From:	pjfarley AT dorsai DOT org (Peter J. Farley III)
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Rebuilding gcc - "c-parse.gperf" not found
Date:	Sun, 31 Aug 1997 20:39:17 GMT
Organization:	None
Lines:	48
Message-ID:	<3409d234.8669182@snews.zippo.com>
References:	<Pine DOT SUN DOT 3 DOT 91 DOT 970831191649 DOT 17131F-100000 AT is>
NNTP-Posting-Host:	news.newsdawg.com
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Eli Zaretskii <eliz AT is DOT elta DOT co DOT il> wrote:
<Snipped>
>gcc2721s.zip is not LFN-clean.  DJ Delorie creates the archive on
>plain DOS machine, where it is not easy at all to make an archive that
>will work on LFN platforms.
>
>You will need to rename all the offending files to the long names that
>the configuration scripts and Makefiles expect to see, then run the
>whole thing again.

Eli,

You misunderstood me, I think.  It's not the files that need renaming
to make the makefile work, it's the makefile itself that still has
long names, when the files themselves have, in fact, all been made 8.3
clean (or so it appears to me so far).  Even the initial
"configur.bat" file in the gnu\gcc-2721 directory calls
"config\msdos\configure" instead of "config\msdos\configur", and
promptly gets "Bad command or filename" as an error message!

Hmm-m-m.  I wonder if it is possible to have a Win95 long-name file
with a nice, normal 8.3 "real name".  E.G., if the long name is
c-parse.gperf, the 8.3 name would still be "c-parse.gpe", not
"c-pars~1.gpe",  but the long name would be "c-parse.gperf".  Probably
not, but it's a nice thought.  Not likely that M$ made it that easy
for DOS folk, is it?

>
>I suggest you also submit names of all the files that needed to be
>renamed to DJ Delorie so that they'll be corrected in future releases.

I will, indeed notify DJ of everything I find.  I'm also writing a
Gawk program to scan through the makefile and clean all the long names
to 8.3, with a double-check to see that the 8.3 actually exists before
performing the rename.  The output will likely be a sed script to
perform all the necessary changes automatically.  I'll post it to the
newsgroup when it's done.  (Yes, I know, it's probably better/faster
done in perl, but you use the tools you know best, and learn the new
ones as and when you can.)

Another thought just occurred to me.  If I turn OFF LFN support, would
that let the makefile work as-is?  Seems logical, since make would be
using the DJGPP libc support, which reads the LFN variable, and so
might return file "c-parse.gpe" when asked for "c-parse.gperf".  I'm
going to try that.  I'll report back on the the results.

----------------------------------------------------
Peter J. Farley III (pjfarley AT dorsai DOT org)

- Raw text -