grep very slow when LANG is set to en_US.UTF-8

[Dan Nicolaescu]

This is on Fedora Core 1 on x86. 
grep-2.5 and grep from CVS

A simple way to reproduce the problem:

cd grep/src
gcc -c -dr dfa.c

now an about 1MB file called dfa.c.00.rtl should exist
(or .01.rtl if using a newer gcc)


do: 
setenv LANG en_US.UTF-8
(this is the way my system is setup by default)

grep   -c "^(insn" dfa.c.00.rtl

This takes forever... 

But: time env LANG=C grep -c "^(insn" dfa.c.00.rtl       3921
0.000u 0.020s 0:00.00 0.0%      0+0k 0+0io 205pf+0w

The LANG setting should not have an influence on the "^(insn" regexp,
yet it makes grep extremely slow. 

I hope this helps.

        --Dan