Skillmon

> Is this quicker? No, it isn't. In fact it's slower by roughly a factor 2 (this strongly depends on the given argument, it might be faster in strange cases,...

Feel free to drop it :) I could write a fast version by doing some caching, setting it up during command-definition etc., but that would turn out to get rather...

> Also, I think that `\expandafter\let\csname pgfk@/pgf/arrow keys/fill/.@def\endcsname\@undefined` is doing the same thing as the `\cs_undefine:c`; am I correct? Yes, that's correct.

@Rmano Here is a parser that parses three numbers separated by a single arbitrary non-digit token (or group). It assumes `YYYYMMdd` (optionally followed by another ``, `` doesn't have to...

@Rmano Your code looks good. The \begingroup\catcode\`\\@=11 \unravel{\endgroup...} is just a scoped `\makeatletter...\makeatother` and not necessary in package code (since your setup for your code level should already deal with...

I don't think this is about the key-name, but about parsing the cs which is the handle's argument. So: ``` \keys_define:nn { foo } { bar .cs_set:N = \__foo_bar:nn }...

@u-fischer and @muzimuzhi: No `a4paper` shouldn't be reported as an unknown option, since it's known to `article`. The effect of an `unknown` handler is to not add unknown options to...

@josephwright what is the current behaviour? (I didn't check on package/class options in quite some time, sorry)

Generally or only if an unknown handler is defined?

@josephwright I'll need some time (my TeX-time is drastically reduced if compared to a few years ago...), and I'll have to take a look into this code and re-understand my...