Date:	Sun, 4 Oct 1998 12:42:13 +0300 (IDT)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender:	eliz AT is
To:	Olivier Perron <perron AT art DOT alcatel DOT fr>
cc:	Olivier Perron <olivier DOT perron AT art DOT alcatel DOT fr>, djgpp AT delorie DOT com
Subject:	Re: /dev/null problems with bash when rebuild from sources
In-Reply-To:	<Pine.GSO.4.05.9810020916001.26036-100000@rtbsci146s>
Message-ID:	<Pine.SUN.3.91.981004124150.1863P-100000@is>
MIME-Version:	1.0

On Fri, 2 Oct 1998, Olivier Perron wrote:

> > (The bug is in the function `_truename' that is called by `stat': the
> > version bundled with stock v2.01 didn't support /dev/foo names.)
> > 
> Yes, but I do *have* the patched libc with the _truename *you* gave me a
> couple of weeks ago when I had to rebuild bash !
> So, I'm completly lost !

You cannot be *completely* lost, since you still have a debugger to
see which library function fails here ;-).

Are you using the patched libc from Nate's site?  If not, get it and
try linking Bash with it.  I might be wrong about `_truename' being
the cause; this provblem was solved a long time ago, so I could
forget.

If the patched libc doesn't help, write a small test program that
calls `stat' on "/dev/null" and see whether it succeeds (it works for
me, but my libc.a is patched more than Nate's).  If it fails, step
into `stat' with a debugger and tell me what do you see.  For all I
know, you could have just discovered another bug.

- Raw text -