On Fri, Aug 14, 2026 at 10:11:46AM +0200, Patrice Dumas wrote: > Hello, > > I propose to generate Info encoded in UTF-8 only from texi2any. This > has two advantages > * simplify the implementation in C (in Perl still need to encode/decode) > * non-ASCII cross-references between manuals are correct in Info reader > > I do not see any disadvantage.
I would be happy with this change. In my experience working on character encoding issues can be very time-consuming. There's a clear reason for the change (simplify the C implementation). UTF-8 Info files are likely to work anywhere that it matters - the stand-alone Info reader even does some degree of recoding in non-UTF-8 locales. > > For the plaintext output, it could be simpler if we did the same, but > encoding the output would probably be easy if we make the change for > Info first. > > -- > Pat >
