Thanks Florian. I used -b instead of -c with en_US.utf8 and that works as
desired as well (on coreutils 8.4). Just out of curiosity, I compiled
coreutils 8.12 and the cut command works as desired using -c and with no
changes to LANG (i.e. at my default of en_US.utf8).

Chetan

On Tue, Jun 14, 2011 at 3:51 PM, Florian Philipp <[email protected]>wrote:

> Am 14.06.2011 22:35, schrieb Valery Mitsyn:
> > On Tue, 14 Jun 2011, Chetan Tiwari wrote:
> >
> >> Greetings,
> >>
> >> I was having trouble using the cut command and wasn't sure if anyone
> >> else on
> >> this list faced a similar problem. Perhaps I am doing something wrong
> and
> >> any help will be much appreciated.
> >>
> >> I am using SL6 and the version of cut installed is "cut (GNU coreutils)
> >> 8.4". I have a text file with the following line:
> >>
> >> abcdefghijklmnopqrstuvwxyz
> >>
> >> When I execute the following statement: "cut -c1-3,24- test.txt", I
> >> get the
> >> correct output: abcxyz. When I run the same command, but with a
> >> --output-delimiter flag (i.e. cut -c1-3,24- test.txt
> >> --output-delimiter="|"), I get the exact same output as before:
> >> abcxyz. The
> >> delimiter (I - pipe symbol) is missing.
> >
> >  What is LANG/LC_ALL enveronment in your shell?
> > If it is something with .UTF-8 try to set it without suffix.
> > LANG=C cut --output-delimiter="|" -c 1-3,24- test.txt
> > should do the trick.
> >
>
> Try to use -b (bytes) instead of -c (characters). It works for me. The
> info page of cut states that it is equal to -c at the moment but that
> info is probably outdated.
>
> While LANG=C works in my tests, oddly enough the results with
> LANG=de_DE.utf8 do not depend on the number of bytes in the delimiter
> character or the input.
>
> >>
> >> When I run the exact same commands on a machine running Ubuntu with "cut
> >> (GNU coreutils) 7.4", I get the desired output: abc|xyz. The same
> problem
> >> occurs with larger data files that I was trying to manipulate as well.
> >>
>
> Version 8.7 in Gentoo also seems to be correct.
>
> Hope this helps,
> Florian Philipp
>
>

Reply via email to