зеркало из https://github.com/mozilla/gecko-dev.git
Bug 1379332: When computing abspos CB content-box size, don't bother subtracting borderpadding if CB is a 0-sized child of a XUL-collapsed frame. r=mats
If border & padding were ignored for sizing the containing block (which can happen, if the containing block is a chlid of a XUL-collapsed frame), then we don't need to subtract border & padding when computing the frame's content-box size for its abspos descendants. MozReview-Commit-ID: JGnzShl8m67 --HG-- extra : rebase_source : 1cead4ce6b403af634c8f9bde0852dce4ee7edc5
This commit is contained in:
Родитель
9ee09abd54
Коммит
828b7077b9
|
@ -1093,6 +1093,14 @@ ReflowInput::ApplyRelativePositioning(nsIFrame* aFrame,
|
|||
}
|
||||
}
|
||||
|
||||
// Returns true if aFrame is non-null, a XUL frame, and "XUL-collapsed" (which
|
||||
// only becomes a valid question to ask if we know it's a XUL frame).
|
||||
static bool
|
||||
IsXULCollapsedXULFrame(nsIFrame* aFrame)
|
||||
{
|
||||
return aFrame && aFrame->IsXULBoxFrame() && aFrame->IsXULCollapsed();
|
||||
}
|
||||
|
||||
nsIFrame*
|
||||
ReflowInput::GetHypotheticalBoxContainer(nsIFrame* aFrame,
|
||||
nscoord& aCBIStartEdge,
|
||||
|
@ -1128,9 +1136,23 @@ ReflowInput::GetHypotheticalBoxContainer(nsIFrame* aFrame,
|
|||
NS_ASSERTION(!(aFrame->GetStateBits() & NS_FRAME_IN_REFLOW),
|
||||
"aFrame shouldn't be in reflow; we'll lie if it is");
|
||||
WritingMode wm = aFrame->GetWritingMode();
|
||||
LogicalMargin borderPadding = aFrame->GetLogicalUsedBorderAndPadding(wm);
|
||||
aCBIStartEdge = borderPadding.IStart(wm);
|
||||
aCBSize = aFrame->GetLogicalSize(wm) - borderPadding.Size(wm);
|
||||
// Compute CB's offset & content-box size by subtracting borderpadding from
|
||||
// frame size. Exception: if the CB is 0-sized, it *might* be a child of a
|
||||
// XUL-collapsed frame and might have nonzero borderpadding that was simply
|
||||
// discarded during its layout. (See the child-zero-sizing in
|
||||
// nsSprocketLayout::XULLayout()). In that case, we ignore the
|
||||
// borderpadding here (just like we did when laying it out), or else we'd
|
||||
// produce a bogus negative content-box size.
|
||||
aCBIStartEdge = 0;
|
||||
aCBSize = aFrame->GetLogicalSize(wm);
|
||||
if (!aCBSize.IsAllZero() ||
|
||||
(!IsXULCollapsedXULFrame(aFrame->GetParent()))) {
|
||||
// aFrame is not XUL-collapsed (nor is it a child of a XUL-collapsed
|
||||
// frame), so we can go ahead and subtract out border padding.
|
||||
LogicalMargin borderPadding = aFrame->GetLogicalUsedBorderAndPadding(wm);
|
||||
aCBIStartEdge += borderPadding.IStart(wm);
|
||||
aCBSize -= borderPadding.Size(wm);
|
||||
}
|
||||
}
|
||||
|
||||
return aFrame;
|
||||
|
|
|
@ -0,0 +1,9 @@
|
|||
<?xml version="1.0"?>
|
||||
<window xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul">
|
||||
<hbox style="visibility: collapse; position: relative">
|
||||
<textbox>
|
||||
<hbox style="position: absolute; width: 10px; height: 10px">
|
||||
</hbox>
|
||||
</textbox>
|
||||
</hbox>
|
||||
</window>
|
|
@ -0,0 +1,9 @@
|
|||
<?xml version="1.0"?>
|
||||
<window xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul">
|
||||
<hbox style="position: relative;visibility: collapse;">
|
||||
<hbox style="padding:5px; border: 5px solid black">
|
||||
<hbox style="position: absolute; width: 10px; height: 10px">
|
||||
</hbox>
|
||||
</hbox>
|
||||
</hbox>
|
||||
</window>
|
|
@ -97,3 +97,9 @@ load 583957-1.html
|
|||
load 617089.html
|
||||
load menulist-focused.xhtml
|
||||
load 716503.html
|
||||
|
||||
# Bug 1379332's tests need stylo to be disabled in order for the
|
||||
# 'visiblity:collapse' to take effect (which is needed to trigger the assertion
|
||||
# that these tests are screening for, in unpatched builds):
|
||||
pref(layout.css.servo.enabled,false) load 1379332-1.xul
|
||||
pref(layout.css.servo.enabled,false) load 1379332-2.xul
|
||||
|
|
Загрузка…
Ссылка в новой задаче