[EMAIL PROTECTED] (Christian Weisgerber) writes: > I don't follow. > From what I understand you set LC_CTYPE=en_US.UTF-8 on your _OpenBSD_ > box, which is invalid, and then PERL_BADLANG to shut up perl. Not > setting LC_CTYPE in the first place looks like a more straightforward > approach.
Sorry, I guess I didn't explain very well at all. "export LC_CTYPE=en_US.UTF-8" does do something for both emacs and xterm. Without it both emacs and xterm have trouble reading make.log's generated by gcc on fedora. With that LC_CYTPE setting both programs correctly display the gcc warnings. Let me see if I can quote a few lines: gmfgetfeature.c: In function ‘gmf_get_feature_name0’: gmfgetfeature.c:222: warning: format ‘%qx’ expects type ‘long long unsigned int’, but argument 5 has type ‘gf’ gmfgetfeature.c:222: warning: format ‘%qx’ expects type ‘long long unsigned int’, but argument 6 has type ‘gf’ With the LC_CYPTE set to utf-8, the back-quote and forward-quote show up like `%qx' and `long long unsigned int` as such: gmfgetfeature.c: In function `gmf_get_feature_name0': gmfgetfeature.c:222: warning: format `%qx' expects type `long long unsigned int', but argument 5 has type `gf' gmfgetfeature.c:222: warning: format `%qx' expects type `long long unsigned int', but argument 6 has type `gf' Without the utf-8 LC_CTYPE, everything between the quotes and the quotes shows up as the single character 'a' with a ^ over it. It basically removes the most valuable part of all the error messages. gmfgetfeature.c: In function `gmf_get_feature_name0': gmfgetfeature.c:222: warning: format a^ expects type a^, but argument 5 has type a^ gmfgetfeature.c:222: warning: format a^ expects type a^, but argument 6 has type a^ So that LC_CTYPE does do something important for the programs I care most about. -wolfgang
