Bug 155516 – ASSERT_NOT_REACHED on imported/w3c/web-platform-tests/html/semantics/embedded-content/the-area-element/area-coords.html

• Status: RESOLVED
• Resolution: DUPLICATE
• Component: Tools / Tests

[Alexey Proskuryakov]

imported/w3c/web-platform-tests/html/semantics/embedded-content/the-area-element/area-coords.html asserts every time.

It's been marked as flaky upon importing, but it doesn't seem useful to have it crash.

[Alexey Proskuryakov]

Updated test results in r198232.

[Alexey Proskuryakov]

rdar://problem/24929497

[Alexey Proskuryakov]

1   com.apple.WebCore             	0x000000010aedcaa1 WebCore::minimumValueForLength(WebCore::Length const&, WebCore::LayoutUnit, bool) + 305
2   com.apple.WebCore             	0x000000010a395dcc WebCore::HTMLAreaElement::getRegion(WebCore::LayoutSize const&) const + 2060
3   com.apple.WebCore             	0x000000010a395159 WebCore::HTMLAreaElement::mapMouseEvent(WebCore::LayoutPoint, WebCore::LayoutSize const&, WebCore::HitTestResult&) + 137
4   com.apple.WebCore             	0x000000010a43fda2 WebCore::HTMLMapElement::mapMouseEvent(WebCore::LayoutPoint, WebCore::LayoutSize const&, WebCore::HitTestResult&) + 242
5   com.apple.WebCore             	0x000000010b3b4d92 WebCore::RenderImage::nodeAtPoint(WebCore::HitTestRequest const&, WebCore::HitTestResult&, WebCore::HitTestLocation const&, WebCore::LayoutPoint const&, WebCore::HitTestAction) + 546
6   com.apple.WebCore             	0x000000010b493bd6 WebCore::RenderObject::hitTest(WebCore::HitTestRequest const&, WebCore::HitTestResult&, WebCore::HitTestLocation const&, WebCore::LayoutPoint const&, WebCore::HitTestFilter) + 118
7   com.apple.WebCore             	0x000000010a6438c0 WebCore::InlineElementBox::nodeAtPoint(WebCore::HitTestRequest const&, WebCore::HitTestResult&, WebCore::HitTestLocation const&, WebCore::LayoutPoint const&, WebCore::LayoutUnit, WebCore::LayoutUnit, WebCore::HitTestAction) + 352
8   com.apple.WebCore             	0x000000010a64cba3 WebCore::InlineFlowBox::nodeAtPoint(WebCore::HitTestRequest const&, WebCore::HitTestResult&, WebCore::HitTestLocation const&, WebCore::LayoutPoint const&, WebCore::LayoutUnit, WebCore::LayoutUnit, WebCore::HitTestAction) + 979
9   com.apple.WebCore             	0x000000010b63874a WebCore::RootInlineBox::nodeAtPoint(WebCore::HitTestRequest const&, WebCore::HitTestResult&, WebCore::HitTestLocation const&, WebCore::LayoutPoint const&, WebCore::LayoutUnit, WebCore::LayoutUnit, WebCore::HitTestAction) + 410
...

[zalan]

It asserts because ".4" string is considered an invalid integral number and newCoordsArray() does not handle this error case very well.
However it seems the spec changed so that now coords attribute can have floating point numbers. (https://html.spec.whatwg.org/multipage/embedded-content.html#attr-area-coords)
So instead of fixing newCoordsArray(), we should move over to floating-point attribute parsing.

[Ahmad Saleem]

(In reply to zalan from comment #4)
> It asserts because ".4" string is considered an invalid integral number and
> newCoordsArray() does not handle this error case very well.
> However it seems the spec changed so that now coords attribute can have
> floating point numbers.
> (https://html.spec.whatwg.org/multipage/embedded-content.html#attr-area-
> coords)
> So instead of fixing newCoordsArray(), we should move over to floating-point
> attribute parsing.

https://searchfox.org/wubkat/rev/dcd42ae35333b577043cbe9e3eda1c4a5c66a471/Source/WebCore/html/HTMLAreaElement.cpp#74

We have 'parseHTMLListOfOfFloatingPointNumberValues(newValue.string());'.

Do we need anything else? We don't have it as flaky expectation for any platform.

We can mark this as 'RESOLVED CONFIGURATION CHANGED' IMO.

[zalan]

progressed at r205095

*** This bug has been marked as a duplicate of bug 161216 ***

[zalan]

actually progressed at r205030 (duplicate bugzilla is still valid)