This requires adding a new overloading of LayoutDevicePixel::ToAppUnits and a
new PixelCastJustification: LayoutDeviceIsParentLayerForRCDRSF.
--HG--
extra : rebase_source : b6d044619918818815b40c1173aa4e0551db28a1
The only reason we had this in the scrollframe at all was so that it could be
saved/restored as part of the frame state when leaving a page and then going
back to it. However we can accomplish this by just reading/writing the resolution
from/to the presshell instead, so there's no need to keep a second copy of it.
--HG--
extra : commitid : J4QBfG2GGjn
The casts in nsCSSProps.cpp (defining kBoxSizingKTable) and in
nsComputedDOMStyle::DoGetBoxSizing (using
nsCSSProps::ValueToKeywordEnum) are a little bit annoying, though aren't
a net reduction in typesafety.
The casts in nsRuleNode.cpp (SetDiscrete) are a little more annoying,
though the change in this patch should be sufficient for converting all
properties -- but that may also mean reducing typesafety a bit for all
properties.
I'd like to find something better to do about them, but I think I'm ok
landing this before doing that. Bug 1224918 covers doing better.
--HG--
extra : commitid : GEVyZB5b23F
ForceInside shifts the rect first, and then clamps if needed. So the displayport doesn't get shrunk unnecessarily.
Bug 1191539 fixed this bug by applying ForceInside to the screen rect of the display port, which happens before the incorrect Intersect call.
It's better to remove the Intersect call and just do ForceInside once at the end to the final display port rect.
Bug 957668 introduced this bug by using Intersect instead of ForceInside when copying the code from AsyncPanZoomController::CalculatePendingDisplayPort when creating the code that computed a displayport rect from displayport margins.
The displayport was accidentally being calculated as too large. This was
leading to a large number of tiles existing at once, causing very high
memory usage.
The patch renames the existing functions (GetBounds(), GetClientBounds(), etc)
by adding an |Untyped| suffix. It then adds typed equivalents, and uses those
typed equivalents in all the call sites where it's easy to do so. The trickier
remaining call sites are converted to use the Untyped-suffix version.
--HG--
extra : rebase_source : 6bfb15bfc4698e2eba7d4db55497299d3dffcd51
We may want to do this for fixed pos frames in all documents (not just root documents). However, this patch only maintains the previous behaviour on purpose.
Instead of returning the root scroll frame if we encountered the root frame (which is the parent of the root scroll frame).
This allows the use of GetNearestScrollableFrame to walk up the frame tree without getting into a infinite loop going from root scroll frame to root frame and back.
This regresses bug 1105823 in that fixed pos frames will no longer find the root scroll frame of their document. The next patch will fix that.
The only other type of frame that will be affected when calling GetNearestScrollableFrame are viewport (root) frames. However, the only user of SCROLLABLE_ALWAYS_MATCH_ROOT (APZCCallbackHelper) calls GetNearestScrollableFrame on the result of a hit test on a display list. Viewport frames never create any display items whose HitTest function could return the viewport frame.
SCROLLABLE_ALWAYS_MATCH_ROOT should take precendence over SCROLLABLE_ONLY_ASYNC_SCROLLABLE as if we fail to find the root scroll frame we will have no usable scroll frame at all.
Removing the "stop at ancestor" parameter from functions that compute AGR meant that nsLayoutUtils::GetAnimatedGeometryRootFor could no longer pass the display item's reference frame as the "stop at ancestor" which meant that the AGR could cross the reference frame for the item, which we don't want. So we make transformed frames into AGRs.
This makes the computation of display items whose frames are transformed tricky. We need the AGR of the transform item to be the ancestor AGR, not the underlying frame of the transform item (which is now an AGR). So we modify nsLayoutUtils::GetAnimatedGeometryRootFor to handle this. (The patch from bug 1205087 didn't suffer from this problem because it special cased the computation of the AGR of transform items. Leaving anybody who called nsLayoutUtils::GetAnimatedGeometryRootFor to get the wrong result.)
The computation of the AGR for scroll metadata in ContainerState::ProcessDisplayItems specifically bypassed nsLayoutUtils::GetAnimatedGeometryRootFor to avoid it's special processing of fixed background items. However we do want the AGR for scroll metadata to do this special processing of transform items. So we add a flag to bypass the fixed background behaviour and use it for the scroll metadata AGR.
This removes the "aStopAtAncestor" argument to agr computing functions. In most cases an AGR was passed for the stop at ancestor, so we'd stop at it anyway since it was an AGR. Most of the remaining cases the root reference frame was passed. And in a few cases something else was passed, which we probably don't want (returning something that isn't an AGR and isn't the root reference frame as an AGR).
The ShouldFixToViewport case is a little tricky. We want to get the AGR of the nearest viewport frame, but only if we don't have to cross our root reference frame to get it. This happens in practice for example when a select dropdown has background-attachment: fixed inside it.
Except for the ShouldFixToViewport bit, this patch is a subset of part 3 in bug 1205087 (which has more changes, and has been temporarily backed out, the remaining bits can hopefully land soon).
The ShouldFixToViewport part is by Timothy Nikkel <tnikkel@gmail.com>
The bulk of this commit was generated with a script, executed at the top
level of a typical source code checkout. The only non-machine-generated
part was modifying MFBT's moz.build to reflect the new naming.
CLOSED TREE makes big refactorings like this a piece of cake.
# The main substitution.
find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
xargs perl -p -i -e '
s/nsRefPtr\.h/RefPtr\.h/g; # handle includes
s/nsRefPtr ?</RefPtr</g; # handle declarations and variables
'
# Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h.
perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h
# Handle nsRefPtr.h itself, a couple places that define constructors
# from nsRefPtr, and code generators specially. We do this here, rather
# than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename
# things like nsRefPtrHashtable.
perl -p -i -e 's/nsRefPtr/RefPtr/g' \
mfbt/nsRefPtr.h \
xpcom/glue/nsCOMPtr.h \
xpcom/base/OwningNonNull.h \
ipc/ipdl/ipdl/lower.py \
ipc/ipdl/ipdl/builtin.py \
dom/bindings/Codegen.py \
python/lldbutils/lldbutils/utils.py
# In our indiscriminate substitution above, we renamed
# nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up.
find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \
xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g'
if [ -d .git ]; then
git mv mfbt/nsRefPtr.h mfbt/RefPtr.h
else
hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h
fi
--HG--
rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
This commit was generated using the following script, executed at the
top level of a typical source code checkout.
# Don't modify select files in mfbt/ because it's not worth trying to
# tease out the dependencies currently.
#
# Don't modify anything in media/gmp-clearkey/0.1/ because those files
# use their own RefPtr, defined in their own RefCounted.h.
find . -name '*.cpp' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
grep -v 'mfbt/RefPtr.h' | \
grep -v 'mfbt/nsRefPtr.h' | \
grep -v 'mfbt/RefCounted.h' | \
grep -v 'media/gmp-clearkey/0.1/' | \
xargs perl -p -i -e '
s/mozilla::RefPtr/nsRefPtr/g; # handle declarations in headers
s/\bRefPtr</nsRefPtr</g; # handle local variables in functions
s#mozilla/RefPtr.h#mozilla/nsRefPtr.h#; # handle #includes
s#mfbt/RefPtr.h#mfbt/nsRefPtr.h#; # handle strange #includes
'
# |using mozilla::RefPtr;| is OK; |using nsRefPtr;| is invalid syntax.
find . -name '*.cpp' -o -name '*.mm' | xargs sed -i -e '/using nsRefPtr/d'
# RefPtr.h used |byRef| for dealing with COM-style outparams.
# nsRefPtr.h uses |getter_AddRefs|.
# Fixup that mismatch.
find . -name '*.cpp' -o -name '*.h'| \
xargs perl -p -i -e 's/byRef/getter_AddRefs/g'
For root scroll frames we need information about the async scrolling (or lack thereof) of the scroll frame before we get to ScrollFrameHelper::BuildDisplayList for the scroll frame. We need it in nsLayoutUtils::PaintFrame and nsSubdocumentFrame::BuildDisplayList. So we factor out all the code responsible for async scrolling decisions into one function we can call from all three places.
The conversion is as follows:
- GraphicsFilter::FILTER_NEAREST == gfx::Filter::POINT
- GraphicsFilter::FILTER_GOOD == gfx::Filter::GOOD
- GraphicsFilter::FILTER_BEST == gfx::Filter::LINEAR
Also typedef GraphicsFilter to gfx::Filter; this will be removed in the next
patch.
These changes mean ToFilter() and ThebesFilter() are no longer needed.
The change breaks some scenarios with APZ scrolling, in particular the code
that layerizes the scroll handoff chain for deeply nested scrollable frames.
--HG--
extra : commitid : KeQX4HyPKXd
In various places SetColor() gets passed an nscolor. These are converted
(either implicitly or explicitly) to a gfxRBGA, and then to a gfx::Color.
This patch changes all these cases to avoid the middle step, by (a)
constructing a gfx::Color directly instead of an nscolor, or (b) by converting
an nscolor with Color::FromABGR().
--HG--
extra : rebase_source : ae423a422b37ada8ca0cb6f130350b4a71d58fc4
gfxIntSize is just a typedef of gfx::IntSize, so this is very mechanical. The
only tricky part is deciding for each occurrence whether to replace it with
IntSize, gfx::IntSize or mozilla::gfx::IntSize; in all cases I went with the
shortest one that worked given the existing "using namespace" declarations.
--HG--
extra : rebase_source : 67fd15f87222b16defa70ef795c6d77dfacf1c36
Remove WrapPreserve3DList() and replaced it by creating a
nsDisplayTransform item for each transformed frame.
- Add an additional item for each top frame extending 3D context to
separate consequence contexts.
- Effective transform of a layer is the accumulation of ancestors in
the same 3D context.
- The layers creating new context and extended by children need a
temporary buffer if it's effective transform is not 2D.
- Clip rects are accumulated along the context chain.
- Visible rects of items are computed from dirty regions of the frame
creating the context and accumulated transforms.
- Bounds of items are computed from accumulated transforms and
accumulated bounds of the descent frames.
- Backface hidden is handled by compositor and BasicLayerManager.
--HG--
extra : rebase_source : 1a894b057a117a8520898c9f83308d92c122b180
Remove WrapPreserve3DList() and replaced it by creating a
nsDisplayTransform item for each transformed frame.
- Add an additional item for each top frame extending 3D context to
separate consequence contexts.
- Effective transform of a layer is the accumulation of ancestors in
the same 3D context.
- The layers creating new context and extended by children need a
temporary buffer if it's effective transform is not 2D.
- Clip rects are accumulated along the context chain.
- Visible rects of items are computed from dirty regions of the frame
creating the context and accumulated transforms.
- Bounds of items are computed from accumulated transforms and
accumulated bounds of the descent frames.
- Backface hidden is handled by compositor and BasicLayerManager.
The special-case was added back when the Fennec dynamic toolbar showing/hiding
behaviour didn't resize the content area. As of bug 1180295 that is no longer
the case, and showing/hiding the dynamic toolbar *does* resize the content area.
Therefore the widget bounds always represents the actual content area and can
be used directly as the composition bounds.
In addition, the special-case code relied on a particular quirk of the Fennec
CSS viewport determination code which was removed in bug 1180267. The quirk was
that the CSS viewport height was always set such that it would cover the
available screen height when the page was zoomed out to the minimum allowed
zoom. This behaviour was technically incorrect as it would ignore any height
properties provided in the meta-viewport tag. Now that this quirk has been
removed, the special-case code breaks because the root frame's height is
independent of the screen height and can result in an incorrect composition
bounds.
--HG--
extra : commitid : 69FDhj50WEN