This is the mail archive of the
libc-hacker@cygnus.com
mailing list for the glibc project.
Re: mysterious wordexp failure
- To: drepper@cygnus.com (Ulrich Drepper)
- Subject: Re: mysterious wordexp failure
- From: Zack Weinberg <zack@rabi.columbia.edu>
- Date: Fri, 02 Apr 1999 13:14:28 -0500
- cc: libc-hacker@cygnus.com
On 02 Apr 1999 09:43:14 -0800, Ulrich Drepper wrote:
>Zack Weinberg <zack@rabi.columbia.edu> writes:
>
>> It's the current egcs CVS tree. Should I try with 1.1.2?
>
>Well, sure you can. But I updated egcs yesterday here in the office
>and it worked. At home I use a different compiler and also no problem.
I've now tracked it to the behavior of the shell on my system. I said
"it's a bug" a few minutes ago but now I'm not sure.
$ var=:abc: IFS=: strace -ewrite /bin/ash -c 'echo $var' >/dev/null
write(1, " abc \n", 6) = 6
$ var=:abc: IFS=: strace -ewrite /bin/bash -c 'echo $var' >/dev/null
write(1, ":abc:\n", 6) = 6
Bash doesn't convert IFS to space in this context, ash does. I don't
know the standard but I would not be surprised to find that both
behaviors were permitted, or that ash was right.
zw