On 28/09/10 21:03, René Berber wrote:
I'm not sure where is the standard, I don't use French but I have used
Spanish and the abbreviations are 3 letters and no dot, for month and day.
Fail2ban has some harcoded French month names, using unicode, in the
code you see things like 'TABLE["Feb"] = [u"Fév"]' which I think stands
for Février:, September doesn't have an entry since, using 3 letters, its
the same. Additional to that, it does locale conversion, from any
configured locally to C which is English using ASCII, perhaps this part
is failing, if you have the correct locale it should have worked (but
syslog being in English does point that the locale is not French, but
English, or C, or none).
Is there some docs on how fail2ban manage locales and why it seems to
recognize only english format ?
I think it is quite common to use 'exotic' (i.e. any other than POSIX, C
or en_*) locales only in user profiles in stead of system wide, so they
don't apply to 'system' processes.
Last time I checked, my glibc included 320+ locales, so imho maintaining
a parser supporting all (or most) of them would take more work than all
of current fail2ban development time available :)