From: Mark Bergman Subject: Re: yet another question about gcc-2.3.3 (fwd) [Porting to DOS] To: djgpp AT sun DOT soe DOT clarkson DOT edu Date: Mon, 8 Feb 1993 00:54:35 -0500 (EST) Forwarded message: >From bergman Thu Jan 21 01:19:38 1993 Subject: Re: yet another question about gcc-2.3.3 To: turnbull AT ecolan DOT sbs DOT ohio-state DOT edu (Stephen Turnbull) Date: Thu, 21 Jan 1993 01:19:38 -0500 (EST) In-Reply-To: <9301202313 DOT AA16297 AT ecolan DOT sbs DOT ohio-state DOT edu> from "Stephen Turnbull" at Jan 20, 93 06:13:24 pm MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3289 ==> Subject: Re: yet another question about gcc-2.3.3 ==> To: djgpp AT sun DOT soe DOT clarkson DOT edu (DJ's GPP mailing list) ==> Date: Wed, 20 Jan 93 18:13:24 EST ==> From: Stephen Turnbull ==> In-Reply-To: <9301202243 DOT AA23415 AT cfa0 DOT HARVARD DOT EDU>; from "Bob Babcock" at Jan 20, 93 5:43 pm ==> X-Mailer: ELM [version 2.3 PL11] ==> ==> > >It seems that there is considerable interest in making djgpp compliant with ==> > >gcc version 2.3.3. It also seems that a principal obstacle to this goal is ==> > >the existence of files in gcc-2.3.3 that have names longer than 8 characters. ==> > > ==> > >My question is, what's the big deal? Change a few file names, change the ==> > >Makefile(s), problem solved (?). ==> > ==> > I don't know if this is the problem, but one thing which would make it more ==> > difficult is if some of the include files have long names. Then source code ==> > changes would be necessary, possibly in many places. One possibility would be ==> > a preprocessor to adjust names in #include statements. (I thought about using ==> > command line args: -Dlongfilename=shrtname, but the C preprocessor won't do ==> > this.) ==> > ==> ==> Given that somebody (Alan E., I think it was) made the paranoid ==> suggestion that this is all a plot by RMS to make it impossible to ==> support Messy-DOG (sic) versions of GCC, maybe what we need to do is ==> write an AWK/Perl script to do it on the entire GCC distribution. ==> Then RMS's plot will fail, since we'll be ready the next time :-> ==> -- Well, I hesitate to mention this, but I've got a really ugly hack (Korn shell scripts, various Unix utils, awk) that runs under the MKSS toolkit that handles this. For example Extract the files from a tar archive save error messages re. long file names Build a table of long_file_names ==> unique short names in the form of a sed script Run fgrep on every file in the tree to find files that contain references to the too-long (now shortened) files. Build a list of these files. Check in (RCS) each file to-be-changed Run sed on the files that need to be changed using the table to convert all references within files to to proper short names. Check in the changed version of each file This is really gross, and has a problem in that the MKS Toolkit's tar isn't smart enough to generate unique directory names, or unique file names if the differences are 3 or more characters past a "." (joe.file1 and joe.file2 both become joe.fil). If someone wants to use my scripts as a basis for a general purpose utility, they are yours for the asking. They are not ready for public consumption as-is, though. If djtar[tx] could generate unique filenames on the fly, <== HINT, HINT I believe that everything that I have written could be run using GNU/pd tools and a hideous DOS batch file or some other script, or even an awk script as the wrapper. ==> ==> Stephen Turnbull ==> The Ohio State University, Department of Economics ==> 410 Arps Hall, 1945 N. High St., Columbus, OH 43210-1172 USA ==> Phone: (614) 292-0654 Fax: ...-3906 Email: turnbull DOT 1 AT osu DOT edu ==> -- Mark Bergman (Biker, Stagehand, (former) Unix user support grunt) 718-855-9148 bergman AT panix DOT com {cmcl2,uunet}!panix!bergman -- Mark Bergman (Biker, Stagehand, (former) Unix user support grunt) 718-855-9148 bergman AT panix DOT com {cmcl2,uunet}!panix!bergman