> Hmmm... I revisited some old documents that use \n[charxxx] requests
^^^^^^^^^^^
You mean \[charxxx], don't you?
> and they are broken... Now I can't access glyphs by number[1] unless
> I use \N'XXX'. Is this intentional?
Yes. \[charXXX] is equivalent to the input character with decimal
code XXX (a misuse of the \[...] escape). They are no longer
hard-coded in the font files so that e.g. latin-2 can be supported as
an input encoding (except something in this direction as soon as I've
finished the tedious work of checking all groff glyph names).
To access glyph indices directly, you should *only* use \N'...'.