Skillmon

Is this really a bug? Should we now start protecting against alignment tokens whenever we read an `N`-type argument anywhere for all functions? When you know you want to assign...

Point taken, should read MWE code more thoroughly. Proposed fix: ```latex \documentclass{article} \ExplSyntaxOn \group_begin: \char_set_catcode_active:N \^^@ \use:n { \group_end: \cs_set_protected:Npn \char_set_active_eq:nN #1 { \group_align_safe_begin: \group_begin: \char_set_lccode:nn { `\^^@ } {#1}...

Hm, still doesn't work for the `\group_align_safe_begin: \char_set_active_eq:NN \@ \prg_do_nothing: \group_align_safe_end:` test case :(

This is the definition of `\ekv@expandafter` that I use in my `expkv-bundle` (hinting at this being very useful). It works in pdfTeX, LuaTeX, XeTeX, and (u)pTeX. Though I'd say the...

`\exp_after:wN` isn't correct either. It simply has no effect on (most?) non-`N`-type material, but "have to" looks too strict. That's why I'd argue it should've never been `:NN` in the...

> So, as efficiency in that case wasn't a concern (pgfplots with gnuplot is the slow part) I opted to easy my life tracking the variables set by each key...

@alceu-frigeri maybe I misunderstood what you meant with a "default/clear" clear key list. I was only referencing that aspect (hence the citation in my original comment), not your overall use...