X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f
Date: Sun, 6 Jan 2002 15:56:08 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Tim Van Holder <tim DOT van DOT holder AT pandora DOT be>
cc: djgpp-workers AT delorie DOT com
Subject: Re: bash's 'test' is inconsistent on XP (causing autoconf testsuite failures)
In-Reply-To: <000001c196b2$00b6f560$d07c76d5@zastaixp>
Message-ID: <Pine.SUN.3.91.1020106155059.2949A-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Sun, 6 Jan 2002, Tim Van Holder wrote:

> Now configure knows that some shells can't source
> /dev/null (or any other special file), and so it
> includes a test:
> 
>   if test -f $cache_file; then
>     ... source the cache file
>   fi
> 
> Unfortunately, our bash (2.05) returns 'true' for
> 'test -f /dev/null'; so it claims that /dev/null is a
> regular file and then complains it isn't.  So either
> 'test' or '.' needs fixing.

"test -f /dev/null" does TRT for me on Windows 98SE and on NT4.  So I'm 
guessing that what you see is due to a broken support for DOS devices on 
W2K and XP.  Try setting LFN=n and see if "test -f" works then.

IIRC, Charles found out that the LFN version of findfirst doesn't set the 
6th bit in the attributes for devices, and the LFN version of _truename 
doesn't return devices with a forward slash (as in "C:/NUL"), so both of 
the methods used by lstat to detect devices don't work.  We must find 
some way to work around this.