On 7/15/10 11:06 AM, Bruce Korb wrote: >>> No, you need to set LC_ALL=C, because LANG has the lowest priority. >> >> That's true enough, but Bruce said he already removed all the LC_ >> variables from his environment, so LANG should work. > > "It's too hard." :) > >> $ locale >> LANG= >> LC_CTYPE="POSIX" >> LC_NUMERIC="POSIX" >> LC_TIME="POSIX" >> LC_COLLATE="POSIX" >> LC_MONETARY="POSIX" >> LC_MESSAGES="POSIX" >> LC_PAPER="POSIX" >> LC_NAME="POSIX" >> LC_ADDRESS="POSIX" >> LC_TELEPHONE="POSIX" >> LC_MEASUREMENT="POSIX" >> LC_IDENTIFICATION="POSIX" >> LC_ALL= > > So, for future reference, put this in my bashrc? > > eval $(locale | sed 's/=.*/=C/;s/^/export /') > > and remove the unset stuff? Of course, this is a GNU C Lib thingey, > so it only works on GNU C Lib based systems.....
Aren't the C and POSIX locales the same on glibc systems? Even if they're not, you should just have to change only LC_COLLATE and possibly LC_CTYPE anyway. > By the way, Chet, once upon a time long ago and far away, I > patched BASH to support BASH_TRACEFD because where I > worked there were 10's of thousands of lines of BASH scripting > that presumed that output to stderr meant command failure. > Well, I've moved on and lo and behold there's more scripting > that makes that assumption. This time, merely thousands of lines. > I'll re-do my old patch if you're ready to deal with it. :) ? bash-4.1 implements BASH_XTRACEFD. Chet -- ``The lyf so short, the craft so long to lerne.'' - Chaucer ``Ars longa, vita brevis'' - Hippocrates Chet Ramey, ITS, CWRU c...@case.edu http://cnswww.cns.cwru.edu/~chet/