On 2026-06-01 02:40, Lasse Collin wrote:
Bruno's tryB.diff used LC_CTYPE, not LC_ALL, and thus didn't affect
number formatting.
Yes, I'm of two minds about this.
In general I'm suspicious about setting only LC_CTYPE. With UTF-8 it
should be safe, but with Shift-JIS etc. you can run into trouble.
However, for this particular case I don't offhand know of any trouble
that would occur in practice.
Conversely, using LC_ALL here (and changing numeric format) shouldn't
cause trouble either, as people don't parse that part of gzip output.
Also, if gzip ever gets around to supporting LC_MESSAGES, we really
should use LC_ALL; and in the meantime LC_ALL is better for Euro-English
locales like en_DK that use English-language messages but "," for
decimal point.
