🚀 go-pugleaf

On Thursday 03 May 2012 21:25, in comp.unix.shell, kaz@kylheku.com wrote:

> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=655293
> 
> I ran into this doing a simple grep job that needed to match upper
> case characters, and so I started Googling.  This was only reported in
> January.
> 
> But the Red Hat people knew about what looks like the same bug two years
> ago. Oops, they didn't share!
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=583011
> 
> (So much for the spirit of collaboration in open source. My distro, my
> patches, screw you!)
> 
> Watch this:
> 
> $ echo a | grep '[A-B]'
> a
> $ echo b | grep '[A-B]'
> $ echo b | grep '[:upper:]'
> $ echo B | grep '[:upper:]'
> $ echo E | grep '[:upper:]'
> $ echo e | grep '[:upper:]'
> e

Hmmm...  A couple of observations

First, IIRC, the grep character classes (such as [:upper:]) syntatically
substitute for the "list of characters" that are enclosed by the square
brackets.

Consequently, the alternate form of '[A-B]' is not '[:upper:]', but instead
is '[[:upper:]]'. That is, the '[:upper:]' is enclosed within a set of
square brackets, just like 'A-B' is.

Thus, your examples that use
  grep '[:upper:]'
should only match the characters ':', 'u', 'p', 'e', or 'r', something that
your final example /does/ show.

Second; I guess that your abberent grep behaviour wrt 'a' is version
dependant. Under GNU grep 2.5.3 (32bit Slackware Linux 12.2), I don't see
the same results. In fact, I see the results you'd properly expect from
grep.

  ~ $ grep -V
  GNU grep 2.5.3

  Copyright (C) 1988, 1992-2002, 2004, 2005  Free Software Foundation, Inc.
  This is free software; see the source for copying conditions.  There is NO
  warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR
  PURPOSE.

  ~ $ for element in a b A B e ;
  > do
  >   echo "Using $element into grep [A-B]"
  >   echo $element | grep '[A-B]'
  >   echo "Using $element into grep [:upper:]"
  >   echo $element | grep '[:upper:]'
  >   echo "Using $element into grep [[:upper:]]"
  >   echo $element | grep '[[:upper:]]'
  > done  
  Using a into grep [A-B]
  Using a into grep [:upper:]
  Using a into grep [[:upper:]]
  Using b into grep [A-B]
  Using b into grep [:upper:]
  Using b into grep [[:upper:]]
  Using A into grep [A-B]
  A
  Using A into grep [:upper:]
  Using A into grep [[:upper:]]
  A
  Using B into grep [A-B]
  B
  Using B into grep [:upper:]
  Using B into grep [[:upper:]]
  B
  Using e into grep [A-B]
  Using e into grep [:upper:]
  e
  Using e into grep [[:upper:]]

HTH
-- 
Lew Pitcher

Lew Pitcher <lpitcher@teksavvy.com> writes:
> kaz@kylheku.com wrote:

>> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=655293
>> 
>> I ran into this doing a simple grep job that needed to match upper
>> case characters, and so I started Googling.  This was only reported in
>> January.
>> 
>> But the Red Hat people knew about what looks like the same bug two years
>> ago. Oops, they didn't share!
>> 
>> https://bugzilla.redhat.com/show_bug.cgi?id=583011
>> 
>> (So much for the spirit of collaboration in open source. My distro, my
>> patches, screw you!)

The fix linked from the RH bug is to upstream grep, so it's not clear in
what sense they "didn't share".
 
> Second; I guess that your abberent grep behaviour wrt 'a' is version
> dependant. Under GNU grep 2.5.3 (32bit Slackware Linux 12.2), I don't see
> the same results. In fact, I see the results you'd properly expect from
> grep.

It looks like it was dependent on both version and locale, as well as on
competence with regexp syntax.

-- 
http://www.greenend.org.uk/rjk/

Kaz Kylheku <kaz@kylheku.com> writes:

> Watch this:
> 
> $ echo a | grep '[A-B]'
> a

I can't reproduce this on Debian.  Neither in lenny nor in squeeze,
not even in etch.  (C and de_DE.UTF-8 locales tested)

What is your 'Debian'?

> $ echo b | grep '[A-B]'
> $ echo b | grep '[:upper:]'
> $ echo B | grep '[:upper:]'
> $ echo E | grep '[:upper:]'
> $ echo e | grep '[:upper:]'
> e

I can reproduce these -- are all expected, as explained in followup.

Regards,
Wolfram.

On 2012-05-04, Wolfram Gloger <wmglo@dent.med.uni-muenchen.de> wrote:
> Kaz Kylheku <kaz@kylheku.com> writes:
>
>> Watch this:
>> 
>> $ echo a | grep '[A-B]'
>> a
>
> I can't reproduce this on Debian.  Neither in lenny nor in squeeze,
> not even in etch.  (C and de_DE.UTF-8 locales tested)

Me neither; I was mistaken about that. Sorry!