[webkit-dev] HTML5 Parsing & MathML

[Adam Barth]

On Tue, Nov 2, 2010 at 9:52 AM, David Carlisle <d.p.carlisle at gmail.com> wrote:
> On 2 November 2010 16:26, Adam Barth <abarth at webkit.org> wrote:
>> On Tue, Nov 2, 2010 at 7:55 AM, David Carlisle <d.p.carlisle at gmail.com> wrote:
>>> Alex Milowski <alex <at> milowski.org> writes:
>>>
>>> sorry for late reply, I'm not subscribed, just saw this in the archives.
>>>
>>>>
>>>> On Fri, Oct 1, 2010 at 12:52 PM, Adam Barth <abarth <at> webkit.org> wrote:
>>>> > Our parser follows the spec (modulo late-breaking spec changes that we
>>>
>>> Actually most mathml in the wild will be mis-parsed by the webkit html5 parser
>>> because of
>>>
>>> https://bugs.webkit.org/show_bug.cgi?id=48105
>>>
>>> but that's hopefully a temporary glitch.
>>
>> Is this a bug in the HTML5 specification or a bug in our
>> implementation of the spec?  If its the former, you might want to file
>> a bug with the HTML working group to resolve the issue.
>>
>> Adam
>
> I'm pretty sure that it is an implementation issue (firefox 4 doesn't
> have this problem for example). Certainly I can't see anything that
> would specify parsing something as simple as
>
>
> <math>
> <mrow>
> <mrow><mn>1</mn></mrow>
> <mi>a</mi>
> </mrow>
> </math>
>
> as a completely different tree:
>
> <math> <mrow> <mrow><mn>1</mn></mrow></mrow> <mi>a</mi> </math>
>
> It makes mathml and svg pretty unusable of course as it's common (very
> common in mathml case) to have elements nested within an element of
> the same name.

Okiedokes.  I've CCed Eric and myself on the bug since we're the
mostly likely folks to fix the issue.  We'd certainly welcome a patch
from you, if you're interested in fixing the issue.

Adam