Re: grep 2.5.1: NUL byte doesn't match a complemented character class

[Jim Meyering]

Joe Wells <address@hidden> wrote:
> Jim Meyering <address@hidden> writes:
>> On some systems, the locale name is spelled slightly differently:
>> [get the proper spelling from the output of "locale -a"]
>
> (By the way, this is irrelevant to the bug in grep, but I believe the
> output of “locale -a” does not give the officially correct locale
> names.  On my system, it says my locale name is “en_US.utf8”.  My
> understanding from reading the standards documents is that the
> officially correct name is “en_US.UTF-8”.  The use of “utf8” occurs
> because glibc has an internal compatibility hack where it downcases
> the charset name and removes hyphens from it before looking up the
> locale on disk and in data structures.)

Maybe you'd prefer s/the proper/a useful/?
I just want to be sure that the locale setting I use will be
recognized by the system at hand, and don't care if it's officially correct.
In pedantic mode, this might be more to your liking:

  if you want to be sure to use a spelling that is recognized on your
  system, one way is to choose from the list output by "locale -a".

>> RHEL5 has the bug [rpm -q grep -> grep-2.5.1-52.2]:
>>
>>   $ printf '\0x' | LC_CTYPE=en_US.UTF-8 grep '[^x]x'
>>   [Exit 1]
>>
>> Debian unstable seems not to have a problem:
>>
>>   $ printf '\0x' | LC_CTYPE=en_US.utf8 grep '[^x]x'
>>   Binary file (standard input) matches
>
> I'm glad you were able to reproduce the bug.  Can you tell if it is in
> grep or the locales or glibc?

I haven't investigated.

>> I've Cc'd address@hidden, since that's the preferred bug-reporting
>> address.
>
> Then I have another bug to report.  The man page for “grep” on my
> system (Ubuntu Dapper Drake) gives address@hidden as the only

That was updated upstream in Nov of 2004.
Of course, it was after the release of grep-2.5.1,
and we're still waiting for 2.5.2.

> bug reporting address.  (And there is no “grep.info” file installed.
> Is there such a file?)  Is this a Debian/Ubuntu bug or a problem in
> the original grep source?

There is most definitely a grep.info file.
If you wonder, check out the upstream site for grep:
  http://savannah.gnu.org/projects/grep

grep.info is generated from grep.texi, here:
  http://cvs.savannah.gnu.org/viewvc/grep/doc/grep.texi?root=grep&view=log

Maybe you haven't installed the documentation.
In any case, it's a Debian/Ubuntu-specific problem.

...
>>> By the way, I am using Ubuntu 6.06 LTS (“Dapper Drake”) with all
>>
>> I would consider upgrading.
>
> Of course.  (But “LTS” is for “long term support”.  One of its main
> advantages is not needing to upgrade.)

You shouldn't expect all of the latest fixes for small things
like this in an older *stable* release.

> Is your point that you think the RBL in question is unreliable and
> shouldn't be used?  (I know RBLs in general are often controversial,
> for exactly the reason we are seeing here.)

Of course.