[Webkit-unassigned] [Bug 170272] MathML: Extra left space for arrows inside <mo>

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=170272

--- Comment #19 from Minsheng Liu <lambda at liu.ms> ---
Here comes the bad news. If we want to move layout code out of paint(), some non-trivial change is required.

For the cases you have mentioned, the DOM involves more than one layer of <mover>. It looks like:
\mover{x + ... + x}{\mover{horizontal brace}{k times}}
The current logic layouts the structure as follows:
1. layout {x + ... + x}.
2. layout {\mover{horizontal brace}{k times}}. (Note the width is incorrect!}
3. stretch the horizontal brace without updating its location.

And fixing everything in operator’s paint().

Despite being ad-hoc—it is unclear to me whether the original fix in paint() is or can be made correct or not—one consequence of this algorithm is that the width of the stretched part of a <mover> element is its original content width. I just checked myself that MathML 3 does not specify the width in such scenario (width is not mentioned at all in the section regarding <munder>/<mover>), but this felt just wrong and would make Inspector very hard to use in inspecting the bounds of those elements. I checked that in Firefox, the width of \mover{horizontal brace}{k times} is the same as that of {x + ... + x}.

So,
1. I think we should change the current behavior and use the more natural width, which involve a non-trivial change to the layout algorithm.
2. Since you are more familiar with MathML standard, does the width of a stretched operator affect its parents? If the standard is unclear in this area, should we send something to the committee?

— Minsheng

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20171114/b25f386f/attachment.html>