17
$\begingroup$

Did MathJax get an update? Or something wrong in Chrome?

Snapshot of question title

In a question on $\mathbb R/\mathbb Q,$ the \mathbb Q seems way bolder than the $\mathbb R,$ and looks strange and wrong. I checked, and all other questions using $\mathbb Q$ look similar. It is almost like it is using the wrong font.

I tried quitting Chrome and starting it again, no luck.

There is a workaround - right-click on an expression, select "Math Settings > Math Renderer > Common HTML".

I am running Chrome on a MacBook:

Version 111.0.5563.146 (Official Build) (x86_64)

My Mac is on MacOS 13.2.1 (22D68).

It is not happening on my iPad in either Chrome or Safari. Nor on Safari on my Mac. Just wondering if it is just me - something about my computer or Chrome.

Here is what it looks like in Safari:

enter image description here

$\endgroup$
10
  • 5
    $\begingroup$ I can confirm the wrong look in Chrome on macOS. Does not happen with Safari or Firefox – perhaps report it here: github.com/mathjax/MathJax/issues ? $\endgroup$
    – Martin R
    Apr 10, 2023 at 16:48
  • 3
    $\begingroup$ Thanks. It looks like there hasn't been a new MathJax release since last year, so it seems more likely it is a Chrome issue? @MartinR This release of Chrome came out 4/4/2023. $\endgroup$ Apr 10, 2023 at 17:13
  • 3
    $\begingroup$ I have no idea. But note that you can choose a different MathJax renderer. With “Common HTML” it renders correctly in my Chrome browsers. $\endgroup$
    – Martin R
    Apr 10, 2023 at 17:20
  • 2
    $\begingroup$ I had a brief period a day or two ago with mathbb characters looking "off". The only thing I noticed that I had to switch my JavaScript permissions (I use NoScript), so that "cloudflare.com" was allowed, and not (something like) "d29842934872.cloudflare.com", which is what had been required for months, if not years. Probably not the same thing you're experiencing, just throwing a possible cloudflare reorganization possibly affecting things into the mix. (Oh yeah, Firefox, Windows 11). $\endgroup$
    – JonathanZ
    Apr 10, 2023 at 17:22
  • 2
    $\begingroup$ Yep, selecting Common HTML renderer fixes things. @MartinR $\endgroup$ Apr 10, 2023 at 17:29
  • 1
    $\begingroup$ So, is one of the renderers recommended over the others? Or is the idea that there are so many browser/OS configurations that the answer is "just keeping trying 'em until you get one that works"? $\endgroup$
    – JonathanZ
    Apr 10, 2023 at 19:35
  • 2
    $\begingroup$ presumably, either the prior versions of Chrome was somehow using a different renderer by default, or the new version broke something that makes the default render work this way. Hard to tell which, but it seems more likely to me that Chrome changed behavior of the old default renderer. @JonathanZsupportsMonicaC $\endgroup$ Apr 10, 2023 at 19:40
  • 2
    $\begingroup$ It may also be relevant that Stack Exchange uses an outdated version of MathJax: meta.stackexchange.com/q/385913/196432, math.meta.stackexchange.com/q/35316/42969 $\endgroup$
    – Martin R
    Apr 10, 2023 at 21:09
  • 3
    $\begingroup$ Does this answer your question? Why using \mathcal command produce a square? $\endgroup$
    – Alex M.
    Apr 12, 2023 at 15:40
  • 3
    $\begingroup$ It is known that the combination of the HTML-CSS renderer, Chrome and Mac Ventura does not display some notations correctly. See here and here for other similar problems. $\endgroup$
    – Alex M.
    Apr 12, 2023 at 15:43

1 Answer 1

3
$\begingroup$

The HTML-CSS renderer in MathJax v2 tries to detect whether there are local copies of the MathJax or STIX fonts and will use those in place of web-based fonts if they are available (to reduce the amount of information that a user must download over the network). Apple has shipped the STIX1 fonts as part of Mac OS and iOS for a number years, but in Ventura, they have switched to the STIX2 fonts. These are packaged differently (the STIX1 fonts came in 29 separate fonts, but STIX2 combines these into 5, one each for regular, bold, italic, and bold-italic, plus a math font). In order to try to keep things backward compatible, Apple still allows access to the STIX1 fonts by (what appears to be) mapping the original 29 fonts into the new 5 fonts. This is largely effective, but not completely. In particular, the calligraphic characters are not part of the Unicode standard, and so STIX includes them in the Private Use Are (PUA) of their fonts. Unfortunately, the location in the PUA Changed between STIX1 and STIX2, and Apple mapping from the old fonts to the new ones doesn't take that into account.

The upshot is, when MathJax asks the local system whether STIXGeneral-Regular (the old STIX1 font) is available, Ventura says "yes" and MathJax believes that the 29 original fonts are there, and tries to use them in the layout that the original fonts actually have. But in Ventura, the calligraphic characters are not in the same location, so you get the missing character symbol instead.

This issue only affects the HTML-CSS output, since it is the only one that uses local STIX fonts. So if you switch to the CommonHTML or SVG output (using the MathJax contextual menu if you are a user, or changing the configuration file if you manage the website), or configuring MathJax to not look for local STIX fonts, then that should show these characters as they should be. Also, version 3, and the upcoming version 4, doesn't use local STIX fonts, and so updating to a current version of MathJax would also resolve the issue.

$\endgroup$
2

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