Closed Bug 1311270 Opened 9 years ago Closed 8 years ago

Implement rendering of mask-clip/mask-origin: fill-box|stroke-box|view-box|no-clip values

Categories

(Core :: Layout, defect, P2)

Product:
Core
Component:
Layout
Type:
defect

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla53
Tracking Flags:
Tracking Status
firefox53 --- fixed

People

(Reporter: ethlin, Assigned: u459114)

References

(Blocks 1 open bug)

Details

Attachments

(9 files, 4 obsolete files)

Bug 1311270 - Part 1. Move geomery box computing from nsCSSClipPathInstance to nsLayoutUtils.
58 bytes, text/x-review-board-request
heycam
: review+
TYLin
: review+
Details
Bug 1311270 - Part 2. Rename StyleClipPathGeometryBox to StyleGeometryBox and extend contained values.
58 bytes, text/x-review-board-request
heycam
: review+
TYLin
: review+
Details
Bug 1311270 - Part 3. Replace {bg|mask}-origin/{bg|mask}-clip constants by StyleGeometryBox.
58 bytes, text/x-review-board-request
heycam
: review+
TYLin
: review+
Details
Bug 1311270 - Part 4. Handle rendering of mask-clip:no-clip.
58 bytes, text/x-review-board-request
mstange
: review+
Details
Bug 1311270 - Part 5. Handle rendering of mask-clip:fill-box/stroke-box/view-box.
58 bytes, text/x-review-board-request
mstange
: review+
Details
Bug 1311270 - Part 6. Handle mask-origin.
58 bytes, text/x-review-board-request
mstange
: review+
Details
Bug 1311270 - Part 7. Rename local variable in GetImageLayerClip and ComputeImageLayerPositioningArea.
58 bytes, text/x-review-board-request
mstange
: review+
Details
Bug 1311270 - Part 9. Test cases for new mask-clip/mask-origin values.
58 bytes, text/x-review-board-request
heycam
: review+
TYLin
: review+
Details
Bug 1311270 - Part 8. Add assertion for margin-box.
58 bytes, text/x-review-board-request
mstange
: review+
Details
Per Bug 1303623 comment 4, create another bug for rendering issues of mask-clip/mask-origin in nsCSSRendering + nsDisplayMask + nsSVGIntergrationUtils.
Blocks: mask-image
I hope I can fix this bug in 53.
Assignee: nobody → cku
Status: NEW → ASSIGNED
Priority: -- → P2
Depends on: 1319667
Update WIP patches. Several things that still unfinished 1. mask-origin: fill-box/ stroke-box/ view-box/ margin-box 2. Test case for all these new mask property values.
Attachment #8813093 - Attachment is obsolete: true
Attachment #8813798 - Attachment is obsolete: true
Attachment #8813801 - Attachment is obsolete: true
Comment on attachment 8813092 [details] Bug 1311270 - Part 1. Move geomery box computing from nsCSSClipPathInstance to nsLayoutUtils. https://reviewboard.mozilla.org/r/94614/#review95612 I assume ComputeSVGReferenceRect(), ComputeHTMLReferenceRect(), and the logic inside ComputeGeometryBox() are simply copied and pasted to nsLayoutUtils. r=TYLin ::: layout/base/nsLayoutUtils.cpp:9253 (Diff revision 3) > + return r; > +} > + > +/* static */ nsRect > +nsLayoutUtils::ComputeGeometryBox(nsIFrame* aFrame, > + StyleClipPathGeometryBox geometryBox) Nit: Please fix this indentation.
Attachment #8813092 - Flags: review?(tlin) → review+
Comment on attachment 8813092 [details] Bug 1311270 - Part 1. Move geomery box computing from nsCSSClipPathInstance to nsLayoutUtils. https://reviewboard.mozilla.org/r/94614/#review95616 ::: layout/base/nsLayoutUtils.h:2870 (Diff revision 3) > * next line exists, null otherwise. > */ > static bool IsInvisibleBreak(nsINode* aNode, nsIFrame** aNextLineFrame = nullptr); > > + static nsRect ComputeGeometryBox(nsIFrame* aFrame, > + StyleClipPathGeometryBox geometryBox); Nit: aGeometryBox. ::: layout/base/nsLayoutUtils.cpp:9144 (Diff revision 3) > return lineNonEmpty; > } > + > +static nsRect > +ComputeSVGReferenceRect(nsIFrame* aFrame, > + StyleClipPathGeometryBox geometryBox) Nit: Again, aGeometryBox. ::: layout/base/nsLayoutUtils.cpp:9219 (Diff revision 3) > + return r; > +} > + > +static nsRect > +ComputeHTMLReferenceRect(nsIFrame* aFrame, > + StyleClipPathGeometryBox geometryBox) Nit: aGeometryBox. ::: layout/base/nsLayoutUtils.cpp:9253 (Diff revision 3) > + return r; > +} > + > +/* static */ nsRect > +nsLayoutUtils::ComputeGeometryBox(nsIFrame* aFrame, > + StyleClipPathGeometryBox geometryBox) Nit: Also aGeometryBox.
Comment on attachment 8813094 [details] Bug 1311270 - Part 2. Rename StyleClipPathGeometryBox to StyleGeometryBox and extend contained values. https://reviewboard.mozilla.org/r/94618/#review95624 r- because I feel it'll be better not to remove margin-box support for clip-path in this patch. ::: layout/base/nsLayoutUtils.cpp (Diff revision 3) > - case StyleClipPathGeometryBox::NoBox: > - case StyleClipPathGeometryBox::Border: > - case StyleClipPathGeometryBox::Content: > - case StyleClipPathGeometryBox::Padding: > - case StyleClipPathGeometryBox::Margin: > + case StyleGeometryBox::NoBox: > + case StyleGeometryBox::Border: > + case StyleGeometryBox::Content: > + case StyleGeometryBox::Padding: > + case StyleGeometryBox::Fill: { > - case StyleClipPathGeometryBox::Fill: { margin-box is required by clip-path. We should keep this. ::: layout/base/nsLayoutUtils.cpp:9231 (Diff revision 3) > - case StyleClipPathGeometryBox::Margin: > - r = aFrame->GetMarginRectRelativeToSelf(); > - break; > + case StyleGeometryBox::NoBox: > + case StyleGeometryBox::Border: > + case StyleGeometryBox::Fill: I think StyleGeometryBox::Margin is still needed. ::: layout/style/nsStyleConsts.h:143 (Diff revision 3) > -enum class StyleClipPathGeometryBox : uint8_t { > - NoBox, > +// background-origin, mask-clip and mask-origin. > +enum class StyleGeometryBox : uint8_t { > Content, > Padding, > Border, > - Margin, > + Margin, // clip-path reference-box only. Better to mention Bug 1260094 here, and say something like "Although margin-box is reqired by mask-origin and mask-clip, we don't implement that due to the lack support of other browsers. ::: layout/style/nsStyleConsts.h:149 (Diff revision 3) > - Fill, > - Stroke, > - View, > + Fill, // mask-clip, mask-origin and clip-path reference-box only. > + Stroke, // mask-clip, mask-origin and clip-path reference-box only. > + View, // mask-clip, mask-origin and clip-path reference-box only. > + NoClip, // mask-clip only. > + Text, // background-clip only. > + NoBox, // Depend on which kind of element this style value applied on, Nit: s/Depend/Depending/ ::: layout/style/nsStyleConsts.h:151 (Diff revision 3) > - View, > + View, // mask-clip, mask-origin and clip-path reference-box only. > + NoClip, // mask-clip only. > + Text, // background-clip only. > + NoBox, // Depend on which kind of element this style value applied on, > + // the initial value of a reference-box can be different. > + // For a HTML element, the initial value of reference-box is Nit: s/a HTML/an HTML ::: layout/style/nsStyleConsts.h:151 (Diff revision 3) > + // For a HTML element, the initial value of reference-box is > + // border-box; for a SVG element, the initial value is fill-box. Nit: s/a HTML/an HTML/, and s/a SVG/an SVG/
Attachment #8813094 - Flags: review?(tlin) → review-
Comment on attachment 8813095 [details] Bug 1311270 - Part 3. Replace {bg|mask}-origin/{bg|mask}-clip constants by StyleGeometryBox. https://reviewboard.mozilla.org/r/94620/#review95632 r=TYLin with the inline comments addresssed. ::: layout/style/nsComputedDOMStyle.h (Diff revision 3) > > already_AddRefed<CSSValue> GetCSSShadowArray(nsCSSShadowArray* aArray, > const nscolor& aDefaultColor, > bool aIsBoxShadow); > > - already_AddRefed<CSSValue> GetBackgroundList( Any reason to convert the two methods GetBackgroundList() and GetRoCSSValueList() to *non-static* function in cpp? They are both marked /\*static\*/ which it is untrue. I'll prefer them remain as methods. ::: layout/style/nsComputedDOMStyle.cpp:1805 (Diff revision 3) > > return val.forget(); > } > > + > +/*static*/ template <typename T> Nit: We usually don't have a space between template and <> ::: layout/style/nsStyleStruct.h:802 (Diff revision 3) > // or an ImageValue.) > mozilla::Position mPosition; // [reset] > Size mSize; // [reset] > - uint8_t mClip; // [reset] See nsStyleConsts.h > + StyleGeometryBox mClip; // [reset] See nsStyleConsts.h > MOZ_INIT_OUTSIDE_CTOR > - uint8_t mOrigin; // [reset] See nsStyleConsts.h > + StyleGeometryBox mOrigin; // [reset] See nsStyleConsts.h Nit: Please align StyleGeometryBox, and the comment.
Attachment #8813095 - Flags: review?(tlin) → review+
Comment on attachment 8813095 [details] Bug 1311270 - Part 3. Replace {bg|mask}-origin/{bg|mask}-clip constants by StyleGeometryBox. https://reviewboard.mozilla.org/r/94620/#review95632 > Nit: Please align StyleGeometryBox, and the comment. MOZ_INIT_OUTSIDE_CTOR StyleGeometryBox mOrigin These two lines is broken from one line, so this identation is needed.
Comment on attachment 8814304 [details] Bug 1311270 - Part 9. Test cases for new mask-clip/mask-origin values. https://reviewboard.mozilla.org/r/95564/#review95646 ::: layout/reftests/w3c-css/submitted/masking/mask-origin-3-ref.html:6 (Diff revision 2) > +<!DOCTYPE html> > +<html> > + <head> > + <meta charset="utf-8"> > + <title>CSS Masking: mask-origin: mask positioning area</title> > + <link rel="author" title="Astley Chen" href="mailto:aschen@mozilla.com"> Please correct the author name.
Attachment #8814304 - Flags: review?(tlin) → review+
Comment on attachment 8813094 [details] Bug 1311270 - Part 2. Rename StyleClipPathGeometryBox to StyleGeometryBox and extend contained values. https://reviewboard.mozilla.org/r/94618/#review95650
Attachment #8813094 - Flags: review?(tlin) → review+
Attachment #8813799 - Flags: review?(mstange)
Attachment #8813800 - Flags: review?(mstange)
Attachment #8813802 - Flags: review?(mstange)
Attachment #8814302 - Flags: review?(mstange)
Comment on attachment 8813092 [details] Bug 1311270 - Part 1. Move geomery box computing from nsCSSClipPathInstance to nsLayoutUtils. https://reviewboard.mozilla.org/r/94614/#review95652 s/geomery/geometry/ in the commit message.
Attachment #8813092 - Flags: review?(cam) → review+
Comment on attachment 8813094 [details] Bug 1311270 - Part 2. Rename StyleClipPathGeometryBox to StyleGeometryBox and extend contained values. https://reviewboard.mozilla.org/r/94618/#review95656 ::: layout/style/nsStyleConsts.h:145 (Diff revision 4) > Content, > Padding, > Border, > - Margin, > - Fill, > - Stroke, > + Margin, // XXX Bug 1260094 comment 9. > + // Although margin-box is required by mask-origin and mask-clip, we > + // don't implement that due to lack support of other browser. Nit: s/lack support of other browser/lack of support in other browsers/. ::: layout/style/nsStyleConsts.h:152 (Diff revision 4) > + NoBox, // Depending on which kind of element this style value applied on, > + // the initial value of a reference-box can be different. > + // For an HTML element, the initial value of reference-box is > + // border-box; for an SVG element, the initial value is fill-box. > + // Since we can not determine the initial value at parsing time, > + // set it as NoBox so that we make a decision later. > + // clip-path reference-box only. Nit: I would say "default" rather than "initial", because "initial" sounds like it is talking about the CSS property initial value (which for clip-path is 'none').
Attachment #8813094 - Flags: review?(cam) → review+
Comment on attachment 8813095 [details] Bug 1311270 - Part 3. Replace {bg|mask}-origin/{bg|mask}-clip constants by StyleGeometryBox. https://reviewboard.mozilla.org/r/94620/#review95660 ::: layout/painting/nsCSSRendering.cpp:3350 (Diff revision 4) > // Background images are tiled over the 'background-clip' area > // but the origin of the tiling is based on the 'background-origin' area > // XXX: Bug 1303623 will bring in new origin value, we should iterate from > - // NS_STYLE_IMAGELAYER_ORIGIN_MARGIN instead of > - // NS_STYLE_IMAGELAYER_ORIGIN_BORDER. > + // StyleGeometryBox::Margin instead of > + // StyleGeometryBox::Border. Remove this comment? I guess it's out of date now that we don't have margin-box for background. ::: layout/style/nsComputedDOMStyle.cpp:1819 (Diff revision 4) > +nsComputedDOMStyle::GetBackgroundList(StyleGeometryBox nsStyleImageLayers::Layer::* aMember, > + uint32_t nsStyleImageLayers::* aCount, > + const nsStyleImageLayers& aLayers, > + const KTableEntry aTable[]) The function isn't too long, so it's not so bad, but it would be nice to avoid the duplication if possible. Especially if later we convert other image layer properties (like background-attachment, etc.) to be enum classes. I think just making GetBackgroundList a template function with a template argument T used in place of StyleGeometryBox in the aMember type should work. ::: layout/style/nsComputedDOMStyle.cpp:1828 (Diff revision 4) > + uint8_t index = static_cast<uint8_t>(aLayers.mLayers[i].*aMember); > + val->SetIdent(nsCSSProps::ValueToKeywordEnum(index, aTable)); I think you shouldn't need to cast to a uint8_t here, since there is a templated nsCSSProps::ValueToKeywordEnum function. ::: layout/style/nsRuleNode.cpp:6735 (Diff revision 4) > +template <> > +struct BackgroundItemComputer<nsCSSValueList, StyleGeometryBox> > +{ > + static void ComputeValue(nsStyleContext* aStyleContext, > + const nsCSSValueList* aSpecifiedValue, > + StyleGeometryBox& aComputedValue, > + RuleNodeCacheConditions& aConditions) > + { > + aComputedValue = > + static_cast<StyleGeometryBox>(aSpecifiedValue->mValue.GetIntValue()); > + } > +}; I think you could avoid this copy of BackgroundItemComputer too, if you want, by making the uint8_t version a bit more general, like: template<typename T> struct BackgroundItemComputer<nsCSSValueList, EnableIf<IsIntegral<T>::value || IsEnum<T>::value>::Type> { static void ComputeValue(..., T& aComputedValue, ...) { aComputedValue = static_cast<T>(...); } }; Again, we'll probably need to do that later anyway, once we convert more image layer properties to enum classes. ::: layout/style/nsStyleStruct.h:802 (Diff revision 4) > // or an ImageValue.) > mozilla::Position mPosition; // [reset] > Size mSize; // [reset] > - uint8_t mClip; // [reset] See nsStyleConsts.h > + StyleGeometryBox mClip; // [reset] See nsStyleConsts.h > MOZ_INIT_OUTSIDE_CTOR > - uint8_t mOrigin; // [reset] See nsStyleConsts.h > + StyleGeometryBox mOrigin; // [reset] See nsStyleConsts.h We could still align the comment, though...
Attachment #8813095 - Flags: review?(cam) → review+
Attachment #8814303 - Attachment is obsolete: true
Attachment #8814303 - Flags: review?(jfkthame)
Blocks: 1300384
Comment on attachment 8813799 [details] Bug 1311270 - Part 4. Handle rendering of mask-clip:no-clip. https://reviewboard.mozilla.org/r/95160/#review96618
Attachment #8813799 - Flags: review?(mstange) → review+
Comment on attachment 8813800 [details] Bug 1311270 - Part 5. Handle rendering of mask-clip:fill-box/stroke-box/view-box. https://reviewboard.mozilla.org/r/95162/#review96620 ::: layout/painting/nsCSSRendering.cpp:1903 (Diff revision 5) > MOZ_ASSERT(!aDirtyRect->IsEmpty() || aDirtyRectGfx->IsEmpty(), > "second should be empty if first is"); > } > > +static bool > +IsSVGStyleGeomertyBox(StyleGeometryBox aBox) Geomerty -> Geometry, here and in other places ::: layout/painting/nsCSSRendering.cpp:1922 (Diff revision 5) > + if (aForFrame->IsFrameOfType(nsIFrame::eSVG) && > + (aForFrame->GetType() != nsGkAtoms::svgOuterSVGFrame)) { Could this check be replaced with (aFrame->GetStateBits() & NS_FRAME_SVG_LAYOUT)? I don't really know the exact differences between the two. Ah, I see that further down you assert that these are actually the same thing. So I think it would be good to just use NS_FRAME_SVG_LAYOUT in both places, and maybe even drop the asserts (up to you). ::: layout/painting/nsCSSRendering.cpp:1959 (Diff revision 5) > + nsPoint toStrokeBoxOffset = clipArea.TopLeft() - strokeBox.TopLeft(); > + aClipState->mBGClipArea = > + nsRect(aBorderArea.TopLeft() + toStrokeBoxOffset, clipArea.Size()); This doesn't seem quite right. Can you explain what each of these rects is relative to, and why subtracting and adding the points in this way is the right thing to do? What would happen if you just did aClipState->mBGClipArea = clipArea;?
Comment on attachment 8813802 [details] Bug 1311270 - Part 6. Handle mask-origin. https://reviewboard.mozilla.org/r/95166/#review96624 ::: layout/painting/nsCSSRendering.cpp:3469 (Diff revision 5) > + nsPoint toStrokeBoxOffset = > + bgPositioningArea.TopLeft() - strokeBox.TopLeft(); > + > + return nsRect(toStrokeBoxOffset, bgPositioningArea.Size()); So is this assuming that the returned rect needs to be relative to the stroke box? Is that the case?
Comment on attachment 8814302 [details] Bug 1311270 - Part 7. Rename local variable in GetImageLayerClip and ComputeImageLayerPositioningArea. https://reviewboard.mozilla.org/r/95560/#review96626
Attachment #8814302 - Flags: review?(mstange) → review+
Comment on attachment 8813802 [details] Bug 1311270 - Part 6. Handle mask-origin. https://reviewboard.mozilla.org/r/95166/#review96624 > So is this assuming that the returned rect needs to be relative to the stroke box? Is that the case? I would say relative to aBoderArea(which is border-box for HTML element and stroke-box for SVG element). But since aBorderArea == stroke-box, the answer is yes.
Comment on attachment 8813800 [details] Bug 1311270 - Part 5. Handle rendering of mask-clip:fill-box/stroke-box/view-box. https://reviewboard.mozilla.org/r/95162/#review96620 > Could this check be replaced with (aFrame->GetStateBits() & NS_FRAME_SVG_LAYOUT)? I don't really know the exact differences between the two. > > Ah, I see that further down you assert that these are actually the same thing. So I think it would be good to just use NS_FRAME_SVG_LAYOUT in both places, and maybe even drop the asserts (up to you). I should change that assertion. Unlike other SVG elements, <svg> is the only SVG element associate with CSS layout box. I learn that from heycam's review. [1] https://bugzilla.mozilla.org/show_bug.cgi?id=1289011#c25 > This doesn't seem quite right. Can you explain what each of these rects is relative to, and why subtracting and adding the points in this way is the right thing to do? > What would happen if you just did aClipState->mBGClipArea = clipArea;? clipArea - According to mask-clip prop value, it can be view-box, fill-box or stroke-box in *svg user space*. strokeBox - stroke-box in *svg user space*. aBoderArea - Depend on the caller, it can be stroke-box area in the *refernence-frame's space* or in *svg user space*, or anything else. The coordinate space of the returning value(aClipState->mBGClipArea/mDirtyRect etc) - be the same with aBoderArea. we can not always set "aClipState->mBGClipArea = clipArea" directly, since clipArea is in svg user space, but according aBoderArea we need mBGClipArea be in reference-frame's space sometimes. I will add more comments to explain detail in code.
Attachment #8815630 - Flags: review?(mstange)
Comment on attachment 8815630 [details] Bug 1311270 - Part 8. Add assertion for margin-box. https://reviewboard.mozilla.org/r/96498/#review96790
Attachment #8815630 - Flags: review?(mstange) → review+
(In reply to C.J. Ku[:cjku](UTC+8) from comment #72) > > This doesn't seem quite right. Can you explain what each of these rects is relative to, and why subtracting and adding the points in this way is the right thing to do? > > What would happen if you just did aClipState->mBGClipArea = clipArea;? > > clipArea - According to mask-clip prop value, it can be view-box, fill-box > or stroke-box in *svg user space*. > strokeBox - stroke-box in *svg user space*. > aBoderArea - Depend on the caller, it can be stroke-box area in the > *refernence-frame's space* or in *svg user space*, or anything else. > The coordinate space of the returning > value(aClipState->mBGClipArea/mDirtyRect etc) - be the same with aBoderArea. > > we can not always set "aClipState->mBGClipArea = clipArea" directly, since > clipArea is in svg user space, but according aBoderArea we need mBGClipArea > be in reference-frame's space sometimes. > I will add more comments to explain detail in code. Thanks, I'll need to think about this some more. Sorry that this is taking so long.
Comment on attachment 8814304 [details] Bug 1311270 - Part 9. Test cases for new mask-clip/mask-origin values. https://reviewboard.mozilla.org/r/95564/#review100990 Sorry, forgot about this one.
Attachment #8814304 - Flags: review?(cam) → review+
Comment on attachment 8813800 [details] Bug 1311270 - Part 5. Handle rendering of mask-clip:fill-box/stroke-box/view-box. https://reviewboard.mozilla.org/r/95162/#review101128 Ok, it makes sense to me now. I think it would be easier to understand if it were written like this: ``` nsRect clipArea = ...; nsRect strokeBox = (backgroundClip == StyleGeometryBox::Stroke) ? clipArea : nsLayoutUtils::ComputeGeometryBox(aForFrame, StyleGeometryBox::Stroke); nsRect clipAreaRelativeToStrokeBox = clipArea - strokeBox.TopLeft(); aClipState->mBGClipArea = clipAreaRelativeToStrokeBox + aBorderArea.TopLeft(); ``` but I can see how that would be slightly slower.
Attachment #8813800 - Flags: review?(mstange) → review+
Comment on attachment 8813802 [details] Bug 1311270 - Part 6. Handle mask-origin. https://reviewboard.mozilla.org/r/95166/#review101132 ::: layout/painting/nsCSSRendering.cpp:3494 (Diff revision 7) > + nsLayoutUtils::ComputeGeometryBox(aForFrame, > + StyleGeometryBox::Stroke); > + toStrokeBoxOffset = bgPositioningArea.TopLeft() - strokeBox.TopLeft(); > + } > + > + return nsRect(toStrokeBoxOffset, bgPositioningArea.Size()); Please add a comment that says that for SVG frames the return value is relative to the stroke box.
Attachment #8813802 - Flags: review?(mstange) → review+
Pushed by cku@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/6d1cf53de2b4 Part 1. Move geomery box computing from nsCSSClipPathInstance to nsLayoutUtils. https://hg.mozilla.org/integration/mozilla-inbound/rev/3f11d4c79554 Part 2. Rename StyleClipPathGeometryBox to StyleGeometryBox and extend contained values. https://hg.mozilla.org/integration/mozilla-inbound/rev/9fc272b52c9a Part 3. Replace {bg|mask}-origin/{bg|mask}-clip constants by StyleGeometryBox. https://hg.mozilla.org/integration/mozilla-inbound/rev/36b977bb8bbe Part 4. Handle rendering of mask-clip:no-clip. https://hg.mozilla.org/integration/mozilla-inbound/rev/6509a168badf Part 5. Handle rendering of mask-clip:fill-box/stroke-box/view-box. https://hg.mozilla.org/integration/mozilla-inbound/rev/1b0ac433a100 Part 6. Handle mask-origin. https://hg.mozilla.org/integration/mozilla-inbound/rev/d83d4dd9f36a Part 7. Rename local variable in GetImageLayerClip and ComputeImageLayerPositioningArea. https://hg.mozilla.org/integration/mozilla-inbound/rev/0627bbd0b8c3 Part 8. Add assertion for margin-box. https://hg.mozilla.org/integration/mozilla-inbound/rev/c82b1188b47a Part 9. Test cases for new mask-clip/mask-origin values.
Depends on: 1341149
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: