Move from LaTeXML's particularities to more standard HTML/CSS source

[fred-wang]

Initially, the MathML in HTML5 implementation note was generated from LaTeXML. If we are now directly storing XML/HTML page, I think we don't really need to bother with all LaTeXML's special features (e.g. to emulate LaTeX) like special CSS stylesheets and span with custom ltx classes, rel navigation,

and , etc Actually I'm not familiar with this, but taking one step further I think we want to align HTML/CSS source with other W3C specs (MathML 4, CSS, HTML 5 etc).

@davidcarlisle Do you think you would be able to do that?

[davidcarlisle]

@fred-wang yes the current sketch was just a more or less exact copy of your latexml output as it was a quick sketch just to restyle in w3c format and merge in the schema, and done without asking you but "borrowing" your words so I kept changes to a minimum.

Of course many newer w3c specs use an html/javascript build toolchain eg respec but many are still xslt based (notably mathml/xml/xslt/xquery etc. I think I'd rather keep MathML4 as an XML/xslt based build (although the group should decide that) and if we do It might be best to keep this spec as an xslt based build to make it easier to copy or cross reference sections between the two.

so initially, if you are happy to drop the latex authoring, at least I'd suggest that we clean up the html source of this spec removing any latex remnants but keep the xml "source" being basically html rather than moving it to the xmlspec markup in mathml (the mathml sources would be easier for others to edit if I moved them to use an html based xml markup, which could be considered)

I'll make a first pass at cleanup...

[fred-wang]

@fred-wang yes the current sketch was just a more or less exact copy of your latexml output as it was a quick sketch just to restyle in w3c format and merge in the schema, and done without asking you but "borrowing" your words so I kept changes to a minimum.

Yes, I think we should clean up things and have a final format decided before starting to do more edits.

Of course many newer w3c specs use an html/javascript build toolchain eg respec but many are still xslt based (notably mathml/xml/xslt/xquery etc. I think I'd rather keep MathML4 as an XML/xslt based build (although the group should decide that) and if we do It might be best to keep this spec as an xslt based build to make it easier to copy or cross reference sections between the two.

I don't really have preference but I guess it makes sense to be consistent with what we do for MathML 4.

so initially, if you are happy to drop the latex authoring, at least I'd suggest that we clean up the html source of this spec removing any latex remnants but keep the xml "source" being basically html rather than moving it to the xmlspec markup in mathml (the mathml sources would be easier for others to edit if I moved them to use an html based xml markup, which could be considered)

I'm not sure I understand all of this, but yes basically using HTML-based source sounds good to me.

I'll make a first pass at cleanup...

Thank you

[fred-wang]

Some quick comments:

• I'm not really attached to the "data:text/plain;base64" tab for code snippets, I think using "pre" like in the MathML 4 spec is probably better.
• I'd also like that we decide a convention/format for the name or id of headers, sections etc so that we have more or less stable links we can share (e.g. for spec reference in WPT tests).
• I'm also happy if the bibliography uses the same style/format as https://mathml-refresh.github.io/mathml/appendixh.html

[davidcarlisle]

Yes agreed, I just made a first pass at cleaning out the latex artefacts, I need to somewhat drastically clean up the sources and document the conventions before serious editing starts however even now I think it's usable to start sketching out the basic structure of the document and what it needs to say. I may experiment with moving the mathml sources towards a similar html rather than xmlspec based xml markup but I'll do that in small steps, testing things don't break.. but that would help keep the two documents looking coherent.

…

On Sat, 9 Feb 2019 at 20:36, Frédéric Wang ***@***.***> wrote: Some quick comments: - I'm not really attached to the "data:text/plain;base64" tab for code snippets, I think using " " like in the MathML 4 spec is probably better. - I'd also like that we decide a convention/format for the name or id of headers, sections etc so that we have more or less stable links we can share (e.g. for spec reference in WPT tests). - I'm also happy if the bibliography uses the same style/format as https://mathml-refresh.github.io/mathml/appendixh.html — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/mathml-refresh/mathml-web-profile/issues/3#issuecomment-462077295>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABNcApCD_3fhpyi4aSfaOGmSnBGgjx6oks5vLzFWgaJpZM4ayg-5> .

[fred-wang]

@davidcarlisle For the record, the original source are located at https://gitlab.com/mathml/MathMLinHTML5/tree/master/source

Do you think you could extract the Tikz figures into independent TeX files and automate generation of images? I think it should be possible to convert the PDF output to SVG using e.g. inkscape.

[davidcarlisle]

I'll see what I can do (there is also some experimental direct output to svg from tikz that I could try) Probably not before the weekend

[fred-wang]

@davidcarlisle Thanks. I think after my last commit, SVG formulas are the only places where MathML is still used.

[davidcarlisle]

@Frédéric WANG <fred.wang@free.fr> actually https://gitlab.com/mathml/MathMLinHTML5/tree/master/source is 404 I can see issues in that repo but no code, is it private?

[fred-wang]

@davidcarlisle can you sign in to GitLab with you github account?

[davidcarlisle]

I'm signed in to gitlab, but see nothing I'm https://gitlab.com/davidcarlisle in gitlab.

…

On Thu, 21 Feb 2019 at 22:54, Frédéric Wang ***@***.***> wrote: @davidcarlisle can you sign in to GitLab with you github account? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[fred-wang]

@davidcarlisle can you try again?

[davidcarlisle]

@Frédéric WANG <fred.wang@free.fr> thanks, works now.

…

On Thu, 21 Feb 2019 at 22:57, Frédéric Wang ***@***.***> wrote: @davidcarlisle <https://github.com/davidcarlisle> can you try again? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/mathml-refresh/mathml-core/issues/3#issuecomment-466202442>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABNcAh8C6EG5ewPw4XfhPJaeNxf9tIRtks5vPyRZgaJpZM4ayg-5> .

[davidcarlisle]

tex files for all the svg are checked in to mathml-core/src/figures a script to convert them is checked in to the same place (not integrated to travis yet) but should run in any reasonably current text installation, just requires latex and dvisvgm. A test file including all teh svg is

https://mathml-refresh.github.io/mathml-core/src/figures/svgtest.html

Not yet integrated into the core spec, need to review the generated svg.

[davidcarlisle]

The core spec now uses the svg files generated by svg.

Currently these are generated via dvisvgm --font-format=woff so they have WOFF versions of the the TeX fonts embedded. This is a bit inefficient as each file includes the fonts, and also may not really be necessary, it may be sufficient to simply use the document default fonts for the purposes. They are Ok for now, but leaving this comment as a marker for future possible cleanup

[fred-wang]

@davidcarlisle Thanks. I think in some figures we use simple math (\frac, \sqrt). I guess in some cases we can just use plain text as in the rest of the spec. At least the square root does not render well for me on figure 19.

[davidcarlisle]

@fred-wang does it look OK to you in
https://mathml-refresh.github.io/mathml-core/src/figures/svgtest.html
? (it does for me)
that's the same svg but referenced by <img> rather than inlined into the html.
which means that there is an error in my code to do the inlining, or as I half suspected there are artefacts coming from inconsistent font subsetting.

Either way this should be fixable without needing to change the tex sources. I'll experiment....

[physikerwelt]

this is work in progress as discussed in the telco on 2019-02-25

[davidcarlisle]

css font class selectors like text.f0 were interfering when all svg inlined. Fixed I hope by

w3c/mathml-core@09b6d10

[fred-wang]

I believe we no longer have LaTeXML stuff after my recent commits.