Karsten Schoelzel, le Tue 09 May 2006 12:20:48 +0200, a écrit : > On Sat, Jan 22, 2005 at 10:02:47PM +0100, Samuel Thibault wrote: > > Thinking twice about it: > > > > Why wouldn't ln_langinfo(CODESET) be used ? It exactly returns what's > > needed: the current charset (instead of merely guessing from the LANG > > value...) > > > Although it is not very obvious nl_langinfo(CODESET) is used if > available and w3m selected ISO-8859-15 if [EMAIL PROTECTED] on my system. > > As long as LC_ALL, LC_CTYPE or LANG are non-null and not "" the > following happen: > main.c:429 wc_guess_locale_charset is called > libwc/charset.c:64 wc_local_to_ces is called
Agreed, althouth the call it at line 60 here. > libwc/charset.c:377 nl_langinfo(CODESET) is used if LANG != "C" There's no such call in the source code here (obtained from unstable distribution by using apt-get source w3m). > I suspect a configuration problem on your side, perhaps enabling > the option "System charset follows locale(LC_CTYPE)" > (-o follow_locale=1) might help. I have this flag set, and it does have effect (LANG=fr_FR.UTF-8 w3m properly sets charsets to UTF-8). Samuel
