Thu 21 Jun 2012 10:42:19 PM UTC, original submission:
Before 7a0ad00f634237b753572378289d76fa8f1c5942:
$ grep 'foo.*\<bar' /tmp/t
fooxxxxx_bar
$
7a0ad00f634237b753572378289d76fa8f1c5942 and after:
$ src/grep 'foo.*\<bar' /tmp/t
$
(/tmp/t contains only that one line.)
The grep that I built to bisect this links to Ubuntu 10.10's libpcre3, which apt-cache show libpcre3 shows as being:
Architecture: amd64
Source: pcre3
Version: 8.02-1
I'm not sure how relevant pcre is, as I'm not using -P.
I don't feel qualified to determine which of the behaviours is correct, but the reason I hunted down this change in behaviour is that the underscore not forming part of the "word" surprised me.
It turns out also that before 7a0ad00, the output depends on whether I set LC_ALL=C or LC_ALL=en_ZA.utf8; 7a0ad00 and after, the output is empty regardless of which of those two locales I set.
In any case, this change of behaviour seems unrelated to the commit message:
commit 7a0ad00f634237b753572378289d76fa8f1c5942
Author: Paolo Bonzini <bonzini@gnu.org>
Date: Mon Apr 19 14:50:23 2010 +0200
dfa: optimize UTF-8 period
* NEWS: Document improvement.
* src/dfa.c (struct dfa): Add utf8_anychar_classes.
(add_utf8_anychar): New.
(atom): Simplify if/else nesting. Call add_utf8_anychar for ANYCHAR
in UTF-8 locales.
(dfaoptimize): Abort on ANYCHAR.
Submitter: Bernd Jendrissek (not logged in)
|