Phelype Oleinik

Something like this should fix the problem: ```latex \ExplSyntaxOn \makeatletter \cs_gset_protected:Npn \__mathcolor_scan_for_scripts:w { \group_align_safe_begin: %

Indeed, I missed a couple `\group_align_safe_end:`. Only one `begin` is needed, but four `end` to cover all four “exits” of `\mathcolor`, when: 1. followed by a “normal” token (the one...

Although this should probably be fixed, it is not of much importance, I think. Those are internal, so the only issue would be if someone did `\scan_new:N \s__cs_mark` (and even...

In short, the issue happens (mostly) because (the equivalent of) `\cs_to_str:N` on the empty csname (`\csname\endcsname`) yields `csname\endcsname` instead of an empty string. Then, testing `\@ifundefined{csname\endcsname}` tests the existence of...

@muzimuzhi It would be nice, but I don't think it can do that in a reasonably efficient way, because the only ways I know to do that expandably are a...

(I added the minimal example for you. Since it requires a file tree, just specify how it should be, and add the necessary files preferrably using `filecontents`, as it's easier...

I will fix this as part of #439 because it requires keeping track of the argument to `\usepackage`, and it requires setting/resetting when a package begins/ends, otherwise a package `foo.sty`...

#439 is a tough one, and I didn't have the time to tackle for the 2021-06-01 release, so instead of a partial fix now I will leave this for the...

One option is to place docstrip guards close to the definition of the actual commands, then it is much easier to keep track of what's missing, and a programmer-friendly interface...

The easy half is done :) (I haven't pushed online yet) Also I've fixed the debug issue in the 2e side, so when this is ready it Should Not Break...