sure why. This condition is now explicitly handled by converting to the referenced image.
- Fixed a bug in _CPDatePickerClock -setEnabled where the sublayers would not update unless the event loop was pumped.
- Check for unchanged value in setters.
- Various code optimizations.
- Removed unnecessary method comments.
- Fixed typo in demo app.
Previously when creating a CGPath with an arc, the path was wrong. It added a line when it wasn't necessary and when it was necessary it added a line to a wrong point.
Now CGPath works like in cocoa (If the specified path already contains a subpath, Quartz implicitly adds a line connecting the subpath’s current point to the beginning of the arc. If the path is empty, Quartz creates a new subpath with a starting point set to the starting point of the arc.)
Test app in Tests/Manual/CGPath/AppController.j
'@ref' is generated to ' function(__input…' with the starting space if, for example, the code is 'return(@ref(a))'. This will create a space between 'return' and 'function(__input…'. When the compiler is generate the code this is not necessary.
Previously the showcase of Aristo and Aristo2 didn't work because several bugs in the class CPDatePicker.
This PR fixes these problems :
- The first problem was when displaying a textual datePicker, the class tried to calculate different things for the calendar (good optimization as well).
- The second problem was about the calendar and the hands. It wasn't possible to display a PatternColor made with an image in a layer, now we use a PatternImage and the method CGContextDrawImage to draw the hands.
Canvas supports CGContextDrawRadialGradient as in cocoa. The CGGradientDrawingOptions is not actually supported.
Test app in Tests/Manual/CGCanvasContext
Without this fix, declaring a type with a nil owner would cause a "DEPRECATED: object cannot be nil" warning to be issued.
This fix now allows a nil owner without warning.
Previously, if the browser provided a language string such as "en-US", CPLocale would try to look up the locale for "en_EN", which wouldn't exist so it'd always select the default language.
This fix uses the browser provided language string fully to select the appropriate locale.
When the language code is not known it still sets the current language code
to the unknown code instead of the defualt 'en_US' code.
This makes the test cases for CPDateFormatter fail when run on a foreign system.
Functions are used everywhere to set width or height in the DOM except in the method lockFocus. This will cause IE8/9 to report error if the width or height is negative in this method.
- Added ~/narwhal/bin to executable paths.
- Hopefully clearer error message when an executable is missing.
- Log PATH when executable is missing.
- Catch parse exceptions when parsing parser.j errors, log the returned text.
Previously, if an image or framework in image@framework was missing, it was just a warning, and conversion would continue.
Now an exception is raised and conversion stops.
Shows the Cappuccino project folder. If Path Finder is installed, it says "Show in Path Finder" and that is used.
Reorganized the menu a bit and updated the help.
Previously users saw the generic "-[CPKeyedUnarchiver decodeObjectForKey:]: cannot decode object of class (NSFoo)" message, which was not particularly instructive.
Now the message is: "NSFoo objects are not supported by nib2cib."
Closes#1923
Saving the project requires the xib to be selected again if it was being edited, which is not so nice.
Now we check the project's modified flag before saving and lazy-create the groups.
Previously, attempting to nib2cib a xib with autolayout would give an error message that there was no class to archive NSLayoutConstraint, which was meaningless to most people.
Now an error message that says exactly why it failed and how to fix it is given.
- The new compiler requires the OBJJ_INCLUDE_PATHS environment variable to be set.
- The new XcodeCapp 3 support for Frameworks/Source requires that directory to be excluded from builds.
Previously, CPStepper was delcared using a @class. If CPStepper was not manually imported anywhere else in the app, it caused a crash.
This patch explicitely import CPStepper.j in _CPDatePickerTextField.j
This reverts commit 345a54e8c9, reversing
changes made to 21301cae21.
This commit is reverted because it causes a non obvious crash when selecting the latest row of a CPOutlineView.
The crash is visible in Tests/Manual/CPOutlineViewTest
The compiler can now run in IE8. Also found the use of the function .indexOf on Arrays in CPView.j.
It is not supported in IE8.
Added a new file (OldBrowserCompatibility.js) in the Objective-J framework that will add functions if they are not supported in the javascript runtime.
There are features of Objective-C 2.0 used in XcodeCapp that cannot be compiled with Xcode 4.2, which is the maximum version that runs on OS X 10.6. Users on 10.6 would see compile errors.
Now a check for OS X 10.7+ is done, and on any version earlier than that, the user is presented with a message telling them to download XcodeCapp from the web site.
It turns out it never worked on 10.6. I had to modify mod_pbxproj.py to make it work. In the process I was able to make it faster by skipping conversion to JSON.
Also, a backup of the pbxproj file is made before saving, just in case.
Whew, a complete overhaul of my previous complete overhaul. Changes in no particular order:
- Redesigned the About box.
- Converted tabs to spaces.
- Using Lumberjack for (much better) logging. When debugging, copious logging goes only to Xcode's console, not to the system log. In the release build, only basic info and errors goes to the system log.
- Reformatted the help using Pages, it's a PDF now (so I can use better fonts).
- Massive optimization across the board. All time-consuming operations are done with NSOperation, which uses Grand Central Dispatch, and can be interrupted. So the XCC menu remains fully responsive during source processing.
- Updates to the Xcode project are coalesced so only one read/update/write operation is done.
- Instead of launching a shell and reading the .profile, the target executable is directly launched, but it must reside somewhere in the default binary path + /usr/local/bin:/usr/local/narwhal/bin:~/bin. If jsc, python, objj and nib2cib cannot be found in that path, the user is alerted and XCC quits.
- Added a preference (true by default) to automatically load the Xcode project when a project is opened.
- Added a preference to turn off per-file processing notifications.
- Added a hidden preference to set the log level. Useful for debugging a user's release build.
- Redesigned the Preferences window.
- Renamed some methods/properties, eliminated some unused properties.
- All of the windows remember their position.
- Moved the fsevent_callback into XcodeCapp.m.
- Fixed a bug in mod_pbxproj.py not setting the type of PBXFileReferences which are directories to "folder".
- Added support for categories in parser.j. Woo-hoo!
- Rewrote pbxprojModifier.py to use a class.
- Renamed the Xcode project group names to "Cocoa Classes" and "Cappuccino Source". Within "Cappuccino Source", framework code is kept in a "Frameworks" group.
- Shadow filenames are all project-relative now, thus *much* shorter.
- pbxprojModifier.py keeps the groups and files in the project in sorted order: Resource folders are at the top, followed by Cocoa Classes, followed by Cappuccino Source. Within Cocoa Classes, non-framework files come first, followed by framework files, followed by xcc_general_include.h.
- If either .XcodeSupport or the .xcodeproj is missing, the other is regenerated to ensure they stay in sync.
- A compatibility version for .XcodeSupport is stored inside it in Info.plist. If that version < the app's compatibility version, the project is reset. This ensures that format changes in the future will not result in projects in an unknown state.
- If the Xcode project cannot be opened, the user is alerted and given the option of regenerating the project.
- Resetting the project deletes all .cibs to force the .xibs to be regenerated.
- All possible FSEvents are dealt with individually now, and in a way that (hopefully) maintains sync between Cappuccino and XCC.
- A file descriptor to the Xcode project is kept open so that If the project path changes it can be relocated. If it does move, the user is alerted ad given the option of reloading the project or quitting.
- If one of the watched paths changes, the project is reloaded.
- Xcode creates temporary files, they are properly filtered out now when handling FSEvents.
- NSRegularExpression was being used before, but that is OS X 10.7 only. NSPredicate is used instead now.
-
Previously, tooltip system was using bubbling event. This was actually making impossible to see a view's tooltip if the parent view also has a tooltip. This patch changes the capturing mode from bubbling to capture.
Previously it wasn't possible to display a toolbar in a sheet window.
Now, a toobar is displayed well in a window. This PR allows every sheet window to have a toolbar, with the PR #1918 just the CPTitleWindow/HUD sheet could have a toolbar as in cocoa.
Fixes#1863
Previously when using a sheet window in a fullPlatform window, the sheet window started the animation under (the layout) the toolbar.
Now, the sheet animation starts below the toolbar as in cocoa.
It also fixed that only a CPBordelessWindow/CPTitleWindow/HUD can have a toolbar.
Previously, when a row with a disclosure control (triangle) was selected, the row was selected
but the control remained in the unselected themestate. Disclosure control themestate was only being
changed when the control itself was clicked on. This fix changes the controls themestate whenever
the row selection changes.
Previously, the tooltip of a control was always shown on the primary platform window. This fixes sets the tooltip window's platform view to [[targetView window] platformWindow].
Now if either .XcodeSupport or the .xcodeproj is missing, both are recreated to ensure they are in sync.
Also, when trying to open the Xcode project, if it is missing or cannot be opened, an alert is displayed and the user is given the option of regenerating the project.
Some code cleanup as well.
Previously, a CPAlert was always shown on the main platform window. This behavior could result in a feeling that the application was hanging because of the modal nature of the alert when users are looking to another platform window. This fix ensure to set the CPAlert window's platformWindow to the current key window using [[CPApp keyWindow] platformWindow].
Previously, when opening a popover on a view from another platform window, the popover
was actually opened in the main window. This fix ensure to set the correct platform window
to the _CPPopoverWindow by setting it to the [[positioningView window] platformWindow];
The first defense against this is to validate the recent menu items to ensure they exist and are directories. If not, the item is disabled.
The next time a project is opened, the missing project is pruned from the recent menu.
The second defense is a check when an attempt is made to open a project.
When added, level of child window is reset to level of the parent.
A user may change it aftewards to the level that may even break
specified relative position.
E.g. if a child window is added below the parent (Normal)
and its level is set to Floating it will be ordered above.
Whenever you change level of the parent window, levels of the child
windows are reset as well ignoring all your changes.
According to the example, if you set level of the parent to Normal+1,
child window' level will also be set to Normal+1 and its relative
order to the parent (below) will be restored.
FILE.exists was being used to test for the existence of the symlink in /Applications, but that doesn't work.
The correct way is to use ln -sfh. Note that the -h was missing before, which prevents an existing target symlink from being followed, which is why the symlink would end up inside XcodeCapp.app.
Previously the CPDatePicker freezed when trying to hide it when it was the firstResponder.
There was an intern problem in the CPDatePickerTextField about the nextKeyView. That's why there was a boundless loop and the freeze.
This PR also fixed a big about the previous and next textField in the CPDatePicker.
It also fixed a bug about the CPDatePicker when it becomes firstResponder.
Fixes#1909
When saving a xib, Xcode creates a temporary file with the name <filename>~.xib. The file event listener would sometimes pick this up and XcodeCapp would attempt to process it, but by the time nib2cib was launched it would be gone, generating an error.
Now .xib files whose filenames end with ~ are ignored.
- Renamed XcodeSupport to .XcodeSupport.
- Added "Reset Project" to menu, it closes the project in Xcode if necessary, removes .XcodeSupport and .xcodeproj, then reloads.
- Added separate preferences for showing errors panel on warnings/errors. You may not want to see the panel if there are category warnings.
- New status icons (again).
- If an error occurs, the status icon turns red when all processing is done.
- Reworked notifications. Objective-J parsing is notified when done. Xib conversion notifies at beginning and end, because it typically takes a few seconds. Individual file notifications are suppressed if a project is being loaded.
The effect is that each minor point release (3.1 -> 3.2) will cause the help to display the first time it is launched, so that the user is encouraged to find out what is new.
XcodeCapp would add a folder reference in the project for user framework Resources directories in Frameworks/Debug, but it didn't add references for framework Resources directories in Frameworks/Source.
- Accidentally deleted the Jakefile. It has been restored, and fixed the case where it would put a symlink to XcodeCapp.app within XcodeCapp.app.
- Minor changes to remove compiler warnings.
Previously any .xcodecapp-ignore file within the monitored directories would be parsed. This commit ensures that only the project's root .xcodecapp-ignore file is parsed.
Previously the number of recent projects was unbounded, which made the recent menu unwieldy.
Now the recent projects list is pruned to the maximum set in the preferences, which defaults to 20.
- Updated project to Xcode 4.
- Reorganized files into preferred Xcode 4 hierarchy.
- Updated code to use properties.
- Eliminated unused/redundant code.
- Significantly optimized project scans.
- Added support for outlets, actions and resources in user frameworks.
- Now using bindings throughout.
- New application icon, Retina ready.
- New status menu icons.
- Reorganized menu.
- Added Preferences… menu item.
- Parsing errors and nib2cib errors are actually logged to the errors panel now.
- When an error occurs, the errors panel automatically opens, unless the "Automatically open Errors & Warnings panel" preference is off.
- Double-clicking an error or selecting an error and clicking Open opens the error in the preferred editor for .j files (or Xcode for .xibs). The cursor jumps to the offending line in the following editors: Sublime Text, TextWrangler, BBEdit, TextMate, Chocolate, MacVim.
- The Errors panel now remembers its position/size.
- When a project is loaded, the errors panel is cleared.
- When a file is modified, any pre-existing errors for that file are removed.
- Updated to Growl framework 2.0.1.
- If Notification Center is available (OS X 10.8+), that is used instead of Growl.
- .xcodecapp-ignore now supports ignoring a directory by suffixing the name with "/".
- .xcodecapp-ignore now supports include expressions prefixed with "!".
- Significantly optimized filename matching against ignored paths.
- .XcodeSupport has been renamed XcodeSupport to make it visible, so that trashing it is easier if things get out of sync.
- The template Xcode project now contains no frameworks or targets, so Xcode does not show any warnings or errors, and the user cannot accidentally try to build.
- Removing a file now removes the file from the Xcode project.
- Shadow files replace forward slash with U+2215 (DIVISION SLASH), which looks like forward slash but is a character that is extremely unlikely to be in a filename.
Previously error messages would always be colorized, which doesn't work well if you are piping the output.
Passing --no-colors turns off all colors in the output.
Previously, the parameters to CPColor colorWithHue:saturation:brightness: could be specified below 0 or above 1.0.
Now the values are clamped so that values below 0 are taken as 0 and values above 1.0 and taken as 1.0.
Previously, - CPColor hsbComponents returned degrees and percents. With the latest change to - CPColor colorWithHue:saturation:brightness: this was inconsistent.
Now - CPColor hsbComponents returns its components as factors in the 0-1.0 range.
Previously, + CPColor colorWithHue:saturation:brightness: took its parameters as degrees and percent (e.g. hue between 0º and 360º, saturation and brightness between 0% and 100%). However, in Cocoa the components are given as factors (0.0-1.0) like other colour methods.
This fix makes + CPColor colorWithHue:saturation:brightness: work like in Cocoa.
In the previous patch the shouldSelectRow delegate was being stored and only called once. This changes its behaviour again to first fire the delegate to see if the row in question can be selected. If not, it will look for the next row (up or down) that may be selected and choose that one.
This commit fixes a typo in CPTableView. `_contentBindingExpicitelySet` should be `_contentBindingExplicitlySet`. The misspelling was consistent, so this should have no functional change.
Previously, navigating through the rows of a tableview with the arrow keys would cause the delegate method `- (BOOL)tableView:(CPTableView)aTableView shouldSelectRow:(int)rowIndex` to fire twice. This was due to the delegate method being called twice in checking whether the row could be selected.
This commit changes it so that the delegate method is called just once and the result stored.
This commit fixes the delegate documentation for `- (void)tableViewSelectionIsChanging:(CPNotification)aNotification`, removing reference to "selection chaining." (should be "selection changing")
This replaces the previous solution with one proposed by @BlairDuncan. It uses the _highlightItemAtIndex: method to maintain the highlight state, rather than setting the _highlightIndex variable directly.
- Added support for CPDateFormatter in Foundation
- Added support for CPTimeZone in Foundation
- Added support fo CPDateFormatter in nib2cib
Test app in Tests/Manual/CPDateFormatter
UnitTest in Tests/Foundation/CPTimeZoneTest.j
UnitTest in Tests/Foundation/CPDateFormatterTest.j
When reloading the table view, the actual loading (-load) is defered
until layout is needed (generally in the next run loop). This is an
advantage because it minimize reloads but in some case it is necessary
to force a reload, for example when we need to access data views, or
manually edit a view, or when we explicitely ask for a reload.
This commit adds _reloadDataViewsImmediately and make use of it when
necessary.
Tests: AppKit/CPTableViewTest -> -testEditCell
Cell-based table views with no identifier set for the table column:
Before this commit, the caching system was asking for a view identified
by the tableColumn UID. If the table column data view was changed
externaly, the previoulsy cached data views were loaded instead of the
new ones. The views are now identified by the -dataview UID.
Test: AppKit/TableTest -testLayout were a custom data view is set.
Previously when a menu item was instantiated outside of the context of the menu itself, its highlight state was maintained even after it was removed from the menu. If the menu was dismissed with the item highlighted, and then the item was used in another menu (e.g., re-added to a context menu for a table row) it would appear highlighted, even though the menu highlight index was not set.
This commit ensures that when removing items from a menu that the highlight state of the underlying view is set to NO as well.
Tests for this behaviour are included as well.
Fixes#1899
Previously the default About panel was constructed in a CIB. Additionally, an Xcode-based XIB/CIB file was constructed to replace it.
This commit creates the default About panel in code and removes the additional CIB/XIB files in the AppKit/Resources panel.
* State *
Previously the Converter class needlessly copied a whole bunch of state from the Nib2Cib class, which was not only poor factoring, but made for easy omissions of state when instantiating a new Converter, such as was done in NSNib.j.
This commit pushes all state not specific to Converter up to Nib2Cib. This allowed removal of redundant code in NSNib -NS_initWithCoder.
* Resources *
Previously it was not possible to use custom images that were located in a framework, because only the app's Resources directory was searched.
This commit automatically searches all frameworks for Resources directories. When a custom image is specified in Xcode, first the app's Resources directory is searched, then all framework Resource directories. Alternately, the specific framework to use can be specified in Xcode by using the form framework@image.
At runtime, the framework is loaded by its identifier (as specified in Info.plist), and if there is no identifier, by searching next to the current AppKit framework.
All of this renders the nib2cib -R option completely obsolete, so it was removed from the NibApplication/Jakefile template.
This commit adds notifications for focus and blur on CPTokenField.
Attached is a modified Manual CPTokenFieldTest that observes these notifications.
Fixes#1396
This fix adds two keys to the default Info.plist in the capp gen template for NibApplication, "CPBundleVersion" and "CPHumanReadableCopyright". These are introduced primarily to fill out the fields in the default "About" panel. (see #1896)
With the death of Atlas CIB files can no longer be edited directly. This commit replaces a CIB-only version of "AppKit/Resources/AboutPanel.cib" with a XIB-derived one that can be edited in Xcode.
This commit also moves some of the formatting of the about panel into the IB file (c.f. the FIXME note).
Previously no default principal class was declared in Info.plist for a framework.
This commit declares the principal class to be the same as the main framework class created with capp gen -t Framework.
Previously, if a CFBundle with the given identifier had been loaded, but no CPBundle had yet been created with the given identifier, -bundleWithIdentifier would return nil instead of a newly constructed CPBundle with the CFBundle's URL.
This commit correctly returns a newly constructed CPBundle when a CFBundle is found.
CFBundle internally stores its URL in relative form, and returned that relative URL in the bundleURL() function. However, in Cocoa an absolute URL is returned. It is reasonable to assume an absolute URL would be expected by some asking for the bundle's URL. Even within CFBundle, most times bundleURL() is called it is followed with absoluteURL().
This commit returns an absolute URL from bundleURL().
Previously the CPDatePicker didn't work with the action and target. Now each time the dateValue changed, the given action is sent to the target.
This commit fix also a bug with moving with tab and the arrows. Before, the left arrow worked liked a tab when the firstElementTextField was selected in a textual CPDatePicker.
Test app in Tests/Manual/CPDatePicker.
When a menu is active, menu selection should change as characters are typed.
This can be seen in the manual CPMenuTest.
It used to work in the past but with all of the changes in the last few months to the compiler I
was not able to track down the exact commit that broke it. The code responsible for
clearing the _keybuffer after a brief delay in typing, was not being called,
resulting in a build up of characters.
This commit moves the check for delay to the interpretKeyEvent and takes care of
clearing the _keybuffer itself and removes that responsibility from the selection method.
Previously, when a text field became the key responder, first it was scrolled to visible, then a check was made to see if the text field was completely within the platform window bounds. The scrolling to visible would unfortunately cause problems because of the interaction between the blur handler in text fields and controls that can be clicked without becoming first responder.
This commit completely forbids text fields from becoming first responder if they are not completely within the platform window. That eliminates any possibility of unwanted browser scrolling of the text field.
This commit fix a bug where declaring the binding option CPNullPlaceholderBindingOption was actually setting the
actual value of a text field instead of just its placeholder. This commit adds a test to determine if the binding source
implements setPlaceholderString:. In that case, we let the value to be nil.
Fixed: Starting a column drag is now faster.
Fixed: When dragging a selected column, selection is now drawn on the dragging view and the cursor is the closed hand.
Fixed: When dragging a table column, underlying columns were sliding according to the tracking location instead of the column lateral edges.
Fixed: In CPTableView, the drop indicator for rows could appear when dragging a column.
This commit creates directly the dragging column instead of relying on
built-in drag&drop. Also fixes a bug where the drop indicator would appear when
dragging a column if some rows were previously drag&dropped.
Fixed: An edited view no longer send its action (view-based) or commit its object value (cell-based) when unexposed.
Fixed: Cell based tables support for editing any control, not just textfield and buttons.
This commit reverts #1478. It appears it is a bad idea to enqueue
visible views. The views generally stay in the cache forever and they
are repeatedly asked to be removed from the table at each load, causing
a performance penalty.
Fixed: Column dragging performance, content binding performance.
After this commit, -reloadDataForRowsIndexes:columnIndexes: reloads the data and the data only.
The -reloadData method no longer tries to reuse the views, instead it just reloads the data for visible views.
To flush and reload the views cache for visible rows, use _reloadDataViews.
To internaly layout the views geometry, use _layouViewsForRowIndexes:columnIndexes:
In CPTableView dragging column code, relayout the views whose frame changed instead of reloading everything.
New: preparedViewAtColumn:row:
These methods allow to enumerate visible data views or data views in specified columns and rows.
-enumerateAvailableViewsUsingBlock: is the counterpart of cocoa's -enumerateAvailableRowViewsUsingBlock: except that it enumerates data views instead of CPTableRowView and the block has an additional column parameter.
This commit reverses the way views are stored and accessed in the table view: rows>columns instead of columns>rows.
Applied the second method where it is relevant, making the code more compact and readable.
Previously it wasn't possible to theme the main menu of the app. With this fix the user can either theme the main menu with the theming system or with the method +setMenuBarAttributes from CPMenu.
Previously, the themed text values for controls were set directly on the control, which made modifying the theme values for text consistently across the controls difficult.
This commit introduces several file-scoped variables that are used to set the text colour, the disabled text colour for both regular and default controls, as well as a consistent text shadow colour for each of these as well.
Previously, the text shadow offset on a control that was active (e.g., a button that was set to default, or the active segment of a segmented control) would have a text shadow applied to it if the button was also disabled.
This commit negates the text-shadow offset so that the text appears correctly.
This commit defines file-level variables for the disabled text colour, the disabled text shadow colour, and the placeholder text colour. It also adds or replaces these values in the controls where appropriate.
Previously, setting a CPComboBox to disabled would not dim the text. This commit fixes the theme state to include a setting for the text colour and text shadow colour.
Previously some disabled controls would be set with an opacity of 0.35, which resulted in controls that were visually too light.
This commit changes the default disabled opacity to 0.5.
Refs #1788
Previously, the enabled/disabled control for NSTextField in Interface Builder was ignored by nib2cib.
This change allows the enabled/disabled state on CPTextField to be set in interface builder.
Previously the Aristo2 checkboxes used dedicated images for disabled controls.
This commit removes the disabled controls and allows the disabled state to be set via CSS opacity on the "regular" checkbox controls.
Previously the Aristo2 checkboxes used dedicated images for disabled controls.
This commit removes the disabled controls and allows the disabled state to be set via CSS opacity on the "regular" checkbox controls.
Previously, when
- Dragging a row in the rule editor
would eventually try to generate a dictionary with key 'indexes' and a
nil value in the method __postRowCountChangedNotificationOfType:indexes:.
Generating an empty dictionary instead when there is no indexes solves
this issue.
Previously, NSNib would create a Converter instance for converting table cell views. Unfortunately Converter did not expect to be instantiated more than once, and the singleton SharedConverter was unconditionally being set in the initializer. This might not have caused problems, except that NSNib did not copy all of the shared Converter's state to the copy, especially the resource path, which caused custom images in table cell views to fail.
This commit correctly ensures that the shared Converter instance is set only once, and also copies all of the shared Converter state to the instance that is used for converting the table cell view.
Long ago it was required to use the -R option with nib2cib if custom images were used. nib2cib now infers the resource path, so there will always be at least a default resource path. Thus the check for no resource path was unnecessary.
This commit removes that check.
Previously, if properties were bound to table columns in a cell-based table, and an object was added/inserted using CPArrayController, a warning was generated about a missing observer.
This commit is a temporary fix, the real fix is to implement more granular notifications during didChangeValueForKey:, which will eliminate the source of this problem.
Fixes#1781.
Previously, when a view-based table:
- Had no delegate.
- Used bindings for its data source, thus obviating the need for makeViewWithIdentifier:owner:.
it would eventually call CPCib -instantiateCibWithOwner:topLevelObjects: with a nil owner (the nil table delegate). This method would try to construct a dictionary with the nil owner, which would generate a deprecation warning, and will in the future fail altogether.
In Cocoa, this is exactly what happens, the owner is nil. And the documentation for -instantiateNibWithOwner:topLevelObjects: clearly states that the owner may be nil.
This commit only adds the owner to the name table dictionary if the owner is non-nil. Since objectForKey: with a non-existent key returns nil, this is the same as storing nil for that key.
For example, in CPTableView before this change, dragged rows could not move up if the table view was near the top of the platform window.
Fixes#1877
Test TableTest/ViewBasedCib/ and drag up the third row.
This bug appears only if a table view have been added to the window.
CPTableView observes first responder changes and moves up in the view
hierarchy to determine the edited data view. If for some reason, a
superview in the hierarchy was nil, we entered an infinite loop.
This commit adds guards to the recursive method that searches the data
view. Also added an early return when the fr changes to a view outside
the table.
Fixes#1875
Previously, CPArray did not allow binding to the collection operator @count. In Cocoa this is allowed.
This commit allow the key path @count to be observed, so that @count can be bound to.
Formatting fixes as well.
Previously, XCC would indiscriminately add files to the Xcode project via the pbxprojModifier.py script. This resulted in duplicate references to a file.
This fix adds a check to see if the file is already included in the project's sourceGroup folder, and will return None if it is.
Fixes#1866
In this commit I've added the currentDocument method to the document controller
Also removed the firstEligibleExistingWindowController method from CPDocument
It did nothing but add confusion as it was not used anywhere else in the frameworks.
And it is not in cocoa.
Example:
#define first Martin
#define second Carlberg
var first##second = 13;
Compiles to:
var MartinCarlberg = 13;
Example 2:
#define _function(inline) function inline { return _##inline; }
#define _CGPointMake(x_, y_) { x:x_, y:y_ }
_function(CGPointMake(x, y))
Compiles to:
function CGPointMake (x, y) {
return {x: x, y: y};
}
Previously, passing nil to the first or last argument of getLineDash:count:phase: would cause a crash.
With this change, these references are simply not written to if nil.
Previously, setLineDash:count:phase: took references for its arguments to match getLineDash:count:phase:. This was however wrong and is not what is done in Cocoa.
This fix makes the setLineDash:count:phase: calling convention the same as in Cocoa.
This fix should in theory make the phase argument work in Firefox, although in Firefox 19.0.2 it doesn't seem to work. Hopefully it'll kick in in a future version of Firefox.
Previously, the enabled state of a CPComboBox could not be set in Interface Builder.
This commit allows Interface Builder to set the enabled state of a CPComboBox.
A modified test in Manual/CPComboBoxTest is included. This test simply adds a new disabled combo box to an existing window, and then checks to see if it is disabled, printing a descriptive message to the Javascript console.
This commit changes the current XCodeCapp menubar icon to use higher contrast colours, making it easier to see when XCC is working or idle.
Previously, the colours were black and dark green. With this change, the colours are black and orange (taken from the Cappuccino logo).
This commit includes the original PSD artwork as well as rendered versions of the icons in PNG format.
Previously CPTableView -_init was trying to reference a theme value, but that doesn't work because the theme creates a CPTableView and during that _init the theme value is not yet defined.
This commit uses lazy instantiation to avoid this problem.
Also, column selection was added to the CPDictionaryControllerTest to better visualize unfocused selection highlight colors.
Previously, the colors used for unfocused selection highlighting were hard coded. In Cocoa, they seem to be desaturations of the focused selection highlight colors.
This commit calculates the unfocused highlight colors from the focused colors.
It also removes some dangling commas.
When a statement ends without a semicolon and a Send message expression is on the next line it was wrongly parsed as a single Send message expression. It should be an ExpressionStatement with a Send message expression inside.
Previously, CPTableView was unconditionally setting the selected theme state for selected data views. When the selection highlight style is CPTableViewSelectionHighlightStyleNone, there is no row selection highlighting done. Since the default theme specifies white text for selected data views, we ended up with white text on a non-highlighted background, which was invisible on non-alternate rows and almost invisible on alternate rows.
This commit unsets the selected theme state flag for any redrawn data views when the selection highlight style is CPTableViewSelectionHighlightStyleNone, thus the text color does change at all when selected.
Fixes#1847
Previously, when a table cell was edited, the table thought it had lost focus and displayed the cell's row highlight in an unfocused state.
This fix updates the _isFocused test to include tests for cell editing in both cell-based and view-based tables.
Previously the gradient in an unfocused selection highlight in CPTableViewSelectionHighlightStyleSourceList style was darker at the top and brighter at the bottom, which is the opposite direction of the gradient when focused.
This commit makes all of the unfocused highlight colors a strict 5% desaturation of the focused colors.
Refs #1839
In CPTableViewSelectionHighlightStyleSourceList selection style, a table view draws the highlight on each row with a top line, linear gradient, and bottom line. Previously, the top and bottom lines were quite dark in relation to the gradient.
This commit makes the top and bottom line brighter, to make the transition to the gradient more subtle.
Refs #1839
CPTableView has no focus ring, and currently gives no visual feedback of its firstResponder status. In Cocoa when the table view resigns firstResponder or its window loses key status, the selection color turns gray.
This commit sets the selection color to gray when a table view resign firstResponder or its window loses key status. A second window and a couple of textfields were added to the CPDictionaryControllerTest to demonstrate the changes.
Fixes#1839
Previously the selection highlight update loop created two selectors and called an extra method for every row being updated.
The selectors are now calculated beforehand, and the method call has been eliminated.
Previously the main menus were farther apart than the Aristo 2 PSD indicated, and too far apart visually.
This commit reduces the horizontal margin between menu items that have originated from Interface Builder. A value of 9.0 for horizontal margin (down from 12.0) reduces the margin on both sides of the menu by 3px, reducing the inter-menu spacing by a total of 6px.
Fixes#1809
Previously, no check was done to ensure a next/previous key view belonged to the same window as the view to which it was being chained. This could lead to an infinite loop.
This commit will only set the next/previous key view if the proposed's views window is either nil or is the same as the receiver's window.
Closes pull request #1851
Previously, if a pull down menu was brought down from a pop up button in pull down mode, clicking the button again would close and reopen the menu.
With this fix the behaviour becomes like in Cocoa, and clicking the button a second time simply closes the menu.
This was caused by the menu opening up below the button. When the menu manager examined this event it saw a click outside of the menu and immediately closed the just opened menu. Then it put the opening event back on the event queue, causing the pop up button to try to open its menu again.
The fix ignores the opening event for purposes of detecting clicks which should close the menu.
Refs #1833.
The class method +themeAttributes returns a dictionary, so CPNull has to be used instead of nil. Previously, if the class theme attribute was used, there was no attempt to convert CPNull to nil, which could have unintended consequences.
Previously more work was done than necessary to initialize custom table text cells.
Now the theme state is set before initializing table cells, which allows fonts and colors to be set correctly. The selected colors still have to be set explicitly.
BREAKING CHANGE:
Because this operates within nib2cib, all xibs should be re-converted.
Previously in CPTableView, after clicking a sortable column, data views of the sorted column ended up in a selected state and a row was selected instead of the sorted column. As a result, the text of the rows would turn white as if they were selected, but there was no highlight, so the text was basically invisible.
This commit fixes that by reversing the order of operations when clicking a column. First the sort is done, then the selection.
Fixes#1337, #1688, #1779.
-tableView:dataViewForTableColumn:row: and -outlineView:dataViewForTableColumn:item: are deprecated in favor of ...viewForTableColumn:...
Using the previous delegate API and not caching the view with an identifier was degrading performance. With this change, developers are encouraged to use the new caching system and the CPTableView method -makeViewWithIdentifier:owner: to get the view.
Also fixed a condition where a view-based outline view was not always asking for the view from its delegate. In some circumstances, an outline view could be considered as view-based instead of cell-based.
Test for deprecated delegate methods in CPOutlineViewViewBasedCib and TableTest/ViewBased examples.
Fixes#1823
Previously _CPThemeAttribute had an ivar called _parentAttribute, which represented the attribute from the theme descriptor, as opposed to the attribute for a particular instance of a view.
This commit renames that ivar to _themeDefaultAttribute, so it is clearer where it gets its values.
Previously, if a cell-based table had a custom font or color in a table cell, when the cell was selected it would not change color. Also, edited cells did not maintain the original font/color.
This commit ensures that table cells uses the theme setting for selected text color, while maintaining all font and color attributes when selected or being edited.
BREAKING CHANGE:
Because this fix operates at the level of nib2cib, xibs that use customized cell-based table text cells should be reconverted.
It used to be part of the theme to make the text in selected table rows bold. I never quite understood that, it really isn't necessary visually. Cocoa doesn't do it, I don't know of any toolkit that does.
Also, fixed sizes were used for text field theme fonts, which did not allow them to track runtime changes to the system font size.
This commit makes selected table row text plain, so selecting a row does not change the metrics of the text, just changes the color, which is much more pleasing visually.
Also, the special CPFontCurrentSystemSize constant, which tracks the actual font size at runtime, is used for text field font sizes.
CPThemeStateSelectedTableDataView and CPThemeStateSelectedDataView were the same, and were not used in any consistent or meaningful way.
This commit eliminates CPThemeStateSelectedTableDataView, only CPThemeStateSelectedDataView is used now.
Previously in CPTableTView, a view being edited was not cached when unloaded, for example when made invisible after scrolling or when calling -reloadData.
This situation was causing a duplicated view to be added at the same place when the data view was loaded again.
After this commit, edited data views are cached and resign their first responder status just before they are added to the queue because they can be reused at another place.
Test: Added a "reload data" button in ViewBasedCib example.
Previously the sheet was made key at the beginning of the animation. It turns out events would not reliably make it to the sheet during animation. The other problem is that if a text field was the first responder, at the beginning of the animation it is offscreen, which means it refuses first responder and sets the first responder to nil.
Now we no longer try to make the sheet key before the animation ends, avoiding the text field problem.
Previously, popovers did not properly stay below the menu bar.
This commit ensures they stay within whatever is the usable content frame for the platform window.
When a transient popover was closed by clicking, in some cases it would not detach from its parent, leaving it in the window list. This caused problems later on.
This commit ensures the popover is detached from its parent when it is ordered out.
Previously, if a window had a sheet, a mouse move event was passed to the sheet and then filtered. This disabled resize cursors for the parent window of a sheet.
This commit allows mouse move events to continue to the parent window so that resize cursors will work on the parent window as well.
Sheets are like child windows, they can never become main. Previously, when an attempt was made to make a sheet the main window, its parent window did not become main as it should have. Thus the title bar would remain dimmed, even though the sheet was the key window.
With this commit, when a sheet is made main, it defers to the parent window.
Previously, if an event occurred during the sheet opening animation that would close the sheet, the animation was not properly cleaned up and the subsequent closing animation would die horribly.
The animation is always cleaned up with this commit.
It is possible to resize the parent window of a sheet while it is open. Previously, the sheet's top shadow was not adjusted when the parent window was resized, which could lead to it being out of synch.
Now the sheet shadow is adjusted whenever the parent window's size changes.
Previously, sheets were only repositioned if the parent window's setFrameOrigin: method was called. This did not cover all possible cases for parent window repositioning.
Now the sheet is repositioned any time the parent window's origin changes.
Previously, Cappuccino was using preprocessor macros internally for the CGPoint/Size/Rect/Inset/Affine functions, as well as for CPRange. These macros had the same name as the corresponding function, but began with _. The functions were actually defined using the macros.
The motivation behind using macros was to increase performance by reducing function calls. However, there were a number of problems with this approach:
- There was an artificial dichotomy between _CG macros and the corresponding CG functions. We never completely replaced CG function calls with _CG macros. In fact, they were often mixed up in the same file. There was an extra burden on the programmer to remember to use the macro instead of the function.
- If a method call was passed as an argument to a macro, performance could actually be significantly *worse* than a function call. For example, _CGGetRectMakeCopy([view frame]) would expand to `{ origin:{ x:[view frame].origin.x, y:[view frame].origin.y }, size:{ width:[view frame].size.width, height:[view frame].size.height } }`. So instead of a single objj_msgSend and a single simple function call, we ended up with 4 objj_msgSend calls, which are way more expensive than simple function calls.
- Because of this expansion problem, to use macros efficiently required us to remember to use variables for all macro parameters. This didn't happen, and shouldn't have to happen.
- Finally, with modern Javascript engines, function call overhead is so small that it really isn't worth using the macros.
This commit eliminates the _CGGeometry, CGAffineTransformation and CPRange macros and replaces them with function calls.
BREAKING CHANGE:
The macros are no longer available. They could only be used with compiled code, but if there is any user code that used them, they will have to be replaced with the corresponding functions.
I'm happy for those who have 27" screens, but on my poor MPB 15" screen, the CPGraphicsTest window was too tall to see everything.
I split the window into two windows so everything can be seen.
The AttachedSheet2 demo was calling runModalForWindow:, which caused the sheet to appear briefly as an application modal.
That call was removed, beginSheet: does everything.
Also:
- Wrapped _attachSheetWindow method into _attachSheet:modalDelegate: method.
- Just to be safe, in CPWindow -_sheetShouldAnimateIn:, the sheet is moved offscreen before being ordered front.
Previously, clicking outside a menu on a control would not close the menu. Also clicking on the menu header again would open a second menu.
This error was caused by the control receiving the click and starting its own tracking, depriving the menu of the mouse up event it was awaiting.
The behaviour to wait for the mouse up was wrong to begin with. Menus should close on mouse down.
This fix closes the menu on left mouse down anywhere but the menu, while still allowing the click to pass through to any control underneath the cursor.
Fixes#1833.
Without this fix, CPTabView would still make view hierarchy changes in `initWithCoder:` despite the fix in f805b6b. Making view hierarchy changes before a cib is fully decoded can easily result in a corrupted hierarchy.
In this case, the decoding of a tab view item view, A, lead to the decoder to try to decode A's next responder, which was the tab view, B. B then tried to set the selected tab, which involved adding A as a subview. But since A wasn't done decoding, A._superview would change later in the process, corrupting this change.
This fix delays `_updateItems` to `awakeFromCib` at which time changing the view hierarchy is safe.
Refs #1409.
Currently, if a window controller has a document, its title is put in menubar. I cannot remember any Cocoa application that ever did this.
This commit removes the synchronizing of the document title with the menubar title.
After bbfaac5, all kinds of strange behavior occurred with window sizing and moving due to overzealous constraining.
Cocoa does not constrain the frame of hidden windows. When a hidden window is ordered in, it is constrained to the usable screen content rect. A visible window has its height constrained when its frame is set. Its origin is constrained such that a minimum margin at the left, right and top is visible, and the top is constrained to be below the menu bar.
This commit fixes a number of problems related to window moving and resizing that were introduced by bbfaac5:
- Sheets are not constrained at all.
- Moving a window does not constrain its size.
In addition, when a window is moved, tracking is relative to the initial click point, which provides proper behavior when the movement is constrained and the mouse keeps moving.
Previously, there were two instances in which a browser would forcibly scroll a text field into view, out of Cappuccino's control:
- A text field is first responder in the key window, is partially or fully offscreen, and you click somewhere else within the same window. If the target of the click does not accept first responder, the text field is refocused.
- A text field is first responder in a non-key window, is partially or fully offscreen, and you make its window the key window. In that case the text field is made first responder and is focused.
In both cases, focusing the text field causes the browser to scroll the viewport out of Cappuccino's control such that the text field is completely onscreen.
With this commit, before a text field is focused, it is checked to ensure it is completely within the usable content rect of the platform window. If not, it refuses first responder. If the window is becoming key, the first responder is set to nil.
Previously, all window subclasses were constrained to the usable screen content. Unfortunately that included the menu bar window, which ended up appearing below itself!
This commit introduces an ivar that indicates whether instances of the window subclass should be constrained or not.
Previously, no bounds were placed on the placement or sizing of windows, which allowed them to extend beyond the usable screen content (the area bounded by the menu bar at the top and the platform window on the other sides).
Cocoa enforces the following restrictions:
- When a window is ordered in, it is forced within the usable screen content (below the menu bar) and its size and width are clipped to the usable screen content as well.
- If a window is already visible, any methods that move or resize the window will ensure the top and bottom of the window are within the usable screen content. The width is unchanged.
This commit implements Cocoa's restrictions in CPWindow.
Fixes#1690
Previously, calling setPostsFrameChangedNotifications:YES or setPostsBoundsChangedNotifications:YES would immediately result in the notification being sent. Cocoa will only send in this context if:
- setPosts[Frame/Bounds]ChangedNotifications:NO was called.
- The frame/bounds changed between that call and the call to setPosts[Frame/Bounds]ChangedNotifications:YES.
We are currently not tracking pending changes, so this commit removes the sending of the notification completely in those methods.
Fixes#1076
Previously, Cappuccino would resize in such a way that a view's size ratio to its superview would not be maintained.
This commit ensures that size ratios are maintained during resizing.
Patch contributed by @davidkhess
Fixes#357
Previously, when a sheet was resized, there were a number of problems:
- Sheets in Cocoa resize their width symmetrically, such that they always remain centered. This was not happening in Cappuccino.
- When a sheet is resized wider than its parent window, the shadow at the top of the sheet was growing wider than the bounds of the parent window's title bar, which makes no sense since the shadow is supposed to be cast by the title bar.
- Sheets had no top border, so when they were resized wider than the parent window, the top edge looked like it was cut off. In Cocoa, sheets have a top border that sits under the title bar of the parent window, and which is visible when the sheet is resized wider than the parent window.
This commit addresses these problems as follows:
- Sheet width is now resized symmetrically.
- Sheet resizing is pinned to the screen bounds.
- The top sheet shadow is pinned to the width of the parent window's content view.
- Sheets now have a top border.
In addition to these changes, a redundant sheet top shadow was eliminated from one of the _CPWindowView subclasses.
Fixes#1846
Previously sheet animation was performed by sizing the sheet from zero to full height when animating in, and then back to zero height when animating out. This caused problems with the resize algorithms, which are not designed to deal with sizing from zero.
Analyzing Cocoa behavior revealed that sheets slide in and out with no resizing.
With this commit, a general purpose mechanism for clipping a window to a rect was added to CPDOMWindowLayer. Clipping was put in CPDOMWindowLayer because that is the class that deals with windows at the global level. CPWindow uses the clipping mechanism to slide the sheet in and out, thus avoiding all resizing problems.
Fixes#1840
Previously the attached sheet shadow (that appears below the title bar of the window a sheet is attached to) was 9x8, and it was quite dark.
With this commit the width has been reduced to 1 (to reduce the image size) and it has been lightened up.
Also changed "height-shadow" theme attribute to "shadow-height".
Previously, sheets could be resized from the top, which would destroy the illusion of being attached to the top of the parent window.
With this commit, sheets are prevented from being resized by the user from the top edge.
Fixed: mouseEntered and mouseExited mixed up for custom views in menu
Previously, when clicking and dragging inside a menu, mouseEntered and mouseExited were sent to the wrong views.
This commit correctly sends mouseExited to the previous hovered view, then mouseEntered to the new hovered view.
Fixed: using keyboard navigation in empty CPCollectionView threw exception
Previously, if an empty CPCollectionView was first responder and keyboard navigation was used, a CPInvalidArgumentException (Range {-1, 1} is out of bounds) was thrown.
This commit checks for an empty collection, eliminating the exception.
Without this fix, a platform window could be resized by clicking just under the menu bar and dragging downwards, even that platform windows are supposed to be the size of the browser window by definition.
Without this fix, the presence of a modal ancestor window to a token field would lock down the autocomplete menu from mouse interaction.
This fix ensures the token field autocomplete menu accepts mouse events even during the presence of modal widows, just like other auxiliary windows such as pop up menus.
Without this fix, opening a transient popover would allow windows other than the modal window (or its children) to be interacted with.
In general, setCallback:forNextEventMatchingMask: did not work right if more than 1 callback was installed.
In the popover case the modal window would install its "any event" callback to control the event loop, but then the popover would add a second callback for mouse down to detect clicks outside of it. This would disrupt the modal event handling and allow mouse clicks (and presumably other events) to escape the modal event catcher.
This fix makes multiple nextEvent callbacks work as one would expect, with later ones taking priority over older ones and reinserted callbacks remaining at a stable priority.
This allows the modal window callback to be at the "bottom" of the handler stack and the popover event handler to be "on top" of that, and to remain that way. In theory more layers of event handling could be layered on top such as 2 simultaneous popovers.
This fix also fixes the dequeue argument in forNextEvent callbacks.
The popover test now added exposes a bug:
1. Click Modal Window to open a modal window.
2. Click Popover to open a popover.
3. Click on the first window.
It should not be possible to interact with this window since a modal window is open, but a bug in the modal event handling when transient popovers are open will allow the event to incorrectly pass through to the window.
AttachedSheet2 is turning into a complete window/modality/sheets/alerts/popover interaction test.
Without this fix, a transient popover would close if the autocomplete menu of a token field contained in it was clicked.
This happened because the popover considered the click to be on a different window. Although not tested, it's likely the same thing would happen if the menu of a combo pop or a pop up button was clicked.
This fix makes it so that any child window of a popover can be clicked without the popover closing.
This change makes it possible to @deref any expression, such as a conditional expression, as long as the expression doesn't have side effects.
The reason not to allow dereferencing of expressions with side effects is that we might need to evaluate the expression twice in certain uses of deref, which is not obvious when you look at the deref operator in plain code.
- Cover the formatting of @"" to nil.
- Check error messages by reference.
- Check output values.
- use @ref() instead of hand coded AT_REF.
Refs #1829.
Without this fix, `- CPNumberFormatter getObjectValue:forString:errorDescription:` would return `YES` for string @"", like if it was properly converted to an object value, but then actually leave object value unchanged.
With this fix object value is set to nil like in Cocoa.
Refs #1829.
New: use dedicated class instead of CPString for selection markers
Previously, selection markers were strings, which allowed the possibility that user data could be mistaken for a selection marker.
With this commit, selection markers are global instances of a private class, removing the possibility of clashes with user data.
Fixed: views from other windows cannot be next/previous key views
Previously, a view from another window could be set as the next/previous key view, which would result in unexpected behavior, including an infinite loop.
This commit ensures the proposed next/previous view belongs to the same window as the view to which it will be chained.
Previously, minimum and maximum were not supported in CPNumberFormatter.
This commit adds support for these in IB and via code. Updated unit tests included.
Fixes#1829
Without this fix, CPDecimal will return NaN for numbers that have leading zeros, e.g., 0123.
This fix changes the matching regex to allow leading zeros to pass. This is then converted to a proper number later on, e.g., "0123" => 123. This is in line with Cocoa behaviour.
This commit also includes updated unit tests.
objj_backtrace_print is only available in debug frameworks. Previously we were not ensuring that objj_backtrace_print was available, which caused an error when running with release frameworks.
Now objj_backtrace_print is checked for availability.
Added documentation for tableView:dataViewForTableColum:row: delegate method. Document other cases, when we expect the data view to have a CPViewNotSizable mask.
ViewBasedCib example: added variable row heights.
Without this fix, CPView's resizeWithOldSuperviewSize: would simply resize subviews at the rate of 1 pixel for every 2 pixel change.
This change introduces a new method of resizing views to a ratio of the size of the superview.
Patch contributed by @davidkhess
Before this fix: environment variables defined `.travis.yml` were still
dependant of internal configuration of the travis worker virtual
machine. See also previous pull request #1728.
With this fix, `.travis.yml` is even more generic and portable.
Previously, if CPApplication -targetForAction:to:from: returned nil, no check was done to see if that failed because there was no action or no valid target. If the item had no item, no target, and to action binding, it would be left enabled, which was incorrect.
This commit fixes the validation to disable the item if targetForAction:to:from: returns nil and the item has an action or target.
This is a second attempt to fix the problems addressed by 69bbe627a9.
Because table cell classes needed to use the same initialization code as the NS_initWithCoder methods, it was necessary to refactor the initialization code into separate NS_initWithCell methods and call them separately from NS_initWithCoder.
Originally the initializer had coerced the parameter to a string, even though it is clearly documented that it should be a string. A later commit removed that coercion.
This commit reinstates the type coercion and thus saves the world from those who would pass non-string types in a parameter that says "withString".
The Javascript .apply() method takes two arguments: A context argument and an array of values. In CPWebScriptObject callWebScriptMethod:withArguments, only an array of arguments was passed to the .apply() method.
This fix inserts `this` as the first argument, and the array of arguments as the second. This has been converted to a pull request from the original issue (#1184)
The original issue also suggested a fix for handling namespaces, which are not incorporated in this patch.
Previously CPAttributedString did not provide an -init method,
with the result that an attributed string created with [[CPAttributedString alloc] init]
would not be correctly initialized.
Also, some methods that took indexes into the string were not checking for an invalid index and were not raising exceptions when the documentation said they should.
Finally, some methods that were supposed to return empty dictionaries were returning nil.
This commit fixes all of these problems. Empty strings should no longer cause errors.
Closes#533
Not clear why this worked before but not now. This fix makes sense regardless.
Without this fix, clicks would not be sent through to the token field menu because its window wasn't the key window.
This fix makes the autocomplete menu declare it does not need to and does not want to become the key window.
Fixes#1807.
Without this fix, grey lines were showing in the background of token field autocomplete menus in recent versions.
This fix removes the grid lines which were never intended to be there to begin with.
Fixes#1806.
Previously, table view cells (non-view-based) were not being decoded with all of the available cell attributes in IB, and many attributes were hard-coded. In addition, there was a bunch of initialization code that has been made redundant by theming.
With this commit, decoding of all control classes has been refactored so that table view cells can use the same code, thus ensuring they decode all available cell attributes. In addition, the redundant code has been eliminated.
It is now possible to apply most table cell attributes such as font, color, alignment, sendsActionOnEndEditing, and so on, directly in IB.
Closes#1803
BREAKING CHANGE:
Changes have been made to table cell decoding in nib2cib. All xibs should be re-converted with `find . -name '*.xib' -exec nib2cib {} \;`.
Without this fix, CGPathMoveToPoint would not actually move to the point if there had been at least one previous CGPath command (a current path existed) but it was not a move to point command.
This fix ensures move to point is effective in all cases.
Fixes#1801.
This fix adds a little more intelligence to the bootstrap.sh script.
If a user is using ZSH or BASH it will recognize those and append to the appropriate config file, creating .profile if necessary.
If the user's shell cannot be read from the $SHELL variable, it will attempt to find the first suitable configuration file in the user's home directory. If none can be found, it will create .profile.
This was based off the Wikipedia article on Unix shells and the configuration files they will read. The only shells that are not supported by this patch should be csh and tcsh.
The implementation of CGPathEqualToPath did not compare all of the subpath types now supported by CGPath. In addition, it referred to subpath properties that no longer exist.
The implementation now compares all properties of all supported subpath types.
Closes#1246
Previously CPCollectionView supported a delegate method collectionViewDidChangeSelection:, which is not defined in Cocoa. Since CPCollectionView is KVO-compliant for selectionIndexes, this delegate method is no longer necessary and is a needless divergence from Cocoa.
The method is now marked as deprecated, and when used will log a deprecation warning.
Closes#715
Previously setContent: was called first, before the rest of the object was initialized. This potentially could have unexpected results because setContent has many side effects.
This commit waits until the object is completely initialized before calling setContent.
In Cocoa, viewWillMoveToSuperview: and viewDidMoveToSuperview: are called when views are instantiated from code or from a nib.
Previously in Cappuccino, when a view was instantiated from a cib, the _subviews of the view were directly set from the decoded subview array. This short-circuited all of the normal notifications a view receives when added to a superview, including viewWillMoveToSuperview: and viewDidMoveToSuperview:. As a result, some views that override these methods, such as CPSearchField, were not set up correctly when instantiated from a cib.
With this commit, subviews instantiated from a cib are manually added to their superview, thus ensuring they go through the same cycle as views instantiated from code, and ensuring that viewWillMoveToSuperview: and viewDidMoveToSuperview: are called.
Closes#1699
Previously, the content view of almost all window types
could overlap the frame of the window.
This is a follow on to b28429f which insets the content view
from the frame for all relevant window types.
Refs #1798
Previously, _CPStandardWindowView did not inset the content view rect
from the window frame, thus its contents could overlap the window frame.
Now the content view is correctly inset so that content view content
is clipped to the window frame, except for the corners. Because the
corners are round and the window frame is a background color, they
can still be overlapped. this will be fixed in a later commit.
Closes#1798
Array literals look like `@[a, b, c]`.
This syntax is supported for completeness and source compatibility, but are not terribly useful since standard JavaScript arrays are toll-free bridged to CPArray.
Array literals could be handy if you're replacing CPArray with your own implementation.
Previously, windows with no minimum size could be resized
by the user such that the content view was completely hidden, and then
resized such that the window was effectively turned inside out.
This obviously needed to be fixed.
Now, when the user resizes a window, each _CPWindowView subclass
is asked for the minimum resize size, with the size being additively
set by each subclass. This minimum size takes into account things
like the title bar, divider line, and close button, and always leaves
at least 2px height for the content view.
The user-requested size is then pinned to the minimum resize size.
In the process of fixing this bug, I discovered in the Cocoa docs
that the CPWindow -setFrame: methods should ignore minSize and maxSize.
Other changes:
- Take the divider height into account in contentRectForFrameRect for titled windows.
- Use _CG macros and fix formatting.
Closes#1753
Contrary to what the previous documentation said, removeObjectIdenticalTo: should remove all instances of the argument from the receiver, not just the first one.
This is an improved fix to the corrupted tab view state bug partially addressed by 0f347be.
Without either fix, NSTabView is converted to a CPTabView where tab view item views are subviews of the NSTabView, while in Cappuccino the selected view item is supposed to be a subview of a CPBox.
Without this fix, CPTabView decoding called addSubview: in initWithCoder: which can cause the superview of a view to become set, only to later be overwritten when the superview ivar is decoded.
This fix removes the tab item view as a subview from the tab view before encoding it in nib2cib, and then makes sure not to call addSubview: while still in the initWithCoder stage, but rather later in awakeFromCib.
If a view to be added is already a subview, but doesn't yet "know it" due to being only partially decoded, a corrupted view hierarchy with duplicate views would be set up, which is very hard to debug.
This bug used to be concealed by the fact that we called [superview removeObject:X] when replacing a subview, which caused all duplicates to be removed.
Without this fix, the call to setDocumentView in CPClipView's decoding would try to add the document view as a subview. But when decoding, the clip view is already in the decoded list of subviews. Normally adding a subview twice would just move it to the end of the list of subviews, but this only works for fully decoded subviews where superview is not nil. Since we're in the middle of decoding there's no guarantee superview is set yet.
This fix avoids calling addSubview:.
In the modified test described in #829 performance improves by at least 30% and that's with a significant addSubview: component diluting the test.
This change should improve outline and table view performance.
This fix corrects the call in CPToolbar _setWindow to `removeObserver:name:object`. Before, this was calling the non-existent `removeObserver:object`, which would raise an exception.
This was fixed by @schipmolder.
- NEW: Retrieve the underlying Image element from a CPImage with -image.
- NEW: You can now render any arbitrary drawing to a pattern context and use that as a fill or stroke pattern. See CGContextCreatePatternContext, CGContextSetFillPattern and CGContextSetStrokePattern. Works in all canvas-enabled browsers, including IE 9+.
- NEW: An example of using a custom rendered pattern is in Tests/Manual/PatternFillTest.
- NEW: Test if a CPImage is a single image (vs. three/nine part) with -isSingleImage.
- FIXED: With canvas, we have to track ourselves whether the context has a path or not.
- FIXED: All shapes except rects may not be added to a path with no context. If you attempt to do so, an error is logged.
- FIXED: CGPath was not setting the start and current point correctly in some cases.
- FIXED: CGContextAddPath was not moving to the path's start point at the beginning.
- FIXED: Removed superfluous CGContextClosePath commands, fixed some drawing sequences.
- FIXED: Misc. formatting.
Sorry, these changes are canvas only (including IE 9+)! I am not going to spend the time to port these fixes to VML (IE 8).
- You may now bind radio groups to selectedValue, selectedTag, selectedIndex, enabled, and hidden. enabled and hidden are multiple value bindings.
- Added some API to CPRadioGroup that parallels NSMatrix: selectRadioAtIndex:, selectRadioWithTag:
- Added setEnabled/setHidden, they operate on all of the radios in the group.
- Sample app to demonstrate all new features.
Without this fix, a regular custom NSBox such as a simple box with a 1px border would become several pixels smaller in Cappuccino.
This fix makes sure we only try to adjust the size for primary and secondary type NSBoxes.
Without this fix, a tab view set up in Xcode where a tab but the first was selected could be instantiated in a corrupted state in Cappuccino, where tab selection would have no effect.
This fix ensures that even if the decoded view of the selected tab view item has a superview different from how we do things in Cappuccino, we make sure to set it right.
- With canvas, we have to track ourselves whether the context has a path or not.
- All shapes except rects may not be added to a path with no context.
- CGPath was not setting the start and current point correctly in some cases.
- CGContextAddPath was not moving to the path's start point at the beginning.
Sorry, I am not going to spend the time to port these fixes to VML!
Without this fix, a CPAlert window couldn't actually become the key window which meant it couldn't receive keyboard events such as using the escape key to close the alert or hitting enter to activate the default button.
Furthermore the window could not be moved by clicking its background and dragging.
This fix makes CPAlert windows into CPPanels instead, which can become key windows and is more appropriate since panels are meant for auxiliary windows.
This is a partial fix that should resolve#1745.
-- It removes an extra "-bar" from the progress bar images in the HUD theme.
-- It removes an older "-bar-regular" from file names to match the file names in the regular theme
-- It corrects a misspelled "indeterminate" in the progress indicator
- Get the scroller orientation from the xib.
- Fix the nonsensical calculation of _isVertical when given bounds.
- Since orientation comes from the xib, had to re-nib2cib everything to be safe.
-scrollerOverlay is called in -reflectScrolledClipRect and creating this new
object is VERY expensive, according to my profiling.
This should speed up scrolling across the board!
- CPView and subclasses support multiple-value hidden bindings.
- CPControl and subclasses support multiple-value enabled bindings.
- CPWindow and CPBox support multiple-value title with pattern bindings.
- CPButton supports multiple argument + target bindings.
- CPImageView and CPTextField support multiple-value editable bindings.
- CPMenuItem supports multiple-value enabled bindings.
- Fixed bugs in CPObjectController with simple collection operators.
- CPColorWell uses black as the placeholder color.
- Runtime object attributes from a cib are applied as they are read, not deferred.
- NSNumberFormatter now reads the number style from the xib.
- Normalized some parameter names.
- Formatting.
- Test app for all binding types.
Create the drop indicator in mouseEntered if needed. This is the case where the collection view is a drop destination but not a drag source.
In _updateDragAndDropStateWithDraggingInfo... , handle the special case where the drop destination is 0 (includes item count == 0).
Test app: add an empty collection view acting as a drop destination for the first collectionView.
- Make height 20px, same as in Xcode, so we can actually lay them out correctly without guessing. And they were just too tall at 25px anyway.
- In case the height does change in the future, calculate the correct height from the theme in NSProgressIndicator.
- Updated the theme descriptor to include the default height.
- Updated Aristo 2 PSD progress indicator artwork.
- Moved init of CPProgressIndicatorSpinningStyleColors out of layoutSubviews, it is file global and only needs to get done once.
Also clean up test a little by initialising the array controller in the xib and connecting the buttons there. This is more natural.
Also make each object created in the table numbered to make it easier to see which one was just added.
This fix addresses the following problems:
- fixes a bug with add: where it would try to add the sender object, not the represented object
- changes insert: to insert a new represented object after the currently selected object. If no object is currently selected it behaves like add: and adds a new object to the end.
This patch includes tests.
There's no `selectedTag` getter for CPPopUpButton (although there's such a binding but that's separate). Also don't call private API that no longer exists - use proper public API instead.
Setting the item prototype without first adding the collection view to a view would cause a crash, as would reloadContent.
This fix makes sure the collection view doesn't try to tile itself unless it has a frame.
If the parent or grandparent window of a token field which was showing an autocomplete menu was closed, the menu would stick around even that the token field was gone.
This fix makes the token field autocomplete menu a proper child window which should also ensure it moves as needed if the parent window moves.
Without this fix, a CPBoxOldStyle box with a CPBezelBorder would draw with the correct border colour and fill, but with an additional inner shadow, unlike Cocoa.
This fix makes it so that only primary and secondary boxes draw with the special "well"-like shadow style.
Without this fix, setting a corner radius of 0 on a CPBoxCustom resulted in there being a corner radius anyhow.
This fix makes sure to preserve the corner radius, border width and margins for non-primary style boxes.
In Chrome, a primary CPBox (showing the new recessed rounded corner look) had distorted and jagged edges in Chrome due to the clipping mask.
This fix places the clipping mask on integer coordinates which seems to resolve the problem.
In Aristo 2, the token field tokens would always show the close button, and make a big space for it.
This change makes it so that when the delete button, and the new disclosure button, are hidden, which is most of the time, the token doesn't look like it has a big empty hole on the right side.
- Tighten up hover state code.
- CPCoding support.
- Make sure to initialise button type to `CPTokenFieldDisclosureButtonType` rather than null.
- Don't try to lay out disclosure/delete button in the token token.
Disclosure arrow behaves like in Cocoa: it shows on hover and if clicked displays the token specific menu.
This change also removes the delete button by default. The delete button can be enabled with `[tokenField setButtonType: CPTokenFieldDeleteButtonType]`.
Without this fix, the autocomplete field would try to select the CPNotFound index from time to time which has always been wrong but is now specifically an error in Cappuccino.
The fix is to select nothing instead.
Without this fix, the pulldown button style displays using the same right-hand side widget as the popup button. This fix creates a new set of bezel images specifically for the popup button.
The token delete button becomes visible on mouse over of the token in the standard theme. It could be made to disappear even that the mouse was still over the token by mousing over the button itself and then out.
This fix delegates all hover management to the token alone.
- Removed stray console.error
- Made border of tooltips darker, they were almost invisible, even over a white background
- Added CGAlignStroke and CGAlignCoordinate, they are of general use
This change implements support for the CPTokenField delegate method `tokenField:menuForRepresentedObject:` (and the associated `tokenField:hasMenuForRepresentedObject:`).
* Note that this does not match the Cocoa documentation, but a whole bunch of event handling code relies on this behavior, so I have to keep it the way it was.
* Fixed dequeue flag in CPApplication -runModalSession:
* Added a window title in CPAlertTest
Without this fix the new compiler would treat IBOutlet as the type and expect the next token to be the ivar name.
This fix make the new compiler treat IBOutlet the same as @outlet.
When running on certain platforms the `ulimit` command is not available, which would crash the build script.
This change simply refrains from using the `ulimit` command if it's not available.
Without this fix all regular key events such as `CPKeyDown` would have an undefined location.
Since the location can't be retrieved from the browser on key events, this fix now makes it so that the last seen mouse position is included with each key event.
Before this fix:
* Build PATH has recently changed on Travis CI and hard-coded values in `.travis.yml' were no more valid (the `builds` folder has been renamed as `build`)
* Any contributor that forked the main project repository had to modify the definition of environment variables in order to integrate with http://travis-ci.org
With this fix cappuccino is correctly configured to be built on Travis CI and makes `.travis.yml` more generic (forkable).
Setting ulimit in narwhal.conf isn't necessary, not all targets need a larger limit. In addition, if the hard limit is 512, it can't be set above that once the terminal session starts.
- Added CPWindow -_parentWindowDidOrderInChild to notify a window when it is directly ordered in by a parent.
- Fixed observing of target view frame changes to work in all possible cases.
- Fixed typo in code introduced in last commit in CPApplication -sendEvent.
- Fixed case where a dangling transition function was left when closing a popover.
- Fixed case where closing a popover, moving the target view, then reopening the same popover would not reposition the popover when resizing the target view's window.
- Optimized repositioning of popover on target view change.
- Implemented transitional popovers via trapped mouse down events.
- Implemented dequeue: argument of CPApplication -setTarget:selector:forNextEventMatchingMask:untilDate:inMode:dequeue: and -setCallback:forNextEventMatchingMask:untilDate:inMode:dequeue:.
- Fixed some child window behavior when ordering/closing.
- Eliminated no longer necessary window close notification in CPPopover.
- Fixed window title in theme so it doesn't dim when window is main but not key.
mutableArrayValueForKey: … removeObjectsInArray should remove all instances of the found objects. Without this fix, only the first instance was removed.
- Popovers are implemented as child windows.
- Renamed _CPAttachedWindow/_CPAttachedWindowView to _CPPopoverWindow/_CPPopoverWindowView, since that is its only use.
- The default for CPView -acceptsFirstMouse is now NO, per Cocoa. Subclasses override this as necessary.
- _CPWindowView -hitTest returns self it the mouse is within a resize region, which may be outside the window's frame.
- Fixed an off by one bug in CPDomWindowLayer -insertWindow:atIndex:, where inserting a visible window behind a window it is already behind would cause it to move up one from its intended position.
- Updated the ChildWindows and CPPopover test apps.
This change makes CPDictionary descriptions exactly like Objective-C literals. Later when Objective-J adds support for such literals too, it'll be possible to just copy and paste a CPDictionary description into code.
Also updated CPArray to look similar for consistency.
Without this fix, KVCArray could try to remove an out of bounds index at CPNotFound from its target array.
This fix makes it so that only found objects are removed.
- Made CPDescriptionOfObject smarter about detecting CG objects
- CPObject, CPArray, CPDictionary do a better (as in perfect) job of indenting nested objects
- CPDictionary is displayed using Cocoa-style key = value; notation
Errors are now displayed in a data view showing:
- the file
- the error message
- A button that allows you to open the errored file in the default system editor
The `cs == nil` to `cs === nil` change introduced by 64a0fb2 was incorrect. Since `CPCharacterSet` is used widely through Cappuccino this broke many unit tests and much functionality.
The fix is to again consider both `nil` and `undefined` as cause for initing a new character set.
Without this fix, CPNumberFormatter didn't format negative values well. E.g. with thousands separators on it could generate, "-,999.00".
This fix generally improves handling of negative numbers by CPNumberFormatter.
With the new compiler, messages such as `[a in:b]` could not be sent, the "in:" parameter throwing an unexpected token error. This was true both if "in" was the first argument or any of the subsequent parameters (such as in `[a something:X in:Y]`).
Before this fix, certain errors during compilation such as not being able to load a file, would cause the compilation to fail without stating what went wrong.
This fix prints such errors and avoids crashing later when the result of the compilation is undefined.
- When a window is neither key nor main the title bar is clearly different
- capp_lint ThemeDescriptors.j
- Renamed some layers in Aristo2 PSD to be more descriptive
- New artwork files for window close buttons
- Removed stray Aristo2 image
Without this check, an index set could be initialised with NaN which would later lead to an infinite loop when enumerating the index set.
This error makes the error early and explicit.
- If a window is fixed width or fixed height, then no resize cursor is shown and no resize area is active on the edges that cannot be resized.
- Optimized CG calls in CPWindow.
- Added fixed height window to test app.
This commit provides resizing of windows from all sides. The legacy style, which displays a resize thumb in the lower right, can be restored with [CPWindow setGlobalResizeStyle:CPWindowResizeStyleLegacy].
- Factored out blur handler from CPTokenField into CPTextField.
- CPTextField will no longer lose focus unnecessarily.
- Changed switch style in CPWindow -sendEvent.
- Added menu bar to KeyViewLoop test app.
- It turns out Cocoa DOES recalculate the loop when calling recalculateKeyViewLoop. It does not wait until selectNext/PreviousKeyView is called.
- Cocoa calculates the key view loop for nibs that do not have one, but does not call recalculateKeyViewLoop.
- If there is no first responder and no initial first responder when a window first becomes key, use the first valid key view, not the first geometric key view.
- Made allViews function a method, not sure why it was a function.
- Unit tests now pass.
Before this fix, XCodeCapp always ran under /bin/bash, causing problems when trying to source the files from other shells, like /bin/zsh
This fix reads the shell information directly from the user's environment and XCC to run using that shell.
- Removed unnecessary code at beginning of CPTextField -becomeFirstResponder that might have been a hack to get around a bug I fixed.
- Fixed race condition in setTimeout closure in CPTextField -becomeFirstResponder.
- CPWindow -setInitialFirstResponder now works reliably and follows Cocoa behavior in that if -makeFirstResponder is called with something other than the window before the window is first shown, it will override the initial first responder.
- Like Cocoa, until the first responder is set during window load, the first responder is the window by default, not the content view.
- Optimized search for any view that has a previous/next key view set.
- Sheets can become key windows again.
- CPWindow -recalculateKeyViewLoop now just marks the loop as dirty, per Cocoa docs.
- CPWindow -autorecalculatesKeyViewLoop now behaves per Cocoa, it only has an effect when views are added or removed.
- If the first responder does not have a valid previous/next key view, it does not resign to nil, per Cocoa behavior.
- Code optimization and cleanup.
- Test app (KeyViewLoopTest) that demonstrates various scenarios.
Cappuccino will now continue to support Aristo1
By default Aristo2 will be used, unless you add in your Info.plist
<key>CPDefaultTheme</key>
<string>Aristo</string>
In that case Aristo2 will be use.
If you put
<key>CPDefaultTheme</key>
<string>MyCustomTheme</string>
It will use your custom theme as usual from your app's Resource folder
This has the added benefit of ignoring unknown/unsupported prefixes, providing the property itself is camel-cased.
For example:
"LinearBackground" -> "linear-background"
"Linear-Background" -> "linear-background"
"linear-background" -> "linear-background"
"fooLinearBackground" -> "linear-background"
and
"WebkitLinearBackground" -> "-webkit-linear-background"
"MozLinearBackground" -> "-moz-linear-background"
etc..
For example, the following CSS property names all resolve to the same correct JS equivalent for the current browser:
* CPBrowserStyleProperty('TransformOrigin') -> 'WebkitTransformOrigin'
* CPBrowserStyleProperty('transform-origin') -> 'WebkitTransformOrigin'
- Track mouse exit from window resize rect instead of unconditionally setting arrow cursor.
- Don't show the resize cursors for full platform windows.
- Show the resize down cursor if the top of a window is being resized and it has reached the bottom of the menubar.
- Make windows in CPToolbarTest resizable to test resizing behavior with toolbars.
- Added a full platform window and menubar to the test app.
- The main run loop resets the cursor if it falls outside of all windows during a mouse move.
- Added cursors supported by CSS that are not in NSCursor.
- Code cleanup in CPCursor.
- Cursors now indicate available resize directions if there is a window min/max size.
- Reduced slop to 3, what it is in Cocoa.
- Added a second non-min/max window to the demo.
Without this fix, `CPPanel` followed the same rules as `CPWindow` with regards to becoming the key window. But `CPPanel` is supposed to always be willing to become the key window, as stated in its documentation.
This change makes `CPPanel` always return `YES` for `canBecomeKeyWindow`.
Before this fix, CPAlert subclassed CPView, causing an API mismatch with Cocoa and exposing unexpected methods in the public API which made no sense for a CPAlert such as autoresizing masks. It also lead to CPAlert mistakingly shadowing _window from its CPView superclass.
This fix makes CPAlert inherit from CPObject as it should. The theming, which needs to be applied to a view, was moved to an internal view.
Without this fix, CPAlert windows would not finish closing before sending the `alertDidEnd:returnCode:` delegate message. This would lead to a crash if the delegate method tried to open a new sheet.
This fix makes sure to fully order out the alert sheet before sending the delegate message.
"detect-missing-imports" takes a long, long time to run. Since it only emits warnings it can't affect the result of the continuous integration builds.
Until it can be made a little faster it's probably best to run it by hand with `jake check-missing-imports`.
Without this change, CPApplicationTest's manipulation of window.location.hash was ignored by CPApplication in recent versions, since CPApp in a unit test normally does not have access to the window object.
This change makes it so that CPApp does try to access the window object, even in a unit test scenario (non PLATFORM(DOM)).
Before this fix, CPApp tried to add nil to its internal list of windows, even that nil should not be added to a CPArray. Second, that nil was actually returned by [CPApp windows] rather than an expected list of only CPWindow instances.
This fix makes [CPApp windows] return only actual windows, replaces the internal _windows[0] placeholder with CPNull, and ensures this CPNull is never actually exposed outside of the class.
Without this change, bootstrap.sh outputs a great deal of often superfluous information, like precisely which files are unzipped.
Also, curl and wget display progress information by default.
With this change, by default unzipping is quiet, unless --verbose is specified. curl and wget still display progress information, but --quiet can be specified to prevent this.
Without this fix, token fields generally assumed that if there was anything in the shared editor buffer, it belonged to the current field, even if it wasn't the field that was being edited.
With this fix only the token field actually being edited considers the editor contents.
Before this fix, popoverDidShow: would not be sent to popover delegates when the app ran in Firefox due to Firefox now recognising the -webkit-transition property, but not the webkitTransitionEnd event.
Also, before this fix popover animation was only implemented for Webkit based browsers.
This fix resolves that issue and in addition adds full support for animated popovers in all browsers that support CSS transitions. The fix also improves Cappuccino's general support for CSS3 animations.
This prevented text fields and token fields from accepting input when placed in a pop over.
This fix allows the underlying popover window to become the key window.
Without this fix, the scroll handle in the token field autocomplete menu could not be moved using the mouse. Only scroll wheel or keyboard based scrolling was possible.
This fix enables regular mouse interaction with the scrollbar.
Before this change the close button showed at all times, even for token fields which weren't even editable.
This change brings us closer to Cocoa in that there's no token delete button most of the time, and just like in Cocoa there are more controls available on hover. (Although in Cocoa there's a disclosure arrow and we have a close button instead.)
Without this change, clicking a token in a first responder token field would cause it to momentarily lose first responder status and then immediately gain it back.
This fix makes it so that tokens never try to become the first responder, resulting in a more stable token field as tokens are selected and deselected.
Without this change some 3rd party JS libraries such as PDF.js, which rely on the nonstandard setTimeout(f), will not operate correctly.
According to MDN the delay argument is required. However browsers seem to treat setTimeout without delay argument the same as a delay of 0. This fix makes it so that Cappuccino apps do the same.
There is not need to use that much arguments because the result is stored in ivars.
CPCollectionViewItem -copy: set properties correctly even if the collectionView overrides them.
Do not cache removed items in -reloadContent when the item prototype changes.
maxItemSize:
Interpreted as horiz/vertic expandable when zero size and the proto view has a CPViewWidth|HeightSizable. (per cocoa but not documented).
maxItemSize height support.
Fixed: incorrect position when num items < numberOfColumns.
Rewrote layout engine. Splitted the computation of size, columns count, rows count from the actual layout where we set the frame.
This will help when/if we support insertion/mutation of -content or view animations.
We can also be more lazy when the computation results stay the same.
Added shared -init. Whitespace cleaning.
CPCollectionviewCibTest:
Add ability to set the 2 kinds of prototypes, one loading its view from a cib, the other with a view outlet in the same cib.
Loaded protoype: the bound textfield can commit its value to the model.
UI for setting maxNumberOfColumns|Rows min|maxItemSize
Due to a typo in ce9534c, even standard titled windows returned NO for `canBecomeKeyWindow`. This triggered errors in `CPApplicationTest`.
The fix returns the correct YES for `canBecomeKeyWindow` of titled windows.
With this commit CPCollectionView can load subviews from an external
cib. Outlets, actions and bindings between the view and the
prototype are restaured when the view is loaded.
After recent changes clicking on an autocomplete item in the pop up did nothing. Now this works and in conjunction with the new non-key-window-change update it's likely to work better than before.
Without this fix Cappuccino always returned YES for `canBecomeKeyWindow` for every window. But in Cocoa, only "standard" windows with a title bar and/or resizing return YES.
With this fix Cappuccino better matches Cocoa's behaviour.
If the token field was in a window not placed in the upper left corner of the screen (like regular platform windows are), the autocomplete menu would show up in the wrong spot.
This means there'll be 1 px of spacing both above and below tokens in multi-row configurations. It also places the token much better in the field both in single and multi-row configurations, at least for the Aristo theme.
Without this fix there'd be a white line visible on top of the token field when scrolling between multiple lines of tokens in a short field.
In addition the scroll view was 1 px short which made it possible to scroll tokens up and down by 1 px even when there was only a single row of tokens which should fit in the field.
This fix expands the clip rect upwards by 1 px to bring it all the way up to the first row of bezel pixels, ensuring sure a single row of tokens fits fully in a 1 row token field.
E.g. in a two window application, if a token field was the first responder and another window was made active and then the original window made active again, the token field would no longer be the first responder.
Without this fix, the tab key press would be propagated to the browser, which would blur the focused input element, even when that input element was the one the new first responder intended to use.
Resizing for all tab view types can now be tested by clicking "Cycle Tab View Type" up to 6 times and resizing the window for each mode. (Some modes are not implemented.)
IB sets `autoresizesSubviews` to NO for tab views by default. In Cocoa this doesn't make any difference because `NSTabView` ignores this setting.
This fix makes it so that Cappuccino's tab view also works with `autoresizesSubviews` off.
The window views driving non standard windows (such as borderless or bridge windows) did not take the toolbar into account when calculating the content rect.
Apart from being wrong, this could lead to the content view overlapping the toolbar if the content view was set after the toolbar was prepared.
Perhaps the thought was that the base _CPWindowView shouldn't make assumptions about the placement and sizing of a toolbar and leave that up to subclasses, but the _CPWindowView tile method already did.
Before, the popoverDidShow: delegate methods was called during the CSS animation (if any)
Problem is that is we are processing big amount of data, the animation may hang for a bit in the middle, giving an impression of jerkiness.
Now, popoverDidShow: is called at the end of the CSS animation if popover is animated.
If the formatter rejected a change and prevented the first responder status from being surrendered, the text field would still stop listening to key window changes and break first responder behaviour when switching between windows.
A line breaking, vertically centred or bottom aligned `_CPImageAndTextView` would not properly recalculate text position when the height of the text changed due to the width of the field changing.
This could cause, for example, table header text to be misaligned if the table was initialised with a zero height.
Without this fix, a matrix with a target-action pair would crash when the cib was loaded due to the matrix having been replaced with a `CPView` (which cannot take a target-action pair).
Now if the matrix is for radio buttons, the action connection for the matrix is made to the `CPRadioGroup` instead and works as expected when radio selections are made.
Without this feature windows could easily end up being placed outside of the screen/browser when loaded in Cappuccino due to e.g. the window position being 100 px from the bottom of a 1418 px tall screen but the browser only being 800 px.
This commit adds support for all "initial position" window masks in Interface Builder except "centre". Position can be specified in current screen space coordinates and then adjusted proportionally to the browser size, or adjusted while preserving a left or right margin and a top or bottom margin.
`CPTableView`'s `selectionIndexes` binding shouldn't be set automatically unless the content binding was also set automatically (through a table column binding).
Assume that manual content binding implies a desire to manually bind selection indexes (and sort descriptors) as well.
With this option an array controller driven table view can be sorted by clicking the table column headers without manually setting up sort descriptor prototypes.
If `CPCreatesSortDescriptorBindingOption` is enabled on a table column value binding, sort descriptors will be created automatically. The default value for this option is YES.
- Attributes set in IB are now obeyed.
- Added accessor methods to programmatically style _CPTableColumnHeaderView.
Since we don't have an NSCell and _CPImageAndTextView is supposed to be private, it's better to have the styling methods in _CPTableColumnHeaderView.
you can set CAPP_NOSUDO=1 in your environment to disable any sudo action.
This means that if a command failed as current user, jake
will not try to run it with sudo,and will simply return the error code
Both controls and cells can have tags in IB, but Cappuccino has no cells. If the control has a tag, that takes precedence. Tag reading was moved to NSControl from NSView, since NSView does not have a tag in Cocoa.
will not scroll without the NSScrollView knowing about it
Issue #1675 and maybe a little of issue #1301
This fix will not work without pull
request #1678 - Fixed scrollRectToVisible in CPView
Without this optimisation nil or 0x0 sized image slices in three part and nine part images result in useless empty divs.
Eliminating these divs should improve performance and reduce memory usage.
Although exact memory usage is hard to measure, in a large app the number of divs on screen went from 1012 to 984, a reduction of about 2.7%.
Avoid sending notifications about toolbar items' enabled state needlessly.
Tiling is expensive, but even if the toolbar item ignores redundant `setEnabled:` calls, the cost of notifications alone can really add up when there are many toolbar items.
`CPEvent timestamp` is now in seconds so the double click delay has to be adjusted. Also slightly increased double click interval to account for lower precision.
This would make the popover open with the wrong coordinates if the positioning view was in a window different than the main window, such as a utility window which can't become the main window.
E.g. a dictionary is expressed as @{key1: value1}, an array like @[value2, value2], a string like @"<value>".
Also don't put all child objects into quotation marks - that makes them all look like strings.
Apparently Rhino doesn't have `<input element>.setAttribute` and an error would be thrown in `CPCompatibility.j`: `Cannot find function setAttribute in object [object Object].`
We were compensating for a bug in Safari, older versions of Chrome and Firefox in our input placement. But new Chrome and neither IE 8 nor IE 9 needed it which resulted in the text being off in those browsers.
Locking down the line height and vertical alignment explicitly seems to work in Safari 6, Chrome 21, Firefox 14 and IE 8 (although Chrome and IE exhibit some unrelated horizontal jumping.)
This was caused by making text display generally 1 px taller to allow for fractional pixels at the bottom to show. This should however not be allowed to affect positioning (even if there is a fractional pixel below vertically centred text the text should not move up 0.5 px.)
Only the first responder of the key window should receive text input. This is important in multi-window apps or windows with sheets.
In the multi window case it should be possible to switch between two windows with first responder text fields and keep right on typing. First responder status should not be lost, but new typing should always go into the first responder text field of the key window.
For sheets, a text field which was active when the sheet opened should not receive input nor look active while the sheet is open.
If a window contained a text field which was the initial first responder, and the window was displaying a sheet, clicking on the window would cause the text field to alternate between being the first responder and not.
This was caused by the window `orderFront:` setting its first responder to the initial responder if there wasn't any responder yet. This would cause the text field to focus, even that the window wasn't the key window, and then on the next click the text field would blur which would make the text field automatically resign its first responder status.
The solution is to leave the first responder status alone if a window which isn't the key window is `orderFront:`ed. This appears to be more sensible UX in general. You don't expect the first responder of non-key windows to change when you click it's background.
If a custom font was used with the CSS @font-face syntax, Cappuccino would calculate the wrong sizes for strings throughout. This could result in e.g. cut off sizeToFit labels.
The sizing tests were being performed inside of an iframe and that iframe would not have the same CSS @font-face declarations. For maximum fidelity all string size tests are now performed in the main document, just off screen.
Events are not so reliable, there are many flaws and special cases (like trashing a folder etc)
In order to simplify the process is the following
- We receive an event
- We tidy up the shadow files
- We check if the given path exists
- If it's a folder, we continue
- If it's a file and it exists : we run handleFileModification
- If it's a file and it doesn't exist: we run handleFileRemoval
Sometimes clicking a token would cause the token field to scroll to some other spot. This appears to have been caused by the inactive input element gaining focus.
We no longer need the input element for keyboard navigation so we can hide it entirely when it's not used to input a new token. Also, avoid focusing the input element needlessly.
If an autocomplete choice was made followed by immediately focusing on another control, the token field would reclaim first responder status a moment later.
A non bezeled text field is now bezeled when edited and unbezeled when editing ends - including when you hit enter (but text has to be modified, see comment in code)
The extra comma causes IE to think it was passed an array with 4
values and causes CPColorWithImages to return a NinePartImage even
though it only has three parts.
Before that, IB views were reusable only if the delegate method was not implemented.
If you create a view in code, you still need to make it reusable by calling -setIdentifier:
Woow ... now viewBasedCib loads the views x3 times faster.
When you add an observer to an object first time, its class is
implicitly changed to a KVO_originalClassName (subclass of original
class). This subclass adds willChange…/didChange… methods for
observable properties.
If you send willChange… before you add an observer, it does nothing.
But if you send didChange… just after, the app will crash, because
new KVO_originalClassName nerver receives willChange…
The idea is to maintain counter of all received willChange… messages
(per key) and decrease it in didChange…
When KVO_originalClassName is created and didChange… is received
(without opening willChange… to new class), exception is not thrown
immediately, but the counter is checked first.
If it's greater than 0, then didChange… just closes
an unboserved willChange… Otherwise exception is thrown, as expected.
Text size and theming enhancements
- _CPImageAndTextView adds extra pixels to allow for fractional pixels in -sizeToFit.
- NSButton now allows buttons that are not fixed height, if the theme allows it.
- Documented the mapping between bezel style constants and their IB style names.
- _CPImageAndTextView adds extra pixels to allow for fractional pixels in -sizeToFit.
- NSButton now allows buttons that are not fixed height, if the theme allows it.
- Documented the mapping between bezel style constants and their IB style names.
This would lead to a `[CPPlatformWindow _willPropagateCurrentDOMEvent] unrecognized selector sent to instance` error if an event was sent in a non DOM environment.
- Fixed case where no format strings are specified in xib (which ends up being an empty string).
- Fixed case where malformed strings file would not match regex.
- General cleanup.
Overhaul of system font handling
- System fonts now track the currently configured system font at runtime.
- You can specify CPFontCurrentSystemSize as a system font size to track the currently configured system font size at runtime.
- Theme fonts and nib2cib now determine face and size at runtime, there is no need to recompile if the system font is changed in Info.plist.
- All hard-coded references to 12.0 as a system font size have been replaced by the current system font size.
- Documented the font substitution algorithm and how to use custom fonts in the CPFont class docs.
- See Tests/Manual/FontEnhancementTest for a demo.
Text clip fix
Currently when controls and text fields size to fit text, they use a to determine the correct height/width. Some browsers do not account for fractional pixels that arise from antialiasing. Cappuccino views must clip their contents, so the fractional pixels get clipped, and depending on the font this can be quite noticeable.
This patch adds 1 pixel when necessary to ensure no clipping occurs.
- System fonts now track the currently configured system font at runtime.
- You can specify CPFontCurrentSystemSize as a system font size to track the currently configured system font size at runtime.
- Theme fonts and nib2cib now determine face and size at runtime, there is no need to recompile if the system font is changed in Info.plist.
- All hard-coded references to 12.0 as a system font size have been replaced by the current system font size.
Support value transformers for CPContent and CPContentValues bindings.
Made _CPPopUpContentValuesBinder a subclass of CPPopUpContentBinder and reuse setContent:.
Fixed a bug in selectedIndex binding set and reverse set when insertsNull options is on.
_CPPopUPSelectionBinder : simplified the code for getting the unique selection binder in use.
ojunit and manual tests for contentvalues + value transformer.
CPComboBox and CPTextField enhancements
- Full Cocoa-compliant implementation of CPComboBox and CPComboBoxDelegate.
- Better focus ring for CPTextField and all subclasses.
- CPTextField and its subclasses draw disabled contents differently, per Cocoa.
Container view is responsible for laying out its views.
Container view is a content view of the window.
If there is no container view, view controller's view is a content view of the window.
- Full Cocoa-compliant implementation of CPComboBox and CPComboBoxDelegate.
- Better focus ring for CPTextField and all subclasses.
- CPTextField and its subclasses draw disabled contents differently, per Cocoa.
Make the default grid color in IB show correctly in CPTableView.
- Changed the default grid color from #dce0e2 to #ccc.
- Fixed NSColor to recognize the named color "gridColor" and set it to #ccc.
- NSTableView always saves the grid color from IB.
When you change the top level object, observer MUST receive value
for the key path of the previous top level object as old and
value for the key path of the new top level object as new.
Currently #888888 is being used as the default, mainly because NSColor.j doesn't recognize the "gridColor" named color, and the cached color in the .xib is #888888 for some reason. Actually the default grid color is #CCCCCC.
Nested split views now correctly restore their divider positions by restoring "top down". E.g. a nesting split view is restored before the nested split view so that the nested split view has the correct frame size for the position it is restoring.
This commit also makes it easier to override autosave restores in a subclass or a controller's `viewDidLoad`. Now if `setPosition:ofDividerAtIndex:` is called after the split view's `initWithCoder:` has finished, but before the final autosave restore has occurred, the autosave restore is cancelled. This preserves whatever the modification the programmer intended to make instead of overwriting it with the autosaved position.
This commit also fixes a bug where an autosaved position would not be restored, or would be restored at an unexpected time, if `setFrameSize:` was not sent to the split view shortly after its `initWithCoder:`. (Such a `setFrameSize:` message is often sent due to a window `setFullPlatformWindow:` call in `AppController.j`'s `awakeFromCib`, but it's not necessarily sent.)
This covers the case where the whole split view is resized and the divider moves as a consequence. For example if the split view was made much smaller the fixed position of the divider previously autosaved might not even be inside of the split view frame anymore, and on restore from autosave that side of the split view might unexpectedly become collapsed.
The new behaviour also more closely matches Cocoa's, although there is still a slight difference in implementation.
- nib2cib will now log the path and size of custom image resources in super verbose mode.
- Alternate images for buttons were not being set by nib2cib, that has been fixed.
- Fixed the name of the imagesize binary in the Xcode project.
If the delegate returns nil we're just copying nil, and if it does return an array we only use it for a single read-only `arrayByAddingObjectsFromArray:`.
This fix makes it possible for a toolbar defined in a cib to have a delegate set without defining `toolbarDefaultItemIdentifiers:` like in Cocoa. You might be implementing other delegate methods.
When higlightsBy == CPNoCellMask and showsStateBy == CPChangeBackgroundCellMask
higlighted status must be preserved when a user switch state from Off to On.
If one or more token is selected and the left arrow key is pressed, collapse the selection to the left of the first selected token.
Similarly, on right arrow key collapse to the right of the last token.
Note that shift-left and shift-right have no effect if the first or last token respectively in the field are already selected.
Cmd-A (Mac) or Ctrl-A (other) finishes any ongoing editing and selects all tokens in a token field.
This commit also adds support for the `selectText:` and `selectAll:` methods.
We now leave the default propagation policy of Cappuccino in place, which normally is NO for regular characters but YES for things which could be shortcuts.
If the editor is placed while there is a token selection it can cause the token view to become scrollable to a white area at the bottom even that there is no editing cursor. And even if the editor is not at the bottom, the empty space made for the editor makes no sense when no editing cursor is visible.
If a half typed token is in the field, Cocoa tries to convert it to a representedObject before responding to objectValue. Now CPTokenField does the same.
Token field now triggers autocompletion from textDidChange: which better meets the expectations of Cappuccino programmers, and is closer to how Cocoa does it.
Eliminated custom non-Cappuccino key press handler.
The new menu is much more efficient for the user:
* Less tall items makes it much easier to select among many completions.
* Exactly wide enough for widest item.
* Tall enough to show all items (instead of just 3 as before).
* Uses a maximum width and height defined by the screen edges and some Cocoa-like constants.
The menu should appear above other normal windows and controls like a regular CPMenu. To do this, put the pop up in its own window with the `CPPopUpMenuWindowLevel` window level.
This improves support for represented objects and is used both when an autocompletion suggestion is used or when the name of a token is typed in by hand.
The behaviour of taking an array of strings from `completionsForSubstring` and then asking for a represented object once a choice has been accepted matches the behaviour of Cocoa.
Auto completions are strings, not represented objects. Revert "Fix CPTokenField autocompletion to deal with represented objects"
This reverts commit 8b9a0edeba.
The first method (updateUserDefaultsWithLastEventId) just
sets the default while the other (synchronizeUserDefaultsWithDisk)
forces a synchronization with the disk.
This is a small fix that makes XcodeCapp update the latestEventId value
in the defaults database at the end of each FSEvents callback.
Prior to this fix, the synchronization was performed only when the user
actively decided to stop listening to a project.
* Support for tall custom widgets. You can now finally have that CPTableView you always wanted in your toolbar.
* "Labels only" mode now makes sense, resulting in a narrow text only toolbar instead of a very tall toolbar with centred labels.
* Use any size toolbar icons such as 48x48 and the toolbar adapts automatically (unlike Cocoa).
The height of all items is determined by the tallest item, like in Cocoa. But unlike Cocoa our toolbar is presently fixed height so in order to make this look well for small or large items, the whole row of items is now vertically centred, making the allocation of margin above and below automatic.
Furthermore if a toolbar item has a small view or a small icon, it is now centred within the available space above the label. For example, for a tiny icon in one item next to a normal size icon the tiny icon would be centre aligned with the larger icon.
It looks like issue #1560 will cause a different sorting function to be used for selectors vs descriptors so performance of both should be monitored for regressions.
Make sure the the xCodeSupport/Sources folder is always
present. The objj tool will fail if it's missing.
The Sources folder could be missing if it's not checked
into a repos for example.
Syntax such as function(/*CPString*/anArgument) {} should not generate a warning, and in general inline /**/ comments should be ignored for linting purposes.
Made LintChecker()'s "view" and lint_text's "filename" arguments optional. Filename will default to <stdin> like Python does when you pipe to it. Also return found errors from lint_text as a convenience.
Grouped the 3 ways to set objectValue on a dataview (content binding, column value binding or regular delegate method) under one main method: _setObjectValueFortableColumn:row:ForView:
Also fixed a bug where the dragging rows didn't have their bindings applied. Columns and row dragging are now using the previous method for setting data view objectValue.
Updated TableBindings example: table with content binding vs. columns with value binding.
---
Dont you hate when your logs of arrays and dictionaries display [Object object] when you have JavaScript objects in them. Now with this patch all JavaScript objects will be display with all attributes in a nice good looking way. Log CGRect, CGPoint and other JavaScript objects withour any hassle anymore. For example: console.log([myRect]);
Conflicts:
Foundation/CPObject.j
Tests/Foundation/CPMutableArrayTest.j
---
Dont you hate when your logs of arrays and dictionaries display [Object object] when you have JavaScript objects in them. Now with this patch all JavaScript objects will be display with all attributes in a nice good looking way. Log CGRect, CGPoint and other JavaScript objects withour any hassle anymore. For example: console.log([myRect]);
Don't you hate when your logs of arrays and dictionaries display [Object object] when you have JavaScript objects in them.
Now with this patch all JavaScript objects will be display with all attributes in a nice good looking way. Log CGRect, CGPoint
and other JavaScript objects withour any hassle anymore. For example: console.log([myRect]);
Selections in the tableview can now have round corners. The default is a regular
rect, but if the selecction radius is changed. Nice good looking round rect
selections will appear.
This will make relative paths fail. For example if OBJJ_INCLUDE_PATHS has the path "../../MyLibrary".
MyLibrary will not be found when the path is joined with mainBundlePath without an ending '/'.
This is working when running in Browser.
And in general divisions where the quotient had fewer digits than the dividend when disregarding exponent would fail.
We want to stop dividing when we've gone over all the digits in the dividend and the remainder is 0.
The errors would cause exceptions such as "Not allowed to load local resource: file:///<some local path from the machine that ran nib2cib>" when loading cib driven app through a web server.
A CPOutlineView without a delegate `outlineView:menuForTableColumn:item` method, would not display its context menu as specified by `setMenu:` or in IB.
This rhino-narwhal patch can be applied to add JAVA_OPTS support when running Cappuccino utilities through rhino. This generally improves configurability for special cases, and specifically enables a work around for the "failed to map segment from shared object" error on some platforms.
Set the dragView frame before dataviews are added to it. Otherwise, dataviews will be expanded depending on their autoresize mask and appear cropped on the right.
Added : -makeViewWithIdentifier:owner:
-rowForView: columnForView: less performant tahn cocoa because we cycle through all visible rows. But these methods are generally used once when editing manually.
Added identifier property to CPView.
retro and forward (in IB) compatible with the existing API. You can use IB table view based without the new API. Existing table views / Nib won't break.
CPTableView view based example. Featuring: IB made data views, different views in the same column, subviews binding in IB
When the view is found automatically in the cib, instantiate with _delegate as the owner (cocoa behavior).
- Changed from deprecated image scaling constants to new ones.
- Added support for CPImageScaleProportionallyUpOrDown (thanks BlairDuncan).
- Cleaned up image layout code.
- Added support for User Defined Runtime Attributes in IB.
- Added support for userland NS classes in IB.
- Added support for file-based nib2cib configuration.
- Converted nib2cib to Objective-J from straight Javascript.
- Updated man page for new features.
- Test app.
- Fixed bizarre bug where local variable path was being set as attribute of object in CPURL -pathComponents.
- Added CPURL -URLByDeletingLastPathComponent.
- Linted CFURL.js.
- CFURL constructor should copy when passed a CFURL.
- Added CFURL.createCopyDeletingLastPathComponent.
- Added test for URLByDeletingLastPathComponent.
- Moved common CPArray/CPSet collection KVC operators to _CPCollectionKVCOperators.
- Implemented KVC operator dispatch using Objective-J.
- Fixed infinite loop with empty collection in @min, @max and @sum operators.
- Correctly return valueForUndefinedKey when necessary.
- valueForUndefinedKey reason uses raw description for consistency, class' overridden description may not helpful at all.
- Don't create a forwarder for @ operators with property paths.
- Fixed CPSet -valueForKeyPath to correctly deal with nil/undefined/empty values.
- Added tests for collection KVC operators.
- enumerator -nextObject should always compare against nil for clarity, correctness, and consistency.
- showRelativeToRect:ofView:preferredEdge: must take provide a view, and if the rect is empty it defaults to the view's bounds.
- If If showRelativeToRect is called while a popover is closing, it is ignored.
- If showRelativeToRect is called when the popover is already open, it should just reposition the existing popover.
- close should NOT consult the delegate's popoverShouldClose. Only performClose should do that.
- If a window is closed, all of its popovers immediately order out.
- Updated demo app to cover more cases.
- Note that child popovers are currently not supported, as this requires support for child windows in CPWindow.
- Finally fixed#1457.
- Fixed case where mouse is dragged and then released over a submenu, which should release the menu.
- Fixed case where clicking without dragging, then clicking again on a menu title would not close the menu.
If there's a single item selected and it's not expanded, pressing the left arrow key on the keyboard now selects the item's parent unless the delegate vetoes.
- Removed animationStyle for now.
- Made sure that popoverDidClose is not called until animation is finished.
- Don't create a new attached window unless: there isn't one; the popover behavior has changed; the current attached window is still visible.
- Don't call popoverWillShow if there is not content view controller.
- Removed unused _shown instance variable.
- Removed redundant !_attachedWindow checks followed by _attachedWindow message that returns a BOOL.
- Documentation tweaks.
- Fixed up some demo issues.
- Lion-style popovers now zoom open just like their Lion counterparts (except for opacity, the browser couldn't handle that at the same time).
- Fixed a number of subtle drawing issues caused by not aligning path points correctly.
- The anchor arrow deals correctly with corner cases where the middle of the reference rect is not within the corners of the popover.
- Code cleanup: factored redundant code, removed unused code.
- Added corner cases to the demo app, changed demo app to reuse a single popover.
- Added animationStyle (Lion/iOS) to CPPopover. iOS style does not zoom open, it just appears.
- Tweaked the Lion animation a bit to be a bit less rushed and match Mac OS X timing better.
- Fixed some typos.
This could either be due to autohiding or they usage of overlay scrollbars. Not showing it matches Cocoa and makes sense because it takes up space for no reason without the scrollbar.
The triangle centring was such that for a regular size disclosure control the right pointing triangle (collapsed state) was not drawn on integral coordinates.
ShouldShowOutlineView is not a great delegate method name for a control called an outline view. The original Cocoa name refers to an 'OutlineCell' which we don't have so that's not appropriate either.
Switched to using a somewhat vertically resizable three part image. The centre image part is being compressed which normally would look poor but you can't tell since it's a gradient.
Although the intention might have been to do 'one final calculation run to provide a sensible default', the result is unintuitive and does not match Cocoa's behaviour.
If the key view loop has been manually set up and the user then calls `setAutorecalculatesKeyViewLoop:NO` they don't expect their work to be overwritten.
---
This regards CPSplitViews new delegate method
- (BOOL)splitView: (CPSplitView)aSplitView shouldAdjustSizeOfSubview: (CPView)aSubView
which was introduced in b27a3cdbe5 by Alexander Ljungberg.
This regards CPSplitView's new delegate method
- (BOOL)splitView: (CPSplitView)aSplitView shouldAdjustSizeOfSubview: (CPView)aSubView
which was introduced in b27a3cdbe5 by Alexander Ljungberg.
---
- selectedTabViewItem should return nil when there are no items (i.e. _selectedIndex == CPNotFound).
Added an empty Tab View to CPTabViewNib test app.
Needs a jake sudo-install for changes to take effect.
Updated CPTabViewNib to show a setView: on the current item.
If the view changes on a non selected item, the tabView content will be
updated when this item is selected.
SmartFoldersDemo: disable the size-to-fit like feature in the controller because 1/ CPPredicateEditor should have its own sizeToFit and 2/ It was shadowing other resizing bugs.
* date is (today, yesterday)
* date is less than [text field]
- CPPredicateEditorCibTest example: templates merging in the 'mixed' criterion.
- ojtest AppKit/CPPredicateEditorTest testing the merging feature.
- Cleaned and rearranged code related to the right control action method.
- re-enable tooltips.
In CPRuleEditorCibTest, the delegate has been updated to follow this rule, it also fixes a bug where the predicate was not correctly handling custom control values.
Do not truncate bold labels.
So this is a temp fix as @aparajita has a better way to handle this. As the patch is very straightforward, it will be easy to revert it when a more correct fix will come up. But while waiting, I think it's a good idea to have correct displaying of the bold labels.
This fix also enables `setSelectionHighlightStyle: CPTableViewSelectionHighlightStyleSourceList` in Internet Explorer. Even that the gradient effect can't be rendered and the fallback regular rendering will be used, the proper field value might still be useful for subclasses or saving out to CIBs etc.
Syntax such as [[CPColor colorWithCSSString:@"rgba(0,0,0,0.5)"] cssString] lead to a CSS string incompatible with browsers without the CSS rgba feature.
If one observer of a notification centre notification removed another observer, that other observer would still be notified even that it was no longer observing.
Now the code agrees with its comments.
-handle attempt to close sheet while it is animating in
-fixes Safari 5.1 animate out bug by using a timer to
initiate the in/out process
-allow window shadow to become 0 width/height for smoother animation
If the autocomplete state had not been set yet, tokens could not be finished using the token separator, return or tab. Instead the user typed text remained in its 'non token' state.
After testing in Cocoa, even unexpected keys such as tab can be used when a text field isn't selected. Some keys such as the arrow keys and escape act like key equivalents even with a text field active.
---
Currently CPTextField only updates its internal string value from the actual HTML input element on a keyUp event. If the CPTextField loses focus before the keyUp event is received, the internal value does not get updated and any text entered since the last keyUp event is lost.
This is easily reproduced when editing a text field by quickly entering text and hitting the enter/return key to insert a new line (before the keyUp event has fired), or when holding down a key (to repeatedly enter the same character), and then tabbing out of the text field -- the value gets reset to whatever the value was before the key was held down.
This fix changes CPTextField#insertNewLine and CPTextField#resignFirstResponder to reload the string value from the input element if it has changed (in the same way as if a keyUp event had been triggered)
The tusk ecosystem is not maintained anymore within the Cappuccino community. The new installation system is not very elegant but it makes it very easy to release new versions of Cappuccino, while taking the first step on the road towards node.js.
- Add preferences window
- Allow user to choose to use the 10.6 mode on 10.7
- Allow user to choose if he wants to resume listening at startup
- Allow use to choose if XCC should react to inode modification (file level API)
- Support for clean mirrored path when origin files has some _ in their names
- Other various fixes
This patch contains the following:
- FSEvent is now at files level instead of folders level. This allows to drop the registry in plist and improve the performances
- When you remove a J file, the mirrored H file is removed too
- .xcodecapp-ignore has been added to the FSEvent stream. ignored token are updated in live
- By default XCC ignores all .git, .svn, .hg .xCodeSupport, Frameworks and Build folders
- When you quit XCC without stop listening of the project, at next launch it will resume on the same project
- Help has been added
- Help is displayed at first user launch
- The Error & Warning table doesn't pop ups automatically, instead it displays Growl (you can then open the panel for more informations)
- The code has been refactored into several classes, one for UI thing, another one for XCC worker
- Several bugs has been fixed
- The about window now displays the version number
- The main window has been removed
- Several other small enhancements.
This is XCC 2.0.
The calculation of the number of pixels outside of the clip view assumed fixed, space consuming scrollbars even when overlay scrollbars were used, leading CPScrollView to always think a few more pixels could be revealed.
If there is no vertical scroller, or there is no horizontal scroller, due to autohiding or scroll view settings, a single visible overlay scroller should extend all the way to the edge of its dimension.
The Lion style scrollbars are inconvenient for users without a touch scrolling device, and we do not have a way to detect whether the user has such a device yet.
Switch to cookie based user defaults if HTML 5 local storage exists but is disabled, as is the case when Safari is in 'private browsing' mode.
Don't crash if the local storage quota is reached. Instead log a warning and silently fail to store the new user default.
If more than one subview is collapsed, remember the pre-collapse divider position for each subview separately.
Autosave pre-collapse positions to restore positions when uncollapsing after a reload.
If the right/lower subview is collapsible and the divider is at its uncollapsed maximum position, continue showing the two way resize cursor to indicate additional movement is actually possible. Also show the shrink cursor when the mouse is over the divider to the left/above of a collapsed subview.
Merely clicking a divider without moving it should not result in a splitViewDidResizeSubviews: message. Fixed errors with setPosition:ofDividerAtIndex:'s no change detection.
Only call splitViewDidResizeSubviews: after the subviews have actually finished being resized. Make sure to call splitViewDidResizeSubviews: when views are collapsed, even if _adjustSubviewsWithCalculatedSize performed no work.
When a collapsed split view subview was restored, e.g. by double clicking the divider, and the collapsed subview was on the right hand side of the divider, the pre-collapse split view position was not restored.
When a divider is dragged to the right past half of the minimum width of a collapsible subview, or the divider is double clicked and the delegate returns YES for shouldCollapseSubview:forDoubleClickOnDividerAtIndex:, the subview is collapsed. This is the mirror operation of collapsing towards the left.
Fixed: if the split view delegate sent non numeric responses when asked to constrain divider positions, subviews would be resized with corrupt size information, which in turn could lead to infinite loops and redraw errors. These bad results are now handled more gracefully.
---
Tusk seems to need OJTest to be alone in order to calculate and/or
download dependencies (which are the Objective-J and Cappuccino
pre-built packages).
---
running tusk install browserjs jake shrinksafe AND extra packages (OJtest, objj and capp)
in a single instruction prevent the installation of the lasts.
This patch splits the procedure in two separate instructions and should fix all problems.
running tusk install browserjs jake shrinksafe AND extra packages (OJtest, objj and capp)
in a single instruction prevent the installation of the lasts.
This patch splits the procedure in two separate instructions
---
Cocoa compliance.
* updated manual test: log a text only when user clicks a radio button butnot when using the CPButton that uses setState:
* ignore .xCodeSupport folders.
---
Store the conversion date before starting to actually convert. This should avoid to loop indefinitly in case of long conversion time.
Also convert tabs to space, and remove trailing spaces.
Store the conversion date before starting to actually convert. This should avoids to loop indefinitly in case of long conversion time.
Also convert tabs to space, and remove trailing spaces.
---
- Enhanced -description method of CPImage and CPColor to show the exact content, especially for pattern colors.
- Added CPColorWithImages() convenience function, analogous to PatternColor() function used in theme descriptors.
- Added @cond/@endcond in CPColor to ignore local variables.
- Sample app included. See the console for samples of -description output for all color/image types.
- 'jake clean' now cleans instead of clobbers.
- Added task 'clean-sprites', cleans sprited images.
- Added task 'clobber-theme', clobbers all traces of Aristo.
- Added missing semicolons, followed JavaScript lint suggestions for using dot notation instead of [""].
In the PlatformWindows test, change the window title both when theWindow is a non platform window and when it is a platform window to verify that the browser window title changes only when it should.
---
if setTitle: is called on a CPWindow with the primaryPlatformWindow as platformWindow,
the main pages title is changed. This patch prevent changing the primaryPlatformWindows title.
This is not optimal, but this is better than changing the title at each window setTitle: call
---
Enabled tokenField:shouldAddObjects:atIndex delegate method and used on _autocompleteWithDOMEvent method to have some control over tokens. Not sure why wasnt enabled, it needed very few code. Maybe no one needed it til now.
This is my first collaboration ever, so Im not sure if is of any use.
- include the window's content view at the start of the key view loop
- set the initial first responder to the window's content view (and keep it that way unless it is explicitly changed)
- CPWindow should return NO from acceptsFirstResponder (Cocoa compliance).
- change the type of setInitialFirstResponder, this is compliant with Cocoa. The new key view loop code will break if initial first responder is anything but a view
- change the implementation of selectNextKeyView:, selectPreviousKeyView:, selectKeyViewFollowingView: and selectKeyViewPrecedingView: to follow to Cocoa's documentation
Tusk seems to need OJTest to be alone in order to calculate and/or
download dependencies (which are the Objective-J and Cappuccino
pre-built packages).
Everything in the README has been fixed except 3 and 4. 4 is outside of the issue scope and 3 works as intended (nearly). The array controller does not have `setAvoidsEmptySelection:NO` so an empty selection should not be possible unless there are no items.
When multiple calls are nested, only the first willChange… and the last didChange… will notify observers, avoiding duplicate notifications for the same change.
if setTitle: is called on a CPWindow with the primaryPlatformWindow as platformWindow,
the main page's title is changed. This patch prevent changing the primaryPlatformWindow's title.
This is not optimal, but this is better than changing the title at each window setTitle: call
- Enhanced -description method of CPImage and CPColor to show the exact content, especially for pattern colors.
- Added CPColorWithImages() convenience function, analogous to PatternColor() function used in theme descriptors.
- Added @cond/@endcond in CPColor to ignore local variables.
- Sample app included.
- New/improved jake tasks: debug, release, all, clean, clobber, test, test-release, help.
- help task shows comprehensive, formatted documentation for the available tasks.
- BlendKit is automatically symlinked into the Frameworks directory if necessary as part of the build process, so that the showcase will always work.
- User can easily change the build directory by changing one line.
- Fixed incorrect placeholders in main.j.
if (theWindow == modalSession._window || [theWindow worksWhenModal])
/*
The special case for popovers here is not clear. In Cocoa the popover window does not respond YES to worksWhenModal, yet it works when there is a modal window. Maybe it starts its own modal session, but interaction with the original modal window seems to continue working as well. Regardless of correctness, this solution beats popovers not working at all from sheets.
*/
if (theWindow == modalSession._window ||
[theWindow worksWhenModal] ||
[theWindow attachedSheet] == modalSession._window || // -dw- allow modal parent of sheet to be repositioned
@return The button's state. A button can have two or three states.
If it has two, this value is either \c CPOffState (the normal or unpressed state)
or \c CPOnState (the alternate or pressed state).
If it has three, this value can be \c CPOnState (the feature is in effect everywhere), \c CPOffState (the feature is in effect nowhere), or \c CPMixedState (the feature is in effect somewhere).
Returns the button's next state.
@return The button's state. A button can have two or three states.
If it has two, this value is either \c CPOffState (the normal or unpressed state)
or \c CPOnState (the alternate or pressed state).
If it has three, this value can be \c CPOnState (the feature is in effect everywhere), \c CPOffState (the feature is in effect nowhere), or \c CPMixedState (the feature is in effect somewhere).
*/
- (CPInteger)nextState
{
@@ -263,7 +304,7 @@ CPButtonImageOffset = 3.0;
}
/*!
Sets the button's next state to \c aState.
Sets the button's state to the next available state.
@param aState Possible states are any of the CPButton globals:
\c CPOffState, \c CPOnState, \c CPMixedState
*/
@@ -294,9 +335,9 @@ CPButtonImageOffset = 3.0;
}
/*!
Sets the title displayed by the button when in its normal state.
@param aTitle The string to set as the button's title. This title is always shown on buttons
that don’t use their alternate contents when highlighting or displaying their alternate state.
Sets the title displayed by the button when in its normal state.
@param aTitle The string to set as the button's title. This title is always shown on buttons
that don’t use their alternate contents when highlighting or displaying their alternate state.
*/
- (void)setTitle:(CPString)aTitle
{
@@ -310,9 +351,9 @@ CPButtonImageOffset = 3.0;
}
/*!
Returns the title displayed on the button when it’s in its normal state.
@returnThe title displayed on the receiver when it’s in its normal state
or the empty string if the button doesn’t display a title.
Returns the title displayed on the button when it’s in its normal state.
@returnThe title displayed on the receiver when it’s in its normal state
or the empty string if the button doesn’t display a title.
*/
- (CPString)title
{
@@ -374,11 +415,19 @@ CPButtonImageOffset = 3.0;
- (void)setShowsStateBy:(CPInteger)aMask
{
// CPPushInCellMask cannot be set for showsStateBy.
if (![aSource respondsToSelector:@selector(numberOfItemsInComboBox:)] ||
@@ -250,7 +278,9 @@ var CPComboBoxTextSubview = @"text",
CPLog.warn("Illegal %s data source (%s). Must implement numberOfItemsInComboBox: and comboBox:objectValueForItemAtIndex:", [self className], [aSource description]);
}
else
{
_dataSource = aSource;
}
}
}
@@ -295,7 +325,7 @@ var CPComboBoxTextSubview = @"text",
{
// Issue the warning first, because removeObjectAtIndex may raise
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@import "CPEvent.j"
@import "CPEvent_Constants.j"
@import "CPPlatform.j"
@@ -38,42 +38,71 @@ CPWindowsOperatingSystem = 1;
CPOtherOperatingSystem = 2;
// Features
CPCSSRGBAFeature = 1 << 5;
CPCSSRGBAFeature = 5;
CPHTMLCanvasFeature = 1 << 6;
CPHTMLContentEditableFeature = 1 << 7;
CPHTMLDragAndDropFeature = 1 << 8;
CPHTMLCanvasFeature = 6;
CPHTMLContentEditableFeature = 7;
CPHTMLDragAndDropFeature = 8;
CPJavaScriptInnerTextFeature = 1 << 9;
CPJavaScriptTextContentFeature = 1 << 10;
CPJavaScriptClipboardEventsFeature = 1 << 11;
CPJavaScriptClipboardAccessFeature = 1 << 12;
CPJavaScriptCanvasDrawFeature = 1 << 13;
CPJavaScriptCanvasTransformFeature = 1 << 14;
CPJavaScriptInnerTextFeature = 9;
CPJavaScriptTextContentFeature = 10;
// In onpaste, oncopy and oncut events, the event has an event.clipboardData from which the current pasteboard contents can be read with event.clipboardData.getData.
CPJavaScriptClipboardEventsFeature = 11;
// window.clipboardData exists and can be read and written to at any time using window.clipboardData.getData/setData.
CPJavaScriptClipboardAccessFeature = 12;
CPJavaScriptCanvasDrawFeature = 13;
CPJavaScriptCanvasTransformFeature = 14;
CPVMLFeature = 1 << 15;
CPVMLFeature = 15;
CPJavaScriptRemedialKeySupport = 1 << 16;
CPJavaScriptShadowFeature = 1 << 20;
CPJavaScriptRemedialKeySupport = 16;
CPJavaScriptShadowFeature = 20;
CPJavaScriptNegativeMouseWheelValues = 1 << 22;
CPJavaScriptMouseWheelValues_8_15 = 1 << 23;
CPJavaScriptNegativeMouseWheelValues = 22;
CPJavaScriptMouseWheelValues_8_15 = 23;
CPOpacityRequiresFilterFeature = 1 << 24;
CPOpacityRequiresFilterFeature = 24;
//Internet explorer does not allow dynamically changing the type of an input element
CPInputTypeCanBeChangedFeature = 1 << 25;
CPHTML5DragAndDropSourceYOffBy1 = 1 << 26;
//Internet explorer does not allow dynamically changing the type of an input element
CPInputTypeCanBeChangedFeature = 25;
CPHTML5DragAndDropSourceYOffBy1 = 26;
CPSOPDisabledFromFileURLs = 27;
// element.style.font can be set for an element not in the DOM.
CPInputSetFontOutsideOfDOM = 28;
// Input elements have 1 px of extra padding on the left regardless of padding setting.
CPInput1PxLeftPadding = 29;
CPInputOnInputEventFeature = 30;
CPFileAPIFeature = 31;
CPAltEnterTextAreaFeature = 32;
/*
When an absolutely positioned div (CPView) with an absolutely positioned canvas in it (CPView with drawRect:) moves things on top of the canvas (subviews) don't redraw correctly. E.g. if you have a bunch of text fields in a CPBox in a sheet which animates in, some of the text fields might not be visible because the CPBox has a canvas at the bottom and the box moved form offscreen to onscreen. This bug is probably very related: https://bugs.webkit.org/show_bug.cgi?id=67203
*/
CPCanvasParentDrawErrorsOnMovementBug = 1 << 0;
// The paste event is only sent if an input or textarea has focus.
CPJavaScriptPasteRequiresEditableTarget = 1 << 1;
// Redirecting the focus of the browser on keydown to an input for Cmd-V or Ctrl-V makes the paste fail.
// foregroundColor and backgroundColor are ignored in Cocoa as well. See http://developer.apple.com/library/mac/#documentation/Cocoa/Reference/ApplicationKit/Classes/NSCursor_Class/Reference/Reference.html
Init a cursor with the given image and hotspot. This is provided
for compliance with Cocoa. Note that foregroundColor and backgroundColor are ignored
(as they are in Cocoa). See http://developer.apple.com/library/mac/#documentation/Cocoa/Reference/ApplicationKit/Classes/NSCursor_Class/Reference/Reference.html
// FIXME Do we need to encode _datePickerTextfield and _datePickerCalendar? As subviews they'll be encoded, but when we decode we recreate them anyhow.
"Geneva","Georgia","Georgia Ref", "Geeza Pro", "Gigi","Gill Sans","Gill Sans MT","Gill Sans MT Condensed","Gill Sans MT Ext Condensed Bold","Gill Sans Ultra Bold","Gill Sans Ultra Bold Condensed",
/* "GlooGun","Gloucester MT Extra Condensed","Goudy Old Style","Goudy Stout","Haettenschweiler","Harlow Solid Italic","Harrington",*/
"Geneva","Georgia","Georgia Ref", "Geeza Pro", "Gigi","Gill Sans","Gill Sans MT","Gill Sans MT Condensed","Gill Sans MT Ext Condensed Bold","Gill Sans Ultra Bold","Gill Sans Ultra Bold Condensed",
/* "GlooGun","Gloucester MT Extra Condensed","Goudy Old Style","Goudy Stout","Haettenschweiler","Harlow Solid Italic","Harrington",*/
[CPException raise:CPInvalidArgumentException reason:@"sides (length: " + sides.length + ") and colors (length: " + colors.length + ") must have the same length."];
var index = [_source indexOfItemWithRepresentedObject:aValue],
offset = [self _getInsertNullOffset];
// If the content binding has the option CPNullPlaceholderBindingOption and the object to select is nil, select the first item (i.e., the placeholder).
// Other cases to consider:
// 1. no binding:
// 1.1 there's no item with a represented object matching the object to select.
// 1.2 the object to select is nil/CPNull
// 2. there's a binding:
// 2.1 there's a CPNullPlaceholderBindingOption:
// 2.1.1 there's no item with a represented object matching the object to select?
// 2.1.2 the object to select is nil/CPNull
// 2.2 there's no CPNullPlaceholderBindingOption:
// 2.2.1 there's no item with a represented object matching the object to select?
// 2.2.2 the object to select is nil/CPNull
// More cases? Behaviour that depends on array controller settings?
@@ -208,12 +268,12 @@ var CPTabViewDidSelectTabViewItemSelector = 1,
*/
- (void)selectNextTabViewItem:(id)aSender
{
if (_selectedIndex === CPNotFound)
if (_selectedTabViewItem === nil)
return;
var nextIndex = _selectedIndex + 1;
var nextIndex = [self indexOfTabViewItem:_selectedTabViewItem] + 1;
if (nextIndex === [_items count])
if (nextIndex === [self numberOfTabViewItems])
// does nothing. According to spec at (http://developer.apple.com/mac/library/DOCUMENTATION/Cocoa/Reference/ApplicationKit/Classes/NSTabView_Class/Reference/Reference.html#//apple_ref/occ/instm/NSTabView/selectNextTabViewItem:)
return;
@@ -226,10 +286,10 @@ var CPTabViewDidSelectTabViewItemSelector = 1,
*/
- (void)selectPreviousTabViewItem:(id)aSender
{
if (_selectedIndex === CPNotFound)
if (_selectedTabViewItem === nil)
return;
var previousIndex = _selectedIndex - 1;
var previousIndex = [self indexOfTabViewItem:_selectedTabViewItem] - 1;
if (previousIndex < 0)
return; // does nothing. See above.
@@ -250,26 +310,36 @@ var CPTabViewDidSelectTabViewItemSelector = 1,
columns = [CPIndexSet indexSetWithIndexesInRange:CPMakeRange(index, [tableView._exposedColumns lastIndex] - index + 1)];
// FIXME: Would be faster with some sort of -setNeedsDisplayInColumns: that updates a dirtyTableColumnForDisplay cache; then marked columns would relayout their data views at display time.
File diff suppressed because it is too large
Load Diff
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Aparajita Fishman
Antoine Mercadal
cacaodev
Alexander Ljungberg
Christophe Serafin
Alexandre Wilhelm
Didier Korthoudt
Didier Korthoudt
Didier Korthoudt
Martin Carlberg
Martin Carlberg
Martin Carlberg
Mathieu Monneyandcacaodev
Roland Schwingel
Malte Tancred
Marek Kresnicki
Christian Passlack
daboe01
william57m
Udo Schneider
Andrew Hankinson
kevin-xu
RoosterDragon
Sebastian Balay
ratijas
Ryan Bannon
Saikat Chakrabarti
Ashley Williams
Alos
Carl Rosell
Carl Rosell
Adam Sowinski
Bruce Doan
Björn Sperber
Blair Duncan
Glenn L. Austin
Sergio Felix
Ilya Kulakov
aparajita
Kjell Nilsson
Tim Lewis
Stefan Wallström
Randy Luecke
Gilles Cornu
Andrew Hankinson
Adam Radabaugh
Stefan EilersandAlexander Ljungberg
Aparajita Fishman
Stephen Ierodiaconou
Miles Tinsley
imageoptimiserandAdam Howard
Johan Stille
Ilya Kulakov
Eric Wong
Darrell Walisser
Alexander Ljungberg
Eduardo Gonzalez
ggsato
Vladimir Shevchenko
Martin Carlberg
Mike Holman
Jiva DeVoe
zittix
Antoine Mercadal
Richard van Zon
Luc Vauvillier
Georges Auberger
Jacek Galowicz
Antoine Mercadal
Antoine Mercadal
Ross Boucher
lynnchin
rafnocx
Andrea D'Amore
Valerii Hiora
James Dessart
lowell
Neer Friedman
dwalisser
Scott RiceandAlexander Ljungberg
Johannes FahrenkrugandAntoine Mercadal
Sjoerd van Geffen
Kevin Jamieson
Nicolas GoyandAparajita Fishman
Klaas Pieter Annema
Samir Gartner
Derek Hammer
Aparajita Fishman
cncolder
