Sender: rich AT phekda DOT freeserve DOT co DOT uk Message-ID: <3E1029E3.E679E723@phekda.freeserve.co.uk> Date: Mon, 30 Dec 2002 11:11:31 +0000 From: Richard Dawe X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.23 i586) X-Accept-Language: de,fr MIME-Version: 1.0 To: DJGPP workers Subject: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Reply-To: djgpp-workers AT delorie DOT com Hello. I was trying to build Emacs CVS against DJGPP CVS yesterday. I had some trouble, because of a typo of the one of the Emacs config headers - src/config.in. But I noticed that the typo was in a section conditional on !HAVE_SIZE_T. Strange, I thought, we have size_t and surely configure would have detected it. It turns out that Emacs builds against the header . describes itself: "/* config.h for DJGPP. This is usually generated automatically from config.h.in by the configure script. However, it is very hard to run that script under MS-DOS, because of its extensive use of Unix shell features. This header file is provided so you can skip the autoconfigure step altogether and go directly to the compilation step (after copying Makefile.in to Makefile and setting the defaults there). There are several parts in this header file, which closely follow the GNU Autoconf procedures." (Rest of description omitted.) I can see that this script was useful in earlier days, but why do we have it now? configure scripts run just fine now. This header could be a maintenance problem. I think we need to do one of the following: (a) remove and force all packages (Emacs?) to use configure; (b) look at the latest autoconf/automake and update . I prefer option (a). Option (b) will lead to DJGPP being behind autoconf/automake tests all the time. Given how long it takes between DJGPP releases, I don't think this is acceptable, especially when configure works. Bye, Rich =] -- Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]