Re: New RTL math support in MathML Core / Chromium browsers

Hi Stephen,

The mirroring for stretchy math operator we added in MathML Core was 
just to do the same technique that exists for normal text (see section 
HTML RTL mirroring at 
https://people.igalia.com/fwang/mathml-operator-mirroring-explainer.html). 
So basically we mirror the base glyph and then apply to the mirrored 
glyph the stretching rules that are described in OpenType MATH / TeX 
which only uses size variants or glyph assembly. This approach is 
probably not flexible enough for the Arabic case.

There is a separate issue about using variable fonts to get better 
stretchy operator support which I believe would better address the use 
case you are mentioning but I believe we lack some data in OpenType MATH 
before we can do that in MathML Core: 
https://github.com/w3c/mathml-core/issues/122

Frédéric Wang

Le 22/07/2025 à 14:31, Stephen Watt a écrit :
> I had one question, independent of this superb achievement, related to 
> glyph rendering:
>
> In many math symbols we have varying width strokes that look as though 
> they are stylistic imitations of using a chisel-edge pen.   In these, 
> stroke thickness depends on stroke orientation.
>
> So the question is:  Does glyph mirroring provide rendering where the 
> strokes in square roots, etc,  are consistent with the kashida, ijam, 
> etc in fonts with varying stroke width?
>
> Stephen
>
> On Tue, Jul 22, 2025 at 8:31 AM Stephen Watt <smwatt@gmail.com> wrote:
>
>     This is fantastic.   Yes, very hearty congratulations!
>
>     Stephen
>
>     On Tue, Jul 22, 2025 at 6:20 AM Deyan Ginev
>     <deyan.ginev@gmail.com> wrote:
>
>         Hi all,
>
>         Congratulations to Harry and Eri for landing this upgrade!
>         And best of success with your continuing MathML Core work.
>
>         Greetings,
>         Deyan
>
>
>         On Tue, Jul 22, 2025 at 5:45 AM Frédéric Wang
>         <fwang@igalia.com> wrote:
>
>             Here is the attachment.
>
>             Le 22/07/2025 à 11:39, Frédéric Wang a écrit :
>             > Hello,
>             >
>             > As some of you probably know, last year Harry Chen had
>             experimented
>             > RTL math support in Chromium as part Igali's Coding
>             Experience
>             > Program. This is just a heads up that I merged his spec
>             changes last
>             > week. My colleague Eri took over the work and landed a
>             patch in Chromium.
>             >
>             > You can find the explainer and Chromium's intent to
>             prototype here:
>             >
>             >
>             https://people.igalia.com/fwang/mathml-operator-mirroring-explainer.html
>             >
>             >
>             https://groups.google.com/a/chromium.org/g/blink-dev/c/iU0jQWJZCEM
>             >
>             > Support is available in the latest Chrome Canary if you
>             enable
>             > experimental web platform features from about:flags (see
>             attached
>             > screenshot).
>             >
>             > WebKit operator mirroring is known to be quite limited.
>             Firefox has
>             > some implementation, but we noticed it currently has
>             some issues. So
>             > we will make progress on them before sending the intent
>             to ship to
>             > Chromium.
>             >
>             > Frédéric Wang
>             >
>             >
>