Joseph Wright

@Soldyesis If you have v2 of `siunitx` then it's more tricky as I didn't used to have a programmers API: that's the obvious reason for `\siunitx_number_parse:nN` not being defined.

The model for `siunitx` to generate 'properly formed' (La)TeX math mode material as an intermediate to then typesetting either in math mode (directly) or in text mode (using appropriate substitutions)....

This setting is really intended for manually-set values, like `\qty[exponent-base = 2]{1e4}{\byte}` or similar.

This is not currently supported. Can you point to a use in the published (professionally-typeset) literature?

@ScientiaEtVeritas That's not a published example, sorry - author-typeset material doesn't really count, I need things typeset by professionals. (Authors ask for all sorts of stuff, doesn't mean editorial offices/typesetters...

Don't we just want to do case folding here? So collect codepoints in all cases, then fold if required whilst comparing?

I think the issue here was that `\bitset_gput_...` should add to the bitset globally, but we are dealing with the index ...

> > It's [`expl3-datatype-analysis.xlsx`](https://github.com/latex3/latex3/blob/main/l3kernel/expl3-datatype-analysis.xlsx). > > I suggest to first make it readable on GitHub and to text editors by converting xlsx to csv/tsv file(s). > > See [related GitHub...

> > It's [`expl3-datatype-analysis.xlsx`](https://github.com/latex3/latex3/blob/main/l3kernel/expl3-datatype-analysis.xlsx). > > not sure I dare to ask, but do I see any takers on my suggestion? I will take a look ...

For the team: see https://lists.dante.de/mailman/private/latex-team/2020-December/032796.html for where `addto` came from.