Bug 18721 – Ideographic comma or full stop, and subsequent ASCII token should be breakable into two lines

• Status: NEW
• Component: Layout and Rendering

[Eric Seidel (no email)]

This is the same as: https://bugs.webkit.org/show_bug.cgi?id=17411 except for Windows.

[Mark Rowe (bdash)]

It sounds like bug 17411 was fixed on Mac but not Windows.  Does bug 17411 still need to be open and is "Depends on" really the correct relationship between these two bugs?

[Alexey Proskuryakov]

Sadly, Bugzilla doesn't have a "related" relationship, so "depends on" is as good as any.

[Jungshik Shin]

I don't understand how it can be different on Windows from on Mac. The fix was in 'break_line.cpp' and it should fix this issue across platforms.   

Does Safari 3.1.1 on Windows include the patch in bug 17411?  

A more dramatic test case is at https://bugs.webkit.org/attachment.cgi?id=16810

[Alexey Proskuryakov]

I do not think bug 17411 was supposed to be fixed in Safari 3.1.1 - Safari 3.1 branched off the trunk earlier. So, it is not fixed in Safari 3.1.1 for Mac either.

I do not think this bug has much in common with bug 17411, other than being a failure on the same test case (yes, the test at bug URL still fails with Windows ToT).

> A more dramatic test case is at https://bugs.webkit.org/attachment.cgi?id=16810

This test works correctly for me with ToT on both Mac and Windows.

Since WebKit is the same on both platforms, the difference is likely to come from a lower level framework, or maybe even some difference in font data - debugging should tell for sure.

[Jungshik Shin]

> I do not think this bug has much in common with bug 17411, other than being a
> failure on the same test case (yes, the test at bug URL still fails with
> Windows ToT).

Bug 17411 is about lines not being broken after ideographic full stop and comma. This bug is also about that. So is bug 15630. 

The test that came with the patch for issue 17411 is rather fragile as a test case for line breaking of ideographic comma and full-stop  in that it depends on the exact metric of a font used to render 'Webkit'. 

I think 'https://bugs.webkit.org/attachment.cgi?id=16810' is more robust and cross-platform so that I think it's better to put that test in. 

[Alexey Proskuryakov]

(In reply to comment #5)
> I think 'https://bugs.webkit.org/attachment.cgi?id=16810' is more robust and
> cross-platform so that I think it's better to put that test in. 

OK. Please file a new bug in Tools/Tests component - this bug is about a Windows-only failure on a different test case, and discussing another test case that doesn't even fail on any platform does not help to resolve it.