Il 17/03/2014 15:49, Norihiro Tanaka ha scritto:
Package: grep
Tags: patch
I ran following test, which used the regex enging in non-UTF8 locale.
$ yes abcd.abc | head -10000 > m
$ time -p env LC_ALL=ja_JP.eucJP src/grep abcd.abd m
real 7.28
user 6.36
sys 0.57
It's extremally slow. When regex engine is used in grep, a text is
splitted by line. However all of buffer is passed to re_search and
re_match. I seem that it's wrong.
Yes, very good catch.
It's likely that the old bytecode matcher didn't care, but the new one
in glibc has to process even the "ignored" part of the buffer to find
the boundaries of multibyte characters.
Paolo
