Avoid potential underflow when evaluating x - y with y > x.
Evaluate it only when y <= x instead.
Avoid clearing indicators in background task
since another watch face is likely active.
Reviewed-by: Matheus Afonso Martins Moreira <matheus@matheusmoreira.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/440
Mechanical switches exhibit temporary voltage fluctuations
when electrical contact is made, which manifest as "bouncing"
between the logical high and low states. Sampling the switch's
state during this period of stability produces invalid results.
The switch must be debounced by ignoring the generated interrupts
until the switch's state has stabilized. This is implemented by delaying
the input events until an empirically determined time has elapsed.
As such this pull request introduces customizable high resolution timers
for debouncing button down and up events.
This software debouncing unfortunately increases the sensor watch's
input latency. This is an acceptable tradeoff due to better usability:
watch faces which require holding down buttons such as the pulsometer
should work much more reliably after this patch.
Reviewed-by: Matheus Afonso Martins Moreira <matheus@matheusmoreira.com>
Tested-on-hardware-by: Krzysztof Gałka <@kshysztof@Discord>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/437
References: https://en.wikipedia.org/wiki/Switch#Contact_bounce
Adds the ability to configure at compilation time
the movement's default birth date and location.
Tested-by: madhogs <madhogs@protonmail.com>
Tested-by: Niehztog <niehztog@gmail.com>
Tested-on-hardware-by: madhogs <madhogs@protonmail.com>
Tested-on-hardware-by: Niehztog <niehztog@gmail.com>
Reviewed-by: Matheus Afonso Martins Moreira <matheus@matheusmoreira.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/387
The movement was checking for scheduled tasks
by comparing for equality their scheduled times to
the current time. However, it is possible that the time
has moved past the scheduled time by the time the
function executes, leading to scheduled tasks not
being executed and therefore to missed deadlines.
Changing it to `<=` fixes the problem by taking that
possibility into account.
Helped-by: Wesley Ellis <tahnok@gmail.com>
Reviewed-by: Matheus Afonso Martins Moreira <matheus@matheusmoreira.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/439
The TOTP face header file documentation contained outdated instructions
for configuring the complication with TOTP credentials. This PR updates
the documentation to match what is expected in `totp_face.c`.
To allow a valid date to be set the `set_time_face` cycles through the
valid days for a given month. July was incorrectly encoded as having 30
days (it has 31!)
Fixes a division by zero bug caused by calling getCodeFromTimestamp
without having initialized the TOTP library with a secret first.
This was happening because the face calls totp_display on activation,
meaning the validity of the secret was not checked since this is
done in the generate function.
Now the validity of the key is determined solely by the size
of the current decoded key. A general display function checks it
and decides whether to display the code or just the error message.
The size of the current decoded key is initialized to zero
on watch face activation, ensuring fail safe operation.
Tested-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
Tested-on-hardware-by: madhogs <59648482+madhogs@users.noreply.github.com>
Signed-off-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/385
Forgot to call watch_display_string on the error message.
Of course the message isn't going to be displayed.
Also, increase the buffer size to 10 characters
and output a space to the last position.
This ensures the segments are cleared.
Tested-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
Tested-on-hardware-by: madhogs <59648482+madhogs@users.noreply.github.com>
Signed-off-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/385
Allocate an unlimited extent 128 byte buffer once during setup
instead of allocating and deallocating repeatedly. A static buffer
was not used because it fails to be reentrant and prevents multiple
instances of the watch face to be compiled by the user.
The advantage is the complete prevention of memory management errors,
improving the reliability of the watch. It also eliminates the overhead
of the memory allocator itself since malloc is not free.
The disadvantage is a worst case default size of 128 bytes was required,
meaning about 90 bytes will be wasted in the common case since most keys
are not that big. This can be overridden by the user via preprocessor.
The key lengths are checked on TOTP watch face initialization
and if any key is found to be too large to fit the buffer
it is turned off and the label and ERROR is displayed instead.
The base32 encoded secrets are decoded dynamically to the buffer
at the following times:
- Face enters the foreground
- User switches TOTP code
Therefore, there is still some extra runtime overhead
that can still be eliminated by code generation.
This will be addressed in future commits.
Tested-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
Tested-on-hardware-by: madhogs <59648482+madhogs@users.noreply.github.com>
Signed-off-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
GitHub-Pull-Request: https://github.com/joeycastillo/Sensor-Watch/pull/385
The TOTP face is working in the simulator but fails on the real hardware
when loaded with lots of codes, just like the LFS version.
This is likely caused by the recent refactoring of the TOTP face
which introduced a declarative credential interface for ease of use.
That's accomplished by decoding the secrets at runtime which increases
the RAM requirements. Users are likely hitting memory limits.
In order to mitigate this, the algorithm is changed from decoding
all of the secrets only once during initialization to on the fly
decoding of the secret for the current TOTP credential only.
This converts this face's dynamic memory usage from O(N) to O(1)
at the cost of memory management when switching faces and credentials
which could impact power consumption. Issue is confirmed fixed by
author of issue who has tested it on real hardware. Fixes#384.
Due to variable key sizes, the memory cannot be statically allocated.
Perhaps there's a maximum key size that can serve as worst case?
Also took this opportunity to restructure the code a bit.
Also added code to check for memory allocation failure.
Reported-by: madhogs <59648482+madhogs@users.noreply.github.com>
Fixed-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
Tested-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
Tested-on-hardware-by: madhogs <59648482+madhogs@users.noreply.github.com>
Signed-off-by: Matheus Afonso Martins Moreira <matheus.a.m.moreira@gmail.com>
GitHub-Issue: https://github.com/joeycastillo/Sensor-Watch/issues/384
