As discussed, we are changing some of the platform advocates for (a)
geographical, and (b) platform familiarity reasons. This will be
reflected in the automatic assignment of reviewers on PRs.
We are pretty much all affected in this shuffle!
@keymanapp-test-bot skip
(cherry picked from commit 9bff83fa00)
Rename ProcessActionsTestParse to NonUpdatable parse. It does
more then just test. Add handling for the invalidate context action
to this parse, so that if we emit a keystroke we invalidate the context
Remove the saving of context and keyboard options
on the non updateable call to process hook.
The windows Text Input Processor calls twice one call
is not updateable. With the core integeration we run
the processor only once, preserving the actions and
context to be applied on the updatable call.
Fixes#7030.
Prevents a stack overflow / loop when the web target is removed from a
keyboard and it has already been tested on web, and the user presses the
Compile button.
- Create and sign source package
- Upload to mentors.debian.net
- Create (and optionally push) commit with updated changelog file
based on stable branch
- Create (and optionally push) commit with updated changelog file
based on master branch
This change also slightly modifies `echo_heading` and outputs a
mark. This will show to the left of the line as well as in the
scroll bar in a terminal in VSCode [[1]] and will help to find the
headings if you have a lot of output in between.
[1]:https://code.visualstudio.com/updates/v1_69#_setmark-sequence-support
(cherry picked from commits ae61410cc9,
2a55ffb024, 9232c20569,
420cc0814d, 9f7b85834a,
and ae0378a57f)
When rebuilding a model, Keyman Developer would query for the
registered model, but a missing simplifyId call meant that it never
matched the model name (due to `.` vs `_`).
Picked up while reviewing code for conversion to C++. The compiler
could, in some circumstances emit garbage code for readonly groups
because we weren't testing all scenarios correctly. In practice, this
would have been rare as readonly groups don't emit characters, but it
should be fixed!
Fixes#6981.
Longpress menus were being triggered on any detected movement on keys on
Android. This would cause rapid typing to periodically fail as the user
would make a single-pixel movement while typing, which would then block
subsequent key events.
This issue was introduced in #6637, which resolved a somewhat related
problem with sticky longpress menus.
The fix is in three parts:
1. Remove the touch movement detection which caused the primary problem
2. Address some logic issues around visibility of `subkeysWindow`
3. Introduce a new 'scroll' gesture handler which detects a negative-y
movement above a 5px threshold, to open the longpress menu.
The 5px threshold is the minimum default used by Keyman Engine for Web.
However, it may not be ideal, and we should consider moving to using the
0.25 x row height value that Keyman Engine for Web uses in a future
update. However, I do not consider this to be a reason to block this
fix, as it would require significant extra engineering.
