[Webkit-unassigned] [Bug 96960] Turn on ENABLE_MATHML for Chromium

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=96960


Dave Barton <dbarton at mathscribe.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |christophe.dumez at intel.com,
                   |                            |davidc at nag.co.uk




--- Comment #6 from Dave Barton <dbarton at mathscribe.com>  2012-09-18 12:36:34 PST ---
Other than checking for crashers, I think our two main issues will be:

1. Most of the MathML tests are pixel tests. I still need to tweak MathML layout sometimes, and when I do, it often triggers a ~1MB patch (sigh). This is almost all test results in LayoutTests/platform/*/mathml/presentation. Actually, so far the only platforms are mac, efl, and gtk. I verify the results in platform/mac, and lately Christophe Dumez usually checks the efl and gtk ones. I'd be happy to check results on various platforms if gardeners want to upload them or e-mail a zip file to me. Actually WebKit MathML isn't used yet in the wild too much, so we don't have to be too paranoid about small changes. If you check the .png files quickly and they look similar to the platform/mac ones, then I'd just rebaseline (.png and .txt) if I were you. If your .png files look bad, then please upload or send your results to me.

2. The biggest platform issue for MathML will doubtless be fonts. We can minimize this problem by ensuring the STIX fonts are on all the bots - then we might even be able to combine some pixel test baselines. On Mac bots the OS must be at least OS X 10.7.4, or else the STIX fonts come out always italic. Can someone verify the OS version on the Chromium Mac bots? There are also some issues on linux - search LayoutTests/platform/efl/Skipped for "mathml"; and see bug 47744; and bug 38377, comment 2 (I need to fix that, I think). But the biggest issue is probably -webkit-line-box-contain on Windows - see <http://code.google.com/p/chromium/issues/detail?id=77095>. Christophe Dumez was able to fix this on linux fairly quickly in bug 93073 - see its comments 8, 11 and 12, and its patch. It'd be great if a Windows font expert could do this on Windows also. Bug 48124 may also be an issue on Windows.

Note the Apple Mac port doesn't enable subpixel layout yet I believe, so its MathML baselines may be slightly different from Chromium.

I think the summary is that it'll be great to test MathML for crashes, and hopefully we can also get it looking good on Mac, Linux, and Windows soon. If not, we may have to turn it off on some platforms for now. But MathML in Chrome is still exciting!

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.