Happens when going into fullscreen when playing a video.
There is a conflict between reinterpret_cast and memcpy
when using uint64_t as a storage. When memcpy'ing a pointer
into uint64_t value, and then reinterpret_cast'ing it back
on Big Endian, you'll get a null pointer value because
pointer was stored in top uint64_t bits, and reinterpret_cast
works on lower bits. Fix makes ReinterpretHelper public -
this is not a good practice to make implementation detail
accessible, but ForEach is already bad anyway providing raw
uint64_t value to the user. I don't think that
std::visit-like interface is possible because there's no
direct link between property and parameter stored. Also
upstream Firefox has this bug too.
This interface inly has a single implementation behind it, which is also
only used in 2 places after the previous commit. That's a lot of
additional complexity and compiler indirection for no good reason.
This change removes the interface and uses direct nsDocShell::Cast calls
instead of going through the interface in the few places left now that
we no longer build on a presentation context for links.
This removes the requirement for there to be a non-null PresShell to
dispatch `Click()` events on `<A>` elements (only), since the exception
to the rule has propagated to the spec.
With these changes it should now be possible do create an anchor and
`Click()` on it from JS without actually first attaching it to the DOM
of the presented document, as abused by scripted downloads in pages
(instead of using the A attribute to custom-name downloads).
This removes a lot of the plumbing for having the platform embed itself
through IPC which was required for B2G running the browser as both
shell and browser application.
However, `-moz-outline-radius` should still override any `border-radius`.
This is primarily done for backward compatibility with some themes.
Additionally, it also allows for more advanced outline control than the CSS spec provides, without breaking spec.
Finally, if the spec is ever updated to include `outline-radius`, we'll be ready to drop the `-moz-` prefix!
Note: BZ 315209 has an inadvertent double-negative in nsDisplayList.cpp:
HasRadius() essentially returns the opposite value it should.
- Renames it to NS_CSS_FULL_CORNERS and uses the one macro for all full
corners. In preparation for giving half corners the same treatment.
- Swaps to prefix operator++
This implementation is different than Mozilla, not indirectly queueing cancel
tasks to be arbitrarily ordered, but rather synchronously close in reverse
top layer order.
Adds --disable-npapi as an additional build option. The default is still to build in NPAPI compatibility, but with this flag, it's finally possible to build without pulling in GTK2 on systems that don't have the package.
This essentially ifdefs everything associated with NPAPI, following various Mozilla removal bugs.
Ref: BZ 1682030, BZ 1251202, BZ 1217665, BZ 1687239
in the fallback case.
This ensures we don't scale the underlying size of the layer beyond what
is close to the current display size. When box shadows get much larger
than this, they start taking so much time to render that successive
frames grow in scale too fast for any inter-frame reuse to be possible.
With this, we avoid that and no longer get crushed by re-rendering
gigantic box shadows every single frame.
See BZ 1383825
Only blur one quadrant of a box-shadow and mirror it to the other quadrants.
This applies only if the corners are symmetrical (square corners or equal
corner radii) otherwise we'll fall back to the old method.
Jeremy Andrews
Sergey Larin
Moonchild
Francis Dominic Fajardo
MeladJM
Andy
FranklinDM
Martok
Brian Smith