Although LaTeX will render equations properly when converted to Portable Document Format (PDF), the results are only useful for people who do not rely on assistive technologies such as screen readers.

Seriously? If someone asks for vietnamese support (for example diacritics stacking) you will tell him “chat with pdfs because all we care is ascii” ?

If we talk about the blind there is a new software called latex2nemeth (which I codeveloped) and which is standard in TeXLive, MikTeX and MacTeX installations. It goes directly from LaTeX to Braille with minimal editing of the LaTeX file. For math it supports Nemeth which is the standard notation for math for the blind in many countries including US. So for such applications the choice of PDF is wrong.

You may say “screenreaders?” This is again wrong choice as long as we talk about Math. You can not read a math document to a blind person and expect him/her to follow. It will not work. Unless we talk about primary school level.

Currently other solutions (libluis for LibreOffice docs, Duxbury for Word and others) fail badly on LaTeX files, at least at the college level Math.

The software (GPL3) was developed for a blind student that insisted to enter the Math department (usually people avoid it because there was no way to have books to read in braille) and she just graduated using latex2nemeth. The first blind Mathematician in Greece after about 60 years.

There’s no reason, in principle, why you can’t have a plugin to display PDF directly within the messaging application - so it **is** at your fingertips. That may not be good for editing back and forth however.

Seeing as we are digressing quite a bit … you will notice that fancy equations in Wikipedia are done as SVG (for good rendering on the screen) *but* the SVG still contains the original LaTeX, which may give better a11y behaviour e.g. convert on the client-side into Nemeth Braille (as per the previous post) and ignore the SVG itself.

To me it comes back to: a messaging protocol and application is a transport mechanism; what the recipient *does* with the message is up to it.

Nitpick: SVG is a fallback, MathML is there by default. At least for me with scripts disabled. MathML is already accessible.

Good point. That’s there too (and it too contains the original LaTeX). Our cup runneth over.

I don’t know whether my browser (Firefox) is *doing* anything with the MathML. Right mouse “Open Image in New Tab” is present and opens the SVG but that doesn’t *prove* that Firefox is ignoring the MathML. Firefox is supposed to have native support for MathML, according to “the internet”.

(Edit: Using the Inspector tool within Firefox, the MathML elements don’t line up properly with what’s on the screen but are approximately right. It is as if it did invisibly render the MathML and then ignored that and displayed the SVG instead.)

Maybe I need to test the default browser on the Librem 5 …

I think that was the case when I came up with a user style to display the MathML. Perhaps they leave it only for the accessibility benefits? Either way, SVGs displayed black-on-black, so I added this CSS and it’s now fine:

```
.mwe-math-mathml-a11y {
opacity: unset;
position: unset;
clip: unset;
overflow: unset;
}
.mwe-math-fallback-image-inline {
display: none;
}
dl dd {
overflow: unset;
}
```

Cool. I used the Inspector tool to change the CSS interactively, using the hints that you have given, and suddenly the layout and exact appearance change subtly even though it’s the same underlying equations - and it appears to be rendering the MathML now and not the SVG.

Like you, I am wondering why “the fallback is the default”. ???