Theoretically this change shouldn't be necessary, however frequent
intermittents suggest that there are still cases where holding a strong
reference here can lead to memory leaks.
Differential Revision: https://phabricator.services.mozilla.com/D109445
This patch:
- adds QM_WARNONLY_TRY/QM_NOTEONLY_TRY macros
- adds QM_WARNONLY_TRY_UNWRAP/QM_NOTEONLY_TRY_UNWRAP macros
- adds QM_OR_ELSE_WARN/QM_OR_ELSE_NOTE sub macros
- replaces non-propagating uses of NS_WARNING with redundant messages by
QM_WARNONLY_TRY
- replaces uses of QM_TRY with orElse by QM_TRY(QM_OR_ELSE_WARN(...))
- replaces uses of QM_TRY inside an extra lambda with QM_WARNONLY_TRY
- replaces uses of QM_TRY with QM_VOID with QM_WARNONLY_TRY.
- replaces uses of QM_TRY with unwanted warnings with QM_NOTEONLY_TRY
- replaces uses of QM_TRY with additional Maybe wrapping for doing a
fallback with QM_TRY(QM_OR_ELSE_WARN(...))
Differential Revision: https://phabricator.services.mozilla.com/D108424
This is needed for bug 1700379, because otherwise we create a reference
frame with the root's scrolled content (the
::-moz-scrolled-page-sequence), and that breaks some display list
invariants.
Always create a canvas frame instead, (doesn't matter when printing
since we print off the page sequence frame directly), and create a
single ::-moz-page-sequence box.
We have to add width: 100% to the UA sheet because we don't get it
automatically set to the scrollport size to by the scrollport anymore.
Otherwise this would regress vertical writing-modes.
Differential Revision: https://phabricator.services.mozilla.com/D109512
This patch changes the value of an empty `<input type=file>` control as obtained from the `FormData` API, which used to be an empty string, to be a `File` object with empty contents, whose name is the empty string and whose type is `"application/octet-stream"`.
Differential Revision: https://phabricator.services.mozilla.com/D106605
Pickup those commits up:
17a63e9 - Fix bug with too small initial skipFract value
6533514 - Improve soundtouch.clear() so that it really clears TDStretch & RateTransposer states
81b0d74 - Correct initial skip value
5e76cf2 - Disable skipping of unaligned SIMD memory offset by default
f38cfa6 - Call "clear()" after changing anti-alias filter on/off
762f560 - Updated versions and documents for release 2.2
1d42d89 - Merge branch 'improve-autovectorization' into 'master'
bf3cec0 - Improvements to help compiler autovectorization
Differential Revision: https://phabricator.services.mozilla.com/D109362
Previously `operator*` returned the iterator, and the templated `operator=` would handle the value insertion.
However that `operator=` prevented a defaulted copy-assignment operator, which is used in some algorithms to overwrite the iterator itself.
So now `operator*` returns a proxy, which implements the templated `operator=` that is used to insert a value. This allows the nsTSetInserter to have a its defaulted `operator=`.
Side-effect: `E` (the hash's KeyClass) is not needed anymore in nsTSetInserter, so it's only templated on the set type.
Tech note: `std::back_insert_iterator` doesn't use a value, but has explicitly-defined `operator=(const typename Container::value_type&)` and `operator=(typename Container::value_type&&)`. Unfortunately, some of our stored value types are references (e.g.: `nsStringHashKey::KeyType` is `const nsAString&`), which, due to reference collapsing, would have made both `operator=`'s have the same signature! That's why this implementation uses a proxy sub-type with a single templated `operator=` that can handle anything.
Differential Revision: https://phabricator.services.mozilla.com/D109587
No behavior change, just compute the different things we need upfront.
At first I was going to use this in the following patch but it ended up
not being needed.
Differential Revision: https://phabricator.services.mozilla.com/D109545
The test was adding a tab with a page that does a console.group call, and then
was refreshing the page and waiting for the next console.group message to be displayed.
The problem seems to be that we weren't waiting for the initial console.group message,
which might mess up the rest of the test.
We're fixing this in the patch, and adding selector to `waitForMessage` so we
are sure to get the console.group message (and not another message, as the location
does have the `hello` world in it, since we're using a data url).
Differential Revision: https://phabricator.services.mozilla.com/D109491
This fixes a regression from bug 1696178 part 5 with dictionary objects.
Object flags on non-last-properties aren't used anywhere, we need to set them on the
last property's shape.
Differential Revision: https://phabricator.services.mozilla.com/D109579
This patch:
- adds QM_WARNONLY_TRY/QM_NOTEONLY_TRY macros
- adds QM_WARNONLY_TRY_UNWRAP/QM_NOTEONLY_TRY_UNWRAP macros
- adds QM_OR_ELSE_WARN/QM_OR_ELSE_NOTE sub macros
- replaces non-propagating uses of NS_WARNING with redundant messages by
QM_WARNONLY_TRY
- replaces uses of QM_TRY with orElse by QM_TRY(QM_OR_ELSE_WARN(...))
- replaces uses of QM_TRY inside an extra lambda with QM_WARNONLY_TRY
- replaces uses of QM_TRY with QM_VOID with QM_WARNONLY_TRY.
- replaces uses of QM_TRY with unwanted warnings with QM_NOTEONLY_TRY
- replaces uses of QM_TRY with additional Maybe wrapping for doing a
fallback with QM_TRY(QM_OR_ELSE_WARN(...))
Differential Revision: https://phabricator.services.mozilla.com/D108424
Those command handlers just check whether the given editor is an `HTMLEditor`
or not. Therefore, we should make them check whether the given editor is
a single line editor or not instead.
Differential Revision: https://phabricator.services.mozilla.com/D108572
