[BlindMath] Please, more talk between blind math users and LaTeX developers

[Jonathan Fine]

Hi Neil

I'll respond and contribute to some of the technical questions you raised
in your open, honest and helpful message. I hope we can have a respectful
difference of opinion on these important matters. I start with this. As I
recall, in 1997 the LaTeX team said that it would develop LaTeX3 to provide
new input document syntax, that aligns with SGML/XML. We still don't have
that. If we did then math accessibility would be much easier, for both PDF
and HTML output.

You wrote:

1. Adding the tagging in a way that doesn't require authors to alter what
they wrote means the deep down guts of the program need to be modified.
2. This is not an easy task, especially when dealing with a large, old
program!
3. If it were easy, it would have been done by now.

I'd like to break the problem down into three main parts.

A. From a LaTeX file produce a PDF document.
B. From a LaTeX file produce a tagged document.
C. Merge the outputs of (A) and (B) into a single document

The whole problem would be much easier if we already had (B) as a starting
point. And (B) would be much easier if we had a new LaTeX input syntax. But
though development of this started in 1997, it seems to have now been
abandoned. I'd like to resume that work.

The LaTeX Team wishes to solve problems largely by writing TeX macros. They
wrote that about 25% of the project would be preliminary work needed to
modernize the document processing functionality of the LATEX kernel. And
then they can start programming.

By document processing they mean text processing rather than typesetting
and PDF generation. Things that languages and systems such as Pandoc,
Python and XSLT already do very well. However hard the task is when using
the best available tools, it becomes much easier if you impose some
reasonable restrictions on the input, and much harder when TeX macros is
the only programming language you use.

Their feasibility study doesn't even consider the idea of using
pre-processing the input to LaTeX, and post-processing the output. The
LaTeX Team love LaTeX and writing TeX macros. I'm grateful to them for the
positive contributions they've made in this area. But I do not think it
wise for us to promote their project without first considering the use of
more suitable tools and existing technologies for the document processing
functionality.

I would also consider alternative means of achieving accessibility, such as
providing just (B), namely producing just a tagged document from LaTeX. The
PreTeX system shows how much can be achieved once one has a suitable tagged
document. I think a good authoring environment for tagged documents is
vital. It could be a restricted subset of LaTeX. Or a markdown variant.

Finally I think that whatever the technology a tagged PDF enhancement will
work only for a subset of present-day valid LaTeX files, except with
massive (and I would say Quixotic effort). It would be best to start with a
safe subset of LaTeX files. Success here will encourage authors to adjust
their way to cooperate with the new system. And provide insight for dealing
with a growing larger subset.

Here's the PreTeXt URL: https://pretextbook.org/. Perhap a LaTeX-like to
PreTeXt XML could be made quite easily.

with best regards

Jonathan