Try out other tools for conversion to HTML

[thesamovar]

• https://wiki.lyx.org/Tools/ELyXer
• LaTeXML
• tth

[rorybyrne]

I tried out LaTeXML on a sample document here.

The nodes could be turned into some abstract representation of a paper, something friendlier than XML. We can probably ignore the fine-grained <XMATH> nodes, and instead pull the raw maths LaTeX from the <Math> node and put it through KaTeX for rendering.

[rorybyrne]

Now that I think about it, having access to the fine-grained maths nodes might also be useful - imagine a special "maths" sidebar that helps the reader to piece the equations together, highlighting the provenance of variables across multiple equations etc..

[rorybyrne]

Taking that a step further, it would be great if the "maths" sidebar could be built by someone else. It takes our intermediary paper-representation as input, and renders some HTML.

I'm not 100% sure if/how that would be achieved though, without using some kind of eval() function to evaluate raw and potentially malicious code.

[thesamovar]

I love the idea of using the fine grained access to math nodes. For example, you could imagine when you hover over a symbol it shows you the definition. In documents built for this system you could explicitly code that, and for legacy documents you could show the first time it's used.

[rorybyrne]

In documents built for this system you could explicitly code that

If I understand correctly, this could take the form of a LaTeX package which defines specific annotations/macros for authors to use?

[thesamovar]

Yes, although I'm not convinced that LaTeX should be the ultimate goal. Too much irrelevant cruft that makes life difficult for translating to web format.

[rorybyrne]

Okay. Changing the source format shouldn't be a hassle (touch wood), and we could in theory support many source formats. Designing the intermediary format of a paper (if we want to go down that route) will be key.

You know better than I do what should be in the intermediary format, so if you have time to think about what it might look like (e.g. as a JSON file), I can start working on parsing the LaTeXML output into it.