RE: sub-problem with regex

[Alexander Stohr]

the suggested change fixed the problem.

here are a few words on my leraning curve:

as seen in manpage the LC_ALL=C subdivides
into several other LC_* settings.
I have to assume that its the LC_CTYPE
submember that is the thing in request.

hmm, on a RH7.3 system where its working:
  LANG=en_US.iso885915
  LC_ALL=<unset>

on the RH beta system where it fails
  LANG=en_US.UTF-8
  LC_ALL=<unset>

so it was the UTF-8 which does break my awak code.
just curios that [[:upper:]] was different from [A-Z].
(and it broke midnight commander grafical chars
on all telnet consoles too!)

on the other side i tried setting IGNORECASE and
saw effects on results on first machine but could
not affect the problematic behaviour on the second.

so i would suggest adding a string hint to both
settings vice versa because it seems the LANG
setting does prevent the other to change behaviour.
two possible causes => two things to look after
=> two or more possible solutions.

Note:
as only beeing a minor part time awk i have had
no success in dumping the script variables into
a file despite doing pretty nice cut and paste.

anyways, the thing is working now.

> -----Original Message-----
> From: Aharon Robbins [mailto:address@hidden]
> Sent: Friday, September 27, 2002 09:56
> To: address@hidden; address@hidden
> Subject: Re: sub-problem with regex
>
>
> This is a problem with the locale you're using.  If you set
>
>       export LANG=C
>
> or
>
>       export LC_ALL=C
>
> before running gawk, it'll work the way you expect.  It's not
> a gawk bug.
>
> Arnold