14
$\begingroup$

The next build will enable Mathjax 2.1 officially. Thanks for your help and feedback!

We will be deploying the latest Mathjax 2.1 beta soon. As with any beta release please post an answer to this question if you find any issues. We will be monitoring this thread closely along with the authors of Mathjax to ensure this release goes smoothly. Here are a list of changes in this release:

MathJax v2.1 is primarily a bug-fix release. Numerous display bugs, line-breaking problems, and interface issues have been resolved. The complete details are available on the MathJax GitHub issue tracker https://github.com/mathjax/MathJax/issues. The following lists indicate the majority of the bugs that have been fixed for this release.

Interface:

  • Make NativeMML output properly handle iOS double-tap-and-hold, and issue warning message when switching to NativeMML output.
  • Use scrollIntoView to handle positionToHash rather than setting the document location to prevent pages from refreshing after MathJax finishes processing the math.
  • Handle positioning to a hash URL when the link is to an element within SVG output.
  • Make href's work in SVG mode in all browsers.
  • Fix problem with opening the "Show Math As" window in WebKit (affected Chrome 18, and Safari 5.1.7).
  • Use MathJax message area rather than window status line for 'maction' with actiontype='statusline' to avoid security restrictions in some browsers.
  • Fix issue where zoom box for math that has been wrapped to the beginning of a line would be positioned at the end of the previous line.
  • Fix a problem where IE would try to typset the page before it was completely available, causing it to not typeset all the math on the page (or in some cases any of the math).
  • Allow decimal scale values in the dialog for setting the scale.
  • Fix SVG output so that setting the scale will rescale the existing mathematics.
  • Add close button to About box and don't make clicking box close it (only clicking button).
  • Make About box show 'woff or otf' when otf fonts are used (since both are requested).
  • Have output jax properly skip math when the input jax has had an internal failure and so didn't produce any element jax.
  • Produce MathJax.Hub signal when [Math Processing Error] is generated.
Line-breaking:
  • Fix problem with SVG output disappearing during line breaks when equation numbers are also present.
  • Fix problem with potential infinite loop when an <mspace> is an embellished operator that causes a linebreak to occur.
  • Allow line breaks within the base of <msubsup> to work so that the super and subscripts stay with the last line of the base.
  • Fix <mfenced> so that when it contains a line break the delimiters and separators are not lost.
  • Allow line breaks at delimiters and separators in <mfenced> elements.
  • Fix issue with line breaking where some lines were going over the maximum width.
  • Fix problem with line breaking inside <semantics>elements.
  • Fix problem with line breaking where the incorrect width was being used to determine breakpoint penalties, so some long lines were not being broken.
HTML-CSS/SVG display:
  • Fix several Chrome alignment and sizing issues, including problems with horizontal lines at the tops of roots, fraction bars being too long, etc.
  • Resolve a problem with how much space is reserved for math equations when a minimum font size is set in the browser.
  • Force final math span to be remeasured so that we are sure the container is the right size.
  • Fix alignment problem in <msubsup>.
  • Fix processing error when rowalign has a bad value.
  • Fix a vertical placement problem with stretched elements in mtables in HTML-CSS, and improve performance for placing the extension characters.
  • Handle spacing for U+2061 (function apply) better.
  • Better handling of primes and other pseudo scripts in HTML-CSS and SVG output.
  • Fixed a problem with <mmultiscripts> in SVG mode that caused processing error messages.
  • Fix misplaced \vec arrows in Opera and IE.
  • Make <mi> with more than one letter have texClass OP rather than ORD in certain cases so it will space as a function.
  • Make HTML snippet handler accept a string as contents, even if not enclosed in braces.
  • Fix spacing for functions that have powers (e.g., \sin^2 x).
  • Fix problem with SVG handling of \liminf and \limsup where the second half of the function name was dropped.
  • Fixed a problem where HTML-CSS and SVG output could leave partial equations in the DOM when the equation processing was interrupted to load a file.
  • Fix problems with <mtable>, <ms>, and <mmultiscripts> which weren't handling styles.
  • Make column widths and row heights take minsize into account in <mtable>.
  • Fix typo in handle-floats.js that caused it to not compile.
  • Fix problem in HTML-CSS output with <msubsup> when super- or subscript has explicit style.
TeX emulation:
  • Allow negative dimensions for \\[] but clip to 0 since this isn't really allowed in MathML.
  • Fixed problem where \\ with whitespace followed by [ would incorrectly be interpretted as \\[dimen].
  • Make jsMath2jax run before other preprocessors so that tex2jax won't grab environments from inside the jsMath spans and divs before jsMath2jax sees them.
  • Fix issue with \vec not producing the correct character for \vec{\mathbf{B}} and similar constructs.
  • Combine multiple primes into single unicode characters.
  • Updated the unicode characters used for some accents and a few other characters to more appropriate choices. See issues #116, #119, and #216 in the MathJax issue tracker on GitHub.
  • Remove unwanted 'em' from eqnarray columnwidth values.
  • Make eqnarray do equation numbering when numbering is enabled.
  • Make vertical stretchy characters stand on the baseline, and improve spacing of some stretchy chars.
  • Make mtextFontInherit use the style and weight indicated in the math, so that \textbf and \textit will work properly.
  • Add \textcolor macro to the color extension.
  • Added RGB color model to the color extension.
  • Automatically load the AMSmath extension when needed by the mhchem extension.
  • Add <<=> arrow to mhchecm extension
  • Fix alignment of prescripts in mhchem to properly right-justify the scripts.
  • Expose the CE object in the mhchem extension.
  • Make autoload-all skip extensions that are already loaded, and not redefine user-defined macros.
  • Fix most extensions to not overwrite user defined macros when the extension is loaded.
  • Ignore \label{} with no label.
  • Make \injlim and friends produce single <mi> elements for thier names rather than one for each letter.
  • Handle primes followed by superscript as real TeX does in TeX input jax.
  • Handle a few more negations (e.g., of arrows) to produce the proper Unicode points for these.
  • Don't produce a processing error when \limits is used without a preceding operator.
MathML Handling:
  • Prevent align attribute on <mtable> from applying to <mover>/<munder>/<munderover> elements.
  • Ignore _moz-math-* attributes in MathML input so they don't appear in MathML output
  • Prevent duplicate xmlns attributes in "Show Math As -> MathML".
  • Fixed a problem in MathML output where dimensions given to <mpadded> with leading +'s could lose the plus and become absolute rather than relative.
  • Fix setTeXclass for TeXatom so that it handles the spacing for relations correctly.
  • Add more CSS to isolate NativeMML output from page.
  • Handle setup of MathPlayer better for IE10, and avoid some IE10 bugs in setting the document namespace for MathML.
Fonts:
  • Fix a problem where bold-script didn't work properly in STIX fonts.
  • Work around Chrome bug with MathJax web fonts that affects some combining characters.
  • Remove dependencies of TeX->MathML conversion on the choice of fonts (TeX versus STIX).
  • For stretchy characters that don't have a single-character version in the MathJax fonts, make sure they are properly sized when not stretched or stretched to a small size.
  • Fix an error with U+u005E (^) which caused it to show as a plus when used as a stretchy accent.
  • Fix a problem with greek letters in STIX font producing the wrong letter (an offset was off by one).
  • Handle more characters in sans-serif-italic and bold-italic STIX fonts.
$\endgroup$
15
  • 1
    $\begingroup$ Did these fixes hopefully address the crashing-in-Chrome issue? $\endgroup$ Commented Oct 1, 2012 at 18:08
  • 2
    $\begingroup$ Is this release a big-fix or a bug-fix release? :-) $\endgroup$
    – Asaf Karagila Mod
    Commented Oct 1, 2012 at 20:29
  • 2
    $\begingroup$ @QiaochuYuan, this version does not do anything specific to address the crashes people have seen in Chrome. I have not been able to reproduce it myself (though I do not mean that to indicate that I don't believe others are having it), and so have not been able to investigate it myself. My personal feeling is that it is a memory management issue in Chrome, and that would make it very difficult to resolve within MathJax itself. It also seems closely tied to the SE preview code, as we have not had any reports of it from other sites. $\endgroup$ Commented Oct 1, 2012 at 20:38
  • $\begingroup$ This version does resolve a number of display issues with Chrome, however, such as this and this. $\endgroup$ Commented Oct 1, 2012 at 20:42
  • $\begingroup$ It would be wonderful if support to theorems like environments is added. $\endgroup$
    – leo
    Commented Oct 6, 2012 at 15:22
  • $\begingroup$ this is Chromium from synaptic on lubuntu 12.10 (beta). The following show up wrong for me: $\mathcal L$. $\endgroup$ Commented Oct 8, 2012 at 1:25
  • $\begingroup$ @kjetilbhalvorsen, what do you see instead? Also, do you have the STIX fonts installed? If you open the MathJax About box (using the MathJxx contextual menu) what font mode does it say it is using (the line right under the "MathJax v2.1" title at the top of the box)? $\endgroup$ Commented Oct 8, 2012 at 23:16
  • $\begingroup$ This is what I can read in the about box: MathJax.js v2.1 TeX Input Jax v2.1 HTML-CSS Output Jax v2.1 mml Element Jax v2.1 MathEvents Extension v2.1 MathMenu Extension v2.1 MathZoom Extension v2.1 TeX/AMSmath Extension v2.1 TeX/AMSsymbols Extension v2.1 TeX/noErrors Extension v2.1 TeX/noUndefined Extension v2.1 tex2jax Extension v2.1 Chrome v20.0.1132.47 $\endgroup$ Commented Oct 9, 2012 at 0:43
  • $\begingroup$ It also says: "using local stix fonts".. This is approximately what I see: ">.-" $\endgroup$ Commented Oct 9, 2012 at 0:45
  • $\begingroup$ @kjetilbhalvorsen, I think I may have found the issue. The fix that was used to work around a WebKit problem with the STIX fonts on Mac OS X Lion was incorrectly being applied to all platforms, not just OSX. I have pushed the fix to the v2.1-beta copy that is being used here at M.SE, so can you check to see if that is working for you now? You might need to clear the cache and restart Chrome to get the updated copy. $\endgroup$ Commented Oct 19, 2012 at 23:52
  • $\begingroup$ I cleared the cache and restarted chromium, seems to work now! Thanks. $\endgroup$ Commented Oct 20, 2012 at 0:20
  • $\begingroup$ @Davide Is there any way for the MathJax preview to locally save drafts, so that an answer can be recovered if the browser/page crashes? The SE draft feature rarely helps out here. Anything, no matter how rough, would be better than nothing. It's still very bad - so bad that I cannot use Chrome (but I often forget and lose work). $\endgroup$ Commented Nov 8, 2012 at 2:23
  • $\begingroup$ Javascript can't save to files (in general) for security reasons, but there is an internal local storage feature that might work for that. The problem is that if the browser crashes, it might not be properly saved, so it might not be reliable. But could be worth a try. Unfortunately, my method of hooking into the preview editor no longer seems to work (since a recent change to the SE code a month or so ago). I have not had the chance to try to get that to work again. $\endgroup$ Commented Nov 8, 2012 at 11:13
  • $\begingroup$ Are you planning on implementing a beta of 2.2 as well? $\endgroup$
    – Asaf Karagila Mod
    Commented May 6, 2013 at 14:57
  • $\begingroup$ Yes, we are planning to go through testing with mathjax 2.2 beta here. $\endgroup$ Commented May 6, 2013 at 15:37

1 Answer 1

5
$\begingroup$

The code

g(x)=\sum_{n\ge 2}\left(\sum_{k\ge 0}\binom{n-1}{2k-1}\right)x^n=x^2\sum_{n\ge 2}\left(\sum_{k\ge 0}\binom{n-1}{2k-1}\right)x^{n-2}=x^2\sum_{n\ge 0}\left(\sum_{k\ge 0}\binom{n+1}{2k-1}\right)x^n\;.

displays correctly here, at least in the preview, but now produces [Math Processing Error] on the main site. (It worked fine when this answer was posted.

$$g(x)=\sum_{n\ge 2}\left(\sum_{k\ge 0}\binom{n-1}{2k-1}\right)x^n=x^2\sum_{n\ge 2}\left(\sum_{k\ge 0}\binom{n-1}{2k-1}\right)x^{n-2}=x^2\sum_{n\ge 0}\left(\sum_{k\ge 0}\binom{n+1}{2k-1}\right)x^n\;.$$

This does not appear to be a browser problem: it occurs both in Firefox 15.0.1 and in Chrome 22.0.1229.79 m (running on Windows 7 32-bit), and repeated reloading of the page does not help.

$\endgroup$
4
  • $\begingroup$ It looks like an issue in the line breaking code. I'll check into it. Thanks for the report. $\endgroup$ Commented Oct 2, 2012 at 11:12
  • $\begingroup$ When this question is shown in the list of linked or related questions (you can try for yourself by clicking on the questions listed among linked and related questions), [Math Processing Error] is shown. See this screenshot. (Davide wrote in a comment to my - now deleted - answer that the reason might be the same as for the above problem. So I am posting this as a comment here.) $\endgroup$ Commented Oct 2, 2012 at 12:33
  • $\begingroup$ Note: for those experiencing this problem, it only affects the HTML-CSS rendering mode, so using the MathJax contextual menu to set Math Settings->Renderer->SVG should give you a proper display until this is fixed. $\endgroup$ Commented Oct 2, 2012 at 14:11
  • $\begingroup$ OK, this issue should now be fixed. It turns out to have been due to not handling elements with empty children (like the sums with lower limits but not upper ones). We pushed the fix to the CDN, so it should now be working for you. $\endgroup$ Commented Oct 4, 2012 at 14:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .