- Using more logical scheme for widget layout. bezel-inset is inset from frame to the bezel image. border-inset is the inset from the frame to the visible border (if any), which usually is the width of a focus ring. content-inset is the inset from the border rect to the content frame, including the width of the border. This scheme makes it much easier to visualize what is happening, and also makes it possible to ensure a consistent margin between the content and the surrounding bezel border. It also allowed me to remove the funky +1 pixel hack in [CPTextField becomeFirstResponder].
- CPTextField (square and rounded) and CPTokenField had their theme insets reworked to fit the new scheme.
- Added missing disabled bezel for rounded textfield.
- All textfield focus rings look much nicer now (more like Cocoa). Previously the actual border of a textfield would shrink when focused, now it stays the same.
- In CPTokenField, wrapped all references to _inputElement in #if PLATFORM(DOM).
- Placeholder state was not being set correctly because a recent change to [CPTextField _updatePlaceholderState] uses the cached _stringValue instead of [self stringValue]. CPTokenField has its own override now that checks the token count.
This patch allow to send 'play' message before
the sound is actually loaded (by remembering the play request
and playing the sound when loaded if needed)
this allow to do:
```objj
var sound = [[CPSound alloc] initWithContentsOfURL:aSoundURL byReference:NO];
[sound play];
```
- [CPArrayController setSelectionIndexes] should NOT obey avoidsEmptySelection. This was preventing an empty selection in a CPTableView. In addition, when calling _selectionDidChange it should suppress notifications, because the reverseSetValueFor:@"selectionIndexes" in [CPArrayController __setSelectionIndexes] does the notifications.
- Added CPContentBinding and CPContentValuesBinding constants.
- [CPObjectController valueForKeyPath:] should return CPNotApplicableMarker for undefined keys.
- Spelling fix and consolidated var blocks in CPKeyValueObserving.j
---
I discovered the following behavior that is not in conformance with Cocoa:
- Backspace, tab, escape, and space are considered key equivalents.
- Function keys (F1 - FNN) are not considered key equivalents.
- CPTextField executes `performKeyEquivalent` twice if a field is first responder.
Build this Cocoa project and debug it: http://d.pr/Xc3h
Take a look at the debug log while typing some keys. You will see that backspace, tab, escape and space do not trigger `performKeyEquivalent`, and that `performKeyEquivalent` is only executed once for the first responder.
Then run the performKeyEquivalentsTest app in this commit and look at the console log while trying keys in the text fields. If you turn the old code on, you will see how `performKeyEquivalent` is executed twice for the first responder.
For reference on key handling code, I consulted this:
http://developer.apple.com/library/mac/#documentation/Cocoa/Conceptual/EventOverview/EventArchitecture/EventArchitecture.html%23//apple_ref/doc/uid/10000060i-CH3-SW11
---
Adds -scrollViewWillScroll: and -scrollviewDidScroll: delegate methods to CPScrollView.
These methods are not in AppKit but exist in UIKit. Useful for lazy loading data in a view, for example a CPTableview.
Tested in the CPScrollView example.
Conflicts:
AppKit/CPScrollView.j
This is a regression from 73b4c33f05.
Because we were only tracking wether the left mouse button went up after a contextual menu (which is opened with the right mouse)
would only close after the second left mouse up.
Aparajita Fishman
Alexander Ljungberg
Antoine Mercadal
Randy Luecke
cacaodev
alexmipego
Brian Donovan
Klaas Pieter Annema
Mike Fellows