Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> writes: > I may be missing something, but this is just the "Accept-Language" part > of this change, i.e. there is no "round-tripping" here of actually doing > the work on the server of doing setlocale(), no? I think the wish is that if Accept-Language is interpreted by the HTTP(s) server and it gets turned into LC_something=<locale> when the HTTP server spawns us, we'd know what language to issue our error messages.