If we re-use the old item we hit the warning because we try to set the very same item as the glass item, but that is fine.
Differential Revision: https://phabricator.services.mozilla.com/D39308
--HG--
extra : moz-landing-system : lando
This requires replacing inclusions of it with inclusions of more specific prefs
files.
The exception is that StaticPrefsAll.h, which is equivalent to StaticPrefs.h,
and is used in `Codegen.py` because doing something smarter is tricky and
suitable for a follow-up. As a result, any change to StaticPrefList.yaml will
still trigger recompilation of all the generated DOM bindings files, but that's
still a big improvement over trigger recompilation of every file that uses
static prefs.
Most of the changes in this commit are very boring. The only changes that are
not boring are modules/libpref/*, Codegen.py, and ServoBindings.toml.
Differential Revision: https://phabricator.services.mozilla.com/D39138
--HG--
extra : moz-landing-system : lando
Not sure if we need this, but the window dragging regions have it, so in case we need to turn off retained display lists we have this and there will be no difference to before the patches in this bug.
Depends on D38589
Differential Revision: https://phabricator.services.mozilla.com/D38590
--HG--
extra : moz-landing-system : lando
For retained display lists it needs to be a weak frame region to areas get removed for modified frames and the rest stick around.
Each display item can contribute an opaque region but WeakFrameRegion can only track one rect per frame. It turns out that we only return rects from GetOpaqueRegion except in one hit testing case.
This still means more than one display item per frame could be contributing to the opaque region, we would miss the second and further rects in that case.
Differential Revision: https://phabricator.services.mozilla.com/D38589
--HG--
extra : moz-landing-system : lando
Currently it's completely unclear at use sites that the getters for `once`
static prefs return the pref value from startup, rather than the current pref
value. (Bugs have been caused by this.) This commit improves things by changing
the getter name to make it clear that the pref value obtained is from startup.
This required changing things within libpref so it distinguishes between the
"base id" (`foo_bar`) and the "full id" (`foo_bar` or
`foo_bar_DoNotUseDirectly` or `foo_bar_AtStartup` or
`foo_bar_AtStartup_DoNotUseDirectly`; the name used depends on the `mirror` and
`do_not_use_directly` values in the YAML definition.) The "full id" is used in
most places, while the "base id" is used for the `GetPrefName_*` and
`GetPrefDefault_*` functions.
(This is a nice demonstration of the benefits of the YAML file, BTW. Making
this change with the old code would have involved adding an entry to every
single pref in StaticPrefList.h.)
The patch also rejigs the comment at the top of StaticPrefList.yaml, to clarify
some things.
Differential Revision: https://phabricator.services.mozilla.com/D38604
--HG--
extra : moz-landing-system : lando
This distinguishes better between the overloaded aspect of the PerFrameKey and the
actual mixed value.
Differential Revision: https://phabricator.services.mozilla.com/D37804
--HG--
extra : moz-landing-system : lando
This static method is assumed to have the same signature as the type's constructor,
and so we must have an implementation of ComputePerFrameKey for each constructor
a display item provides that is called by MakeDisplayItem. Notably this excludes
the MakeClone constructor for a lot of items.
There is a default varargs implementation on nsDisplayItem which everyone
inherits by default, so types which previously didn't overload this method
still don't need to.
Providing an implementation of ComputePerFrameKey on some display item type
shadows the varargs implementation, so one doesn't need to worry about overloading
one constructor but forgetting about another -- if you do, the compiler will only
see the overload and complain that the signature doesn't match.
One slightly annoying result of this is that display items which previously
inherited an overloaded implementation from a superclass now must provide
their own manual implementations. Although as far as I could tell, all of
those cases had a trivial implementation of key=0 (the super class supported
custom keys but the subclasses didn't make use of it).
In those cases I just hardcoded key=0, but it's possible that it would be
better to call into the superclass' implementation to be more robust to changes.
Differential Revision: https://phabricator.services.mozilla.com/D37803
--HG--
extra : moz-landing-system : lando
Also move the impl to the .cpp file to avoid expensive rebuilds when it
is modified.
Differential Revision: https://phabricator.services.mozilla.com/D37589
--HG--
extra : moz-landing-system : lando
Before this patch, we would use fallback for all border images. Now for
all but vector images we will use the WebRender border images
primitives. Vector images are an exception because the fallback is
clever in that it upscales the vector image and clips to only draw the
region it requires. This avoids artifacting but to do something similar
for WebRender as it is currently defined, we would increase our CPU and
memory footprint as we would need to produce the entire vector image
upscaled, not just the parts we need. In the future we should change
WebRender to accept different image resources for each of the segments.
Differential Revision: https://phabricator.services.mozilla.com/D37093
I missed in bug 1487216 that the pres arena memory reporting assumes that the
entry indices are frame class ids, which means that we're reporting some display
list arena entries as frames, which is obviously wrong.
Cleanup a bit nsPresArena to remove the custom id concept, and report also
individual display item type memory usage.
Differential Revision: https://phabricator.services.mozilla.com/D35368
--HG--
extra : moz-landing-system : lando
With retained display lists, a content render root might get marked as not
needing a build, in which case the nsDisplayRenderRoot::CreateWRCommands
function does an early exit. In this case, we don't mark the associated
WebRenderUserData as used during the display list build, which causes it to
get deleted at the end of the transaction. The next transaction that
doesn't early-exit will re-create the WebRenderUserData with a new boundary
object. The compositor therefore thinks it's a brand new thing and, if
conditions are right, could end up destroying and re-creating much of the
APZC tree. That in turn can have effects like discarding paint-skipped
scrolling.
This patch ensures we always touch the WebRenderUserData during the display
list build, so we don't discard it. This problem may still affect nested
nsDisplayRenderRoot instances but I don't think we ever cases where those
occur.
Depends on D36386
Differential Revision: https://phabricator.services.mozilla.com/D36387
--HG--
extra : moz-landing-system : lando
If we decide to just go with an overlay that sits fully over the
window (which I don't personally see a perf problem with right now,
but correct me if you can think of one), then this should be all
we need.
Differential Revision: https://phabricator.services.mozilla.com//D33819
--HG--
extra : rebase_source : 44a5af47f9c10071b0933931fbf3708978f549e4
The patch also removes the layers.mlgpu.enable-container-resizing pref, because
it's dead.
Differential Revision: https://phabricator.services.mozilla.com/D36159
--HG--
extra : rebase_source : e215d584aed18f865d2e8d00a78e76e9b0323e6e
This restores our previous and per-spec behavior. Comparing only ratios was not
correct in the case one of the dimensions was zero and thus not scaled.
Differential Revision: https://phabricator.services.mozilla.com/D35571
--HG--
extra : moz-landing-system : lando
This changes CreateClippedDrawTarget so that instead of taking
a max size and a transform it just takes a user space rect of
the desired bounds.
This change allows the caller to not worry about the computing
a max size based on the current clip. Instead this responsibility
is lowered into the specific backends.
The main motivation for this work is to allow blob recoordination
to create recordings that don't depend on the current clip.
Some additional benefits are that the API is easier to use and
as can be seen simplifies the SVG masking code because it doesn't
need to track surface offsets manually.
It's also an important step towards removing all the uses of
gfxContext::GetClipExtents which will let us get rid of the separate
clipping stack in gfxContext and help us move off of gfxContext
completely.
Most backend implementations of CreateClippedDrawTarget are relatively
simple. DrawTargetCapture is modified to track the current clip rect
so that it can create a new DrawTargetCapture of the appropriate size
without needing to worry about lazy resolution.
Differential Revision: https://phabricator.services.mozilla.com/D33363
--HG--
extra : moz-landing-system : lando
This changes CreateClippedDrawTarget so that instead of taking
a max size and a transform it just takes a user space rect of
the desired bounds.
This change allows the caller to not worry about the computing
a max size based on the current clip. Instead this responsibility
is lowered into the specific backends.
The main motivation for this work is to allow blob recoordination
to create recordings that don't depend on the current clip.
Some additional benefits are that the API is easier to use and
as can be seen simplifies the SVG masking code because it doesn't
need to track surface offsets manually.
It's also an important step towards removing all the uses of
gfxContext::GetClipExtents which will let us get rid of the separate
clipping stack in gfxContext and help us move off of gfxContext
completely.
Most backend implementations of CreateClippedDrawTarget are relatively
simple. DrawTargetCapture is modified to track the current clip rect
so that it can create a new DrawTargetCapture of the appropriate size
without needing to worry about lazy resolution.
Differential Revision: https://phabricator.services.mozilla.com/D33363
--HG--
extra : moz-landing-system : lando
visibleRect should affect both WR/layers, and scaling will only affect
layers.
Differential Revision: https://phabricator.services.mozilla.com/D34525
--HG--
extra : rebase_source : 49273db3e10088c9493f693d5f8b58864dcb01aa