This can be used to test Decoder Doctor on Windows, but disabling WMF through
prefs, to trigger an Decoder Doctor infobar.
MozReview-Commit-ID: 1aRk111VFO7
For making TextInputHandler MOZ_LOG* environment variables aware, TextInputHandler should use LazyLogModule.
MozReview-Commit-ID: 9La229BFaJ1
--HG--
extra : rebase_source : f18f689ea6522f3300c5411bcd9e49d4c31143e4
This is preparation. TISInputSourceWrapper is created before starting XPCOM, however, when its first instance is created, TextInputHandler.mm tries to log all keyboard layouts and IMEs which are installed into the system. This would be problem if it uses LazyLogModule because it's initialized at starting XPCOM.
MozReview-Commit-ID: DWz8TylL175
--HG--
extra : rebase_source : f377530f6325d6fcf8f90be5d6856972e9d312e5
TSFTextStore should use insertion point relative query for getting character rect and caret rect while there is a composition because in such case, TSF must want to query a part of or around the composition. Therefore, it makes sense to use insertion point relative query since it adjusts the offset with the latest content information.
MozReview-Commit-ID: IVjZ4zqFUkr
--HG--
extra : rebase_source : e535ffa2c7649f16ffe7f761a9ed01b061848aa7
IMM always queries character rect or caret rect relative to insertion point. Therefore, it makes sense to use the new feature, insertion point relative query content events for them. Then, IMMHandler doesn't need to care mismatch between its cache and actual content information.
MozReview-Commit-ID: LCCrJkR77I8
--HG--
extra : rebase_source : c75f1b29e34e13df397de351909f07e2a95b6c5f
In e10s mode and during focus is in a remote process, ContentCache handles WidgetQueryContentEvent instead of ContentEventHandler. Therefore, for supporting selection relative WidgetQueryContentEvent in e10s mode, we need to support it in ContentCache too.
MozReview-Commit-ID: L5d5ilmpetG
--HG--
extra : rebase_source : 46d646677ac0a5803398d18efbfd667bc3bf0c17
IMEContentObserver may have cache of normal selection. If it's available, IMEContentObserver should use it for computing absolute offset of WidgetQueryContentEvent whose mInput::mOffset is relative offset to selection.
This patch just improves the performance of such query.
MozReview-Commit-ID: KHLgCc2uQzs
--HG--
extra : rebase_source : 1367aee0aadb88258135690aa5a8591201129c27
Native IME handler may want to query content relative to start of selection (or composition if there is it). Additionally, in e10s mode, insertion point in actual content may be different from the cache in parent. Therefore, in some cases, it does make sense to query content with offset relative to start of selection or composition.
This patch implements it simply and only in non-e10s mode.
Additionally, this fixes a bug of nsQueryContentEventResult::GetOffset() which hasn't been accepted its calls even if the event message is valid (eQueryTextContent, eQueryTextRect and eQueryCaretRect).
MozReview-Commit-ID: 34I7vyTUAgO
--HG--
extra : rebase_source : d79ba0dc3e002f7691495ee1ff8bdb3854d8f6fe
This will only run if they receive a WGL context, and we have a blocklist
entry for this. Since there isn't a reason right now to want WGL without
DXGL, we should tie these together.
MozReview-Commit-ID: ymtVT5dJIZ
Test that paths to test modules, directories, and manifest files are
processed correctly by BaseMarionetteTestRunner.add_test.
Individual test modules should be added
even if they have invalid test names.
If a directory path is provided, all validly-named
test modules discovered by walking the dir should be added.
Manifest files should be parsed and tests added/skipped accordingly.
MozReview-Commit-ID: 2GJFEGGcEW2
--HG--
extra : rebase_source : 44c752f1025be715b44833ced08cc315c5becaa6
We do all validation in our WebGL layer, we only really generate GL_OOM
from the driver, no other errors.
This is opposed to SkiaGL, which may or may not rely on errors returned
from the driver. This is why I am not just turning on ABORT_ON_ERROR for
all GLContexts.
(Also, later, we should be creating no-error contexts on non-DEBUG builds,
so the driver doesn't replicate our validation. ANGLE already supports
this)
MozReview-Commit-ID: A4RSgzUXfHA