In LaTeX should the align environment has an implicit {rlrl...} horizontal alignment. https://ctan.math.illinois.edu/obsolete/info/math/voss/mathmode/Mathmode.pdf (page 25)

We don't quite follow the latex spec as \begin{align} has should have equation numbers, our is more like \begin{align*} with no numbering.

First field are also center aligned rather than right aligned

The same problem occurs with \begin{cases}

f(n) =
\begin{cases}
 n/2, & \text{if }n\text{ is even} \\
 3n+1, & \text{if }n\text{ is odd}
\end{cases}

The conditions should be left alligned.

I had looked into the alignment problems before and it's relatively easy to adjust the output, cf. T350736 However, I don't fully understand what the desired MathML Layout is. One strategy is to copy the settings from the new arxiv HTML rendering, e.g, https://arxiv.org/html/2409.12006v1. Do you think it looks good there?

Also T352608 is related. I now updated https://www.w3.org/wiki/Math_Tools#Tools_for_MathML_Core to have a list of possible realizations in MathML.

For example http://fred-wang.github.io/TeXZilla/ generates for the cases example the follwing.

<mtable displaystyle="false" columnalign="left left">

We generate

<mtable columnspacing="1em" rowspacing="4pt">

I will start with generating the coumnalign attribute, maybe this solves most of the problems already.

Change #1074702 had a related patch set uploaded (by Physikerwelt; author: Physikerwelt):

[mediawiki/extensions/Math@master] Fix: Use align information in matrix constructs

https://gerrit.wikimedia.org/r/1074702

Change #1074702 merged by jenkins-bot:

[mediawiki/extensions/Math@master] Fix: Use align information in matrix constructs

https://gerrit.wikimedia.org/r/1074702

Hum, things seem to have gone back again.

Current rendering from https://en.wikipedia.beta.wmflabs.org/wiki/T375317

It would be good to check for more than 2 fields. Which should be aligned r-l-r-l etc.

https://www.latex-project.org/help/documentation/amsldoc.pdf page 7 has

\begin{align}
x&=y & X&=Y & a&=b+c\\
x’&=y’ & X’&=Y’ & a’&=b\\
x+x’&=y+y’ & X+X’&=Y+Y’ & a’b&=c’b
\end{align}

Rendered as

In T375317#10185633, @SalixAlba wrote:

Hum, things seem to have gone back again.

Current rendering from https://en.wikipedia.beta.wmflabs.org/wiki/T375317

I can not reproduce this.

In T375317#10185640, @SalixAlba wrote:

It would be good to check for more than 2 fields. Which should be aligned r-l-r-l etc.

For that example, it would be columnalign="right left right left right left right left right left right left" and also important columnspacing="0em 2em 0em 2em 0em 2em 0em 2em 0em 2em 0em" (taken from current svg rendering)

Also broken in Firefox with the first field, which should be right aligned.

<math forcemathmode="native">\begin{align}
 z & =  a \\
 f(x,y,z) & = x + y + z
\end{align}</math>

Rendered on beta as

Rather than SVG

I've added this to the end of https://en.wikipedia.beta.wmflabs.org/wiki/T375317

In T375317#10215500, @SalixAlba wrote:

Still broken with chrome and edge, windows. Current rendering of
https://en.wikipedia.beta.wmflabs.org/wiki/T375317

I think this is rather a new/independent bug.

In T375317#10215537, @SalixAlba wrote:

Also broken in Firefox with the first field, which should be right aligned.

<math forcemathmode="native">\begin{align}
 z & =  a \\
 f(x,y,z) & = x + y + z
\end{align}</math>

Rendered on beta as

Rather than SVG

I've added this to the end of https://en.wikipedia.beta.wmflabs.org/wiki/T375317

I'll fix this.

There is a dedicated test, that wants the standard alignment to be left (T348791)

Change #1079626 had a related patch set uploaded (by Physikerwelt; author: Physikerwelt):

[mediawiki/extensions/Math@master] Use the same formatting for align as MathJax

https://gerrit.wikimedia.org/r/1079626

new local rendering

Change #1079626 merged by jenkins-bot:

[mediawiki/extensions/Math@master] Use the same formatting for align as MathJax

https://gerrit.wikimedia.org/r/1079626

I believe it's related to https://issues.chromium.org/issues/40889720, in which Chrome doesn't support the columnalign attribute at all.

See also https://github.com/w3c/mathml/issues/181 and https://github.com/w3c/mathml-core/issues/268 (mathml-core)

Firefox:

@SalixAlba does that look good for you https://codepen.io/physikerwelt/pen/YPyQQgE it just uses CSS following suggestions in https://github.com/w3c/mathml-core/issues/268

That seems good.

I'd like to check the case of multiple fields which for \begin{align} should be aligned right-left-right-left etc.

@SalixAlba I added 5d

https://codepen.io/physikerwelt/pen/YPyQQgE

Looks good to me.

Change #1176552 had a related patch set uploaded (by Physikerwelt; author: Physikerwelt):

[mediawiki/extensions/Math@master] Use polyfill for columnalign

https://gerrit.wikimedia.org/r/1176552

Change #1176552 merged by jenkins-bot:

[mediawiki/extensions/Math@master] Use polyfill for columnalign

https://gerrit.wikimedia.org/r/1176552

It looks good to me now on Firefox and Chrome on beta. However, MathJax does not seem to support CSS, so this has no alignment now.

Im not sure if the alignment in https://de.wikipedia.beta.wmcloud.org/wiki/Intbar#Algin should also be fixed. I think it is not fixed.

I'm using Ubuntu 24.04 and Chrome.

In T375317#11091487, @Christian1985 wrote:

Im not sure if the alignment in https://de.wikipedia.beta.wmcloud.org/wiki/Intbar#Algin should also be fixed. I think it is not fixed.

I think that might be a cache issue. I modified the equation to force the system to recalculate the expressions and thats fixed it.

I tried doing a purge with https://de.wikipedia.beta.wmcloud.org/w/index.php?title=Intbar&action=purge&mathpurge=true that normally fixes things but not here for some reason.