Date: Sun, 10 Jan 1999 13:33:18 +0200 (IST) From: Eli Zaretskii X-Sender: eliz AT is To: "Mark E." cc: djgpp-workers AT delorie DOT com Subject: Re: shell utils 1.16 question In-Reply-To: <199901082041.UAA153364@out5.ibm.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Reply-To: djgpp-workers AT delorie DOT com On Fri, 8 Jan 1999, Mark E. wrote: > > Let's not break things that don't have to be broken. > > It seems to me when we have to recreate configure scripts that > something is already broken. Indeed. But the broken thing is the configure script itself, not the way `test' works in the DJGPP port. So you suggest to fix the wrong part, IMHO. The right way to fix this is to change Autoconf so that the generated scripts automatically work correctly on DOS and Windows. Since "test -x" cannot be used unconditionally (some old shells don't support it), one way that works is this: if test -n "$COMSPEC$ComSpec"; then ac_x=-x; else ac_x=-f; fi (the ComSpec part is for NT) and subsequently use "test $ac_x" instead of "test -f". What Robert suggests is also okay, provided that the check for "test -x" is done when the script (as opposed to Autoconf) actually runs. Since Autoconf was not actively maintained during last several years, I succeeded to convince several maintainers of GNU packages to make such changes in their local Autoconf files. Now, that Autoconf is maintained again, I suggest we join forces to try and push this (and other) DJGPP-related changes into the official Autoconf release.