Add an AMScd extension

[fred-wang]

Just opening an issue for this, as we don't seem to have one yet.

dpvc/MathJax@master...AMScd

[dpvc]

I think this is ready to go, so I'm marking it "Ready for Review".

[fred-wang]

It looks like the packed autoload-all.js file is causing conflicts with the current master/develop branches.

[fred-wang]

I prepared some tests

LaTeXToMathML/AMScd/amscd-1.html
LaTeXToMathML/AMScd/amscd-2.html
LaTeXToMathML/AMScd/amscd-3.html
LaTeXToMathML/AMScd/arrows-1.html
LaTeXToMathML/AMScd/arrows-2.html
LaTeXToMathML/AMScd/min-1.html
LaTeXToMathML/AMScd/min-2.html

I didn't add references yet, because it seems that I get different output with the ShowSource or with the native MathML source. I'm not sure if that's a bug or not.

One issue I noticed is that one can't do something like @<a<< because it will be badly parsed by the HTML parser. Hence people have to write @<{a}<<. I guess we can't fix that, but that's something we should remember to document.

Another thing is about the horizontal stretchy "=", that I think we also use for the chemistry extension. The MathML operator dictionary does not specify the equal sign as stretchy and fonts like e.g. STIX do not seem to have horizontal construction for it. Gecko does not support this and I don't know about MathPlayer. I don't remember when we chose to use the horizontal equal sign but we should probably coordinate with other projects and the Math WG. Perhaps Unicode has a "double horizontal bar" for that purpose that should be included in the math operator dictionary with stretchy=true by default.

[dpvc]

It looks like the packed autoload-all.js file is causing conflicts with the current master/develop branches.

Well, the AMScd branch is very old (pre v2.1) so that is not a surprise. There are two ways we could handle that. First, I could just make a new AMScd branch from develop so that it will merge cleanly now (without the packed versions). Or, we could do the merge, get the conflict, and then perform

git add extensions/TeX/autoload-all.js
git checkout extensions/TeX/autoload-all.js

to resolve the conflict, but get back the original version of the packed autoload-all.js file. Then commit the result.

Either is fine. Since there is really no history in the AMScd branch, we don't really lose anything with a new branch (other than the fact that the work was done earlier). What do you think?

[fred-wang]

I'm fine if you create a new branch (e.g. issue420) with only the changes to the two unpacked files. The packed files with basically a single line with all the code are more likely to get merge conflicts, so I'm happy if we can avoid them during the development.

[dpvc]

One issue I noticed is that one can't do something like @<a<< because it will be badly parsed by the HTML parser. Hence people have to write @<{a}<<

There are two other possible approaches. ONe is to use CDATA, as in

\[<!--[CDATA[
\begin{CD} 
A @>a>b> B\\ 
@VlVrV @AlArA\\ 
C @<a<b<D 
\end{CD}
]]-->\]

or to use spaces, as in

\[\begin{CD} 
A @>a>b> B\\ 
@VlVrV @AlArA\\ 
C @< a < b < D 
\end{CD}\]

(or put the code in one of MathJax's <script type="math/tex;mode=display">, or use <, etc.). If people are using content-management systems, they may handle the conversion to `<`` for them.

[dpvc]

Another thing is about the horizontal stretchy "=" ...

I don't see another alternative in the Unicode characters for this. Of course, stretching the equal sign is easy, since you just have to repeat it (it is its own extender character). I don't know if MathPayer supports it either. I'd be happy if you wanted to coordinate with other projects on supporting the stretchy equals.

[dpvc]

The packed files ... are more likely to get merge conflicts, so I'm happy if we can avoid them during the development.

I've stopped creating them, as per your previous request. This commit was created long before that, however, which is why they are there. I'll make the new branch.

[dpvc]

OK, I made the new issue420 branch.

[fred-wang]

Thanks. I'll write a mail to the Math WG. BTW, I see for example that LaTeXML implements AMScd the same way (using tables and stretchy equal signs):

http://www.maths-informatique-jeux.com/maths/memoire_groupes_quantiques/Ch1.S5.html

[dpvc]

I see for example that LaTeXML implements AMScd the same way

I really don't see any other way to do it, but thanks for the further information.

Should I merge the branch into develop at this point?

[fred-wang]

Should I merge the branch into develop at this point?

There's one thing I didn't have time to check:

I get different output with the ShowSource or with the native MathML source. I'm not sure if that's a bug or not.

Essentially, the attributes that are only added in the CD_env

arraydef: {
columnalign: "center",
columnspacing: CONFIG.colspace,
rowspacing: CONFIG.rowspace,
displaystyle: true

[fred-wang]

(Sorry, I unintentionally pressed enter). I wanted to say that columnalign and displaystyle do not show up in the ShowMath menu but they are in the MathML DOM.