Tables incorrectly baseline-align to first row #891

Closed
opened this issue Aug 18, 2014 · 13 comments

Projects
None yet
4 participants

digama0 commented Aug 18, 2014

 Here is a picture of the "TeX Samples" page on the MathJax homepage: In case it's not obvious, the issue I am referring to is the fact that the table is aligned to the first row, so there is a large white space above the table in the brackets. I can't imagine that this is expected behavior, and if it is I can't figure out what TeX to write to work around this issue. Any ideas as to what is causing this? (Apologies if this is a dupe, I didn't see anything similar.) I believe that the TeX code to generate this image is: \mathbf{V}_1 \times \mathbf{V}_2 = \begin{vmatrix} \mathbf{i} & \mathbf{j} & \mathbf{k} \\ \frac{\partial X}{\partial u} & \frac{\partial Y}{\partial u} & 0 \\ \frac{\partial X}{\partial v} & \frac{\partial Y}{\partial v} & 0 \end{vmatrix} 
Member

pkra commented Aug 18, 2014

 Thanks for reporting this. I'm unable to reproduce this problem. What brower and OS versions are you seeing this with?
Author

digama0 commented Aug 18, 2014

 Firefox version 34.0a1, Windows 8. I can confirm that the issue does not happen on IE 11, don't have any other browsers to test.
Member

pkra commented Aug 18, 2014

 Hm. That's a nightly build, correct? From what I can tell, Aurora is still at 33. I would suggest to file a bug with Mozilla.
Author

digama0 commented Aug 18, 2014

 That's right, this is a Nightly build. Filed with Mozilla.
Member

pkra commented Aug 18, 2014

 Thanks! Here's hoping it's an issue they'll want to fix. If it's still happening when 34 hits Aurora, we'll have to consider working around it.

tschneidereit commented Aug 18, 2014

 I'm afraid we'd rather not change this back. Based on comment 3 on the bug @digama0 filed, this change in behavior is caused by us fixing a spec compatibility bug in our RegExp implementation. Since all other engines implement the behavior we have now, too, it'd, annoyingly, be much better for MathJax to fix this on your side, if possible at all. I'm sorry for the inconvenience this causes.
Member

pkra commented Aug 18, 2014

 Thanks for the update, @tschneidereit. I'm afraid we'd rather not change this back. [...] this change in behavior is caused by us fixing a spec compatibility bug in our RegExp implementation. Thanks for the additional information. Since all other engines implement the behavior we have now, too, it'd, annoyingly, be much better for MathJax to fix this on your side, if possible at all. Hm. Are you sure? I'm only wondering because this problem arises only on Gecko right now (and IIUC the relevant code is not browser dependent).

tschneidereit commented Aug 18, 2014

 Hmm, could this be related to bug 1053944, perhaps?
Member

pkra commented Aug 19, 2014

 @dpvc could you take a look?

Member

dpvc commented Aug 19, 2014

 I think the original comment 3 link indicates the change (from blank string to undefined value) and that it is going to be permanent. It looks like there is a place in MathJax's table alignment code that could be affected by this, and will need to be changed.

dpvc pushed a commit to dpvc/MathJax that referenced this issue Aug 19, 2014

 Forgot to do the corresponding change in SVG output. Issue mathjax#891. 
 944e8ad 

dpvc pushed a commit to dpvc/MathJax that referenced this issue Aug 19, 2014

 Forgot to do the corresponding change in SVG output. Issue mathjax#891. 
 cd5a06d 

dpvc pushed a commit to dpvc/MathJax that referenced this issue Aug 20, 2014

 Merge branch 'issue891' into develop. Issue mathjax#891. 
 1fa3c4f 
Member

dpvc commented Aug 20, 2014

 ==> Merged.

Member

dpvc commented Jan 16, 2015

 I can't reproduce the problem in Firefox 34.0.5 or 35, so am removing the test wanted label.