RE: Bug in uniq 2.0 / 5.0

[Brancato, Mike]

actually, I think it was a problem on my end with line returns.  All files
should have been unix (\n) but apparently one was opened under windows at
some point and saved (\r\n).  sorry for the confusion.

-----Original Message-----
From: Paul Eggert [mailto:address@hidden
Sent: Thursday, July 17, 2003 6:16 PM
To: Brancato, Mike
Cc: address@hidden
Subject: Re: Bug in uniq 2.0 / 5.0


"Brancato, Mike" <address@hidden> writes:

> I tested this with the latest uniq with the same results.

I can't reproduce the problem, either with uniq (GNU textutils) 2.0 or
with uniq (coreutils) 5.0.1.  Here's how I tried to reproduce it:

  $ cat folderlist
  04-00-0002287
  $ cat allfolders
  04-00-0002287
  04-00-0002287
  $ cat folderlist allfolders | sort | uniq -u

The last command didn't output anything, which was what was expected.

Does the 'sort' command actually ensure that all instances of the
04-00-0002287 line are adjacent in its output?  That might explain the
problem; i.e., it might be a problem with 'sort', not with 'uniq'.

When you say "the latest uniq", are you talking about the latest
test version <ftp://alpha.gnu.org/gnu/coreutils/coreutils-5.0.1.tar.gz>?
Or is it some other version?

It could be a problem with your locale.  What is the output of the
"locale" command?  coreutils 5.0.1 "uniq" reports an error if it
detects that the locale is misconfigured; this doesn't suffice to
detect all locale bugs but it's better than nothing.  Perhaps you can
give that version a try.

Another possibility is to fix the locale misconfiguration by setting
LC_ALL="C" in your environment.

Perhaps we need your exact test data; can you make it available?