This fixes typing `n>` with sil_ipa keyboard which no longer output
the expected `ŋ` because pressing the shift key reset the context.
For modifier key presses we now exit `KMX_ProcessEvent::ProcessEvent`
early.
This change also removes a redundant line from engine.c. Clearing
the context happens in `reset_context()` again if necessary.
Fixes#5591.
This change improves capslock handling with a capsAlwaysOff keyboard.
Previously, the caps lock indicator was turned on while holding down
the Caps Lock key. With this change we turn it off immediately. This
is done by forcing caps lock off even if the state already seems to
match.
Also, for capsAlwaysOff keyboards skip further processing of caps
lock key after we turned capslock off again.
Fixes#5465.
Fixes#5561.
If a non-BMP Unicode char was emitted into the internal context, it was
being truncated to a WORD, which caused data loss for multi-group
keyboards.
Includes an update to a unit test to validate this fix.
Moves the responsibility for deletion of markers when the users presses
the backspace key, from `kmx_processor::process_event` to
`KMX_ProcessEvent::ProcessGroup`. This has no effect on usage of Keyman
Core, but:
1. Simplifies the association between context and action further
2. Ensures that `m_actions` and `state->actions` will always be the
same length, making future refactoring of these simpler
3. Allows us to use the index into the m_actions array for Debug events,
which simplifies the interleaving of these for the upcoming Debugger.
When the user presses backspace at start of context (where there may be
markers in the cached context, but no characters), the kmx processor
will delete any markers prior to insertion point before requesting that
the consumer emit the backspace virtual key back to the application, to
allow the application to handle backspace at start of text.
Fixes#5488.
This updates Keyman Core, corresponding tests, and Keyman Engine for
Linux to support deletion of markers through an action, ensuring that
the action queue does not desynchronize with the context.
Adds some tests for Keyman Core for debugging, verifying contents of
the action list after every keystroke in the test.
Also refactors some of the test helper files to make it easier to
diagnose failing tests and share helper functions.
Adds an action_index to each debug event (not all debug events
need it). This allows the debugger to partially execute a rule
including the possibility of multiple updates to context.
Adds debug events to the kmx processor and corresponding unit tests.
This corresponds largely to the way that keyman32 emits debug events,
but with one key change: debug events in keyman32 are processed
synchronously, whereas in Keyman Core, the full set of debug events will
be returned alongside the final action list for the key event.
Given this difference, there will likely need to be some more data
returned in individual debug events, specifically around intermediate
context manipulation, as required for step-by-step debugging. However,
that will come in a subsequent PR.
Part of #5013.
Adds all the debug infrastructure:
* public API headers
* public API implementation classes and functions
* basic unit tests
* helper classes for recording debug events
* minimal implementation of kmx-specific debug event signalling --
just BEGIN and END events.
* tangential: unit test assertion macros and coloured output
As the build environment documentation is duplicated across each of the
Keyman platform projects, it has become somewhat convoluted and
inconsistent. This PR moves all the mac-based and shared documentation
into docs/build, with placeholders for Windows and Linux documentation
to be moved in the next PR.
The idea is that a developer can configure their box to build any of the
available projects for that platform (web and android build on all three
main dev platforms).
Keeping the build documentation in one place makes it much simpler to
keep it consistent and up-to-date as well, as well as keeping us aware
of potential dependency conflicts between projects.
The icing on the cake is the resources/devbox/macos/macos.sh script,
which I have tested and takes a bare metal mac box through to a full
Keyman development environment (except XCode, there's always something!)
The changelogs should match what got uploaded (and accepted) to
Debian so that the changelog reflects the upload history to Debian.
(cherry picked from commit f91b52d3ea)