Align BUG

[mikev3]

Hi, I noticed that there is a bug with \begin{align} when rendering
with MathML. At first the page renders properly, but choose:
1. "Math Renderer -> HTML-CSS"
2. "Math Renderer -> MathML"

And the alignment breaks!

[mikev3]

Forgot this test link! Sorry:

<a href="http://www.rootmath.org/test/align-test">http://
www.rootmath.org/test/align-test</a>

[Davide P. Cervone]

It looks like there is a rendering bug in Firefox that is causing
this. The MathML produced by MathJax is OK, and indeed, if you turn
on the debugger, or do something else that forces a reflow, Firefox
will render it correctly again. I will look into whether I can force
that to occur, but preliminary tests haven't been encouraging.

Davide

[Frédéric WANG]

I guess it's bug
https://bugzilla.mozilla.org/show_bug.cgi?id=491384

--
Frédéric Wang.
Website - Weblog

[Davide P. Cervone]

It does seem similar, but FYI, moving the mouse into the column doesn't fix the alignment (as it does in the bug you reference).  I HAVE seen that behavior in other examples, but this one doesn't seem to have it.

Davide

[Frédéric WANG]

Le 05/08/2011 14:06, Davide P. Cervone a �crit :

> It does seem similar, but FYI, moving the mouse into the column
> doesn't fix the alignment (as it does in the bug you reference). I
> HAVE seen that behavior in other examples, but this one doesn't seem
> to have it.
>
> Davide

Moving the mouse in the column to fix the alignment was reported by Joe
Java, but I can't remember this behavior.

FYI, The bug entry also contains a reduced testcase, here:
https://bug491384.bugzilla.mozilla.org/attachment.cgi?id=427195

[Davide P. Cervone]

I have certainly seen it in other cases where the alignment is
initially wrong. (I know that the Maxwell's Equation array on the TeX
examples page has been one place where that has occurred for me). But
the test given by mikev3 doesn't respond to that (though it does
reformat properly if you zoom in and out).

Davide

On Aug 5, 2011, at 8:12 AM, Frédéric WANG wrote:

[mikev3]

Hi, thanks for the responses. I initially came across this because I
have a page that actually loads improperly to begin with:
http://rootmath.org/calculus/solving-limits If the page is initially
rendered with MathML the alignment is broken for the Summary.

Vidoes Tab -> Click "Summary"

As mentioned, zooming in and out fixes this issue, but the initial
display is wrong.

Mike

[mikev3]

I did some looking around, in the original link: www.rootmath.org/test/align-test
there is a difference in the MathML between the broken alignment and
the correct alignment.


//BROKEN ALIGN
====================
<math xmlns="http://www.w3.org/1998/Math/MathML">
<mtable columnalign="right left right left right left right left
right left right left" rowspacing="3pt"
columnspacing="0.2777777777777778em 2em 0.2777777777777778em 2em
0.2777777777777778em 2em 0.2777777777777778em 2em 0.2777777777777778em
2em 0.2777777777777778em" displaystyle="true">
<mtr>
<mtd>
<mrow>
<munder>
<mo form="prefix" movablelimits="true">lim</mo>
<---- form = "prefix"

//CORRECT ALIGN
====================
<math xmlns="http://www.w3.org/1998/Math/MathML">
<mtable columnalign="right left right left right left right left
right left right left" rowspacing="3pt"
columnspacing="0.2777777777777778em 2em 0.2777777777777778em 2em
0.2777777777777778em 2em 0.2777777777777778em 2em 0.2777777777777778em
2em 0.2777777777777778em" displaystyle="true">
<mtr>
<mtd>
<mrow>
<munder>
<mo movablelimits="true">lim</mo> <----------
the form = "prefix" is removed



This difference is present for every <mo movablelimits="true"> tag.
The broken align includes the 'form="prefix"' the correct align does
not. Sorry for the messy post, if I knew how to upload a file or
screenshot I would do that so I can point you to line numbers ect.

Thanks,
M.

[Frédéric WANG]

Thanks for looking into this.
Did you test with pages that directly use MathML in HTML (without MathJax)?
The form = "prefix" may be added by MathJax without having actually any
effect on the rendering.

Le 05/08/2011 17:18, mikev3 a �crit :

[mikev3]

All I did was view www.rootmath.org/test/align-test Then I viewed the
math source (using mathjax) as MathML. I compared the source of the
broken alignment with the source from the correct alignment. Again
these sources are coming from MathJax. I'm not very tech savy, I've
learned everything I know in pieces on my own so I hope this answers
your question!

M.

[Frédéric WANG]

Thank you for the info, that's what I thought.
Can you please create HTML pages and copy and paste the MathML sources
coming from MathJax (for both the correct and broken alignment), and
tell me what you get with Firefox?

[mikev3]

Ok, I created an HTML page with the copy and pasted MathML source.
Apparently the only formatting difference is the thickness of the line
in the fraction.

www.rootmath.org/test/align-test-html

Thanks,
M.

On Aug 5, 11:36 am, Frédéric WANG <fred.w...@free.fr> wrote:
> Thank you for the info, that's what I thought.
> Can you please create HTML pages and copy and paste the MathML sources
> coming from MathJax (for both the correct and broken alignment), and
> tell me what you get with Firefox?
>
>
>
>
>
>
>

> > All I did was viewwww.rootmath.org/test/align-test Then I viewed the

[mikev3]

correct link: http://rootmath.org/test/align-test-html

[mikev3]

correct link: http://rootmath.org/test/align-test-html

On Aug 5, 12:28 pm, mikev3 <mikevai...@gmail.com> wrote:

[Davide P. Cervone]

Thanks for the detective work. I don't think the form="prefix" is the
culprit, however, as the browser properly displays it after a reflow.
It certainly seems to be a browser bug, not a MathML problem, and one
that occurs only under certain circumstances that are not easy to
generate. I'm glad to have your example that is reproducible, but I
suspect we will have to wait for Firefox to fix this one. You might
consider using the HTML-CSS renderer for Firefox, rather than its
native MathML renderer, as this is more reliable with MathJax, though
somewhat slower.

Davide