This morning, at 9:04 AM, I wrote: > Xindy is more flexible in dealing with accented letters. This is > unlikely to matter much for the LaTeX3 project, but it seems as if the > `sort-rule` tool used for that may well be useful if you wanted to > (e.g.) sort `\__pkg_func:` under P just after `\pkg_func:`, or group > variables together instead of listing them under C, L, & G. This leads in to another question about how to index function variants. Is it better to have all uses of (say) `\bool_if:NTF`, `\bool_if:NT`, `\bool_if:NF`, `\bool_if_p:N` indexed differently, or should they all be linked together? I don't know whether the indexing tools we have can deal with “synonyms”, but there’s an interesting use-case for them if they’re possible. Take a look at source3.pdf, and notice the entry for `\bool_if:N` (sic). There ain’t no such function, but the macrocode-scanning auto-indexing functionality of doc.sty found it anyway, in the line where the actually-existing variants are defined: \prg_new_conditional:Npnn \bool_if:N #1 { p , T , F , TF } I think the index entries should read something like this: \bool_if:N(TF), \bool_if_p:N . . . . . . . 4, 8, 15, 16, 23, 42, _2015_, /37/ \bool_if:n(TF), \bool_if_p:n . . . . . . . 5, 9, 16, 17, 24, 43, _2015_, /38/ (The (TF) should be formatted the way it is in the margins, italicized & red-underlined.) If the LaTeX3 developers think this (or something similar) is the right way to go, I’ll try to research how to make it happen. —Joel