home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Geek Gadgets 2
/
Geek_Gadgets_2_2352.bin
/
lists
/
ade-gcc.archive.9602.gz
/
ade-gcc.archive.9602
/
000102_owner-ade-gcc_Wed Feb 21 12:52:39 1996.msg
< prev
next >
Wrap
Internet Message Format
|
1996-02-28
|
3KB
Return-Path: <owner-ade-gcc>
Received: by fishpond (Smail3.1.29.1 #57)
id m0tpIgL-000gXea; Wed, 21 Feb 96 12:50 EST
Sender: owner-ade-gcc
Received: by fishpond (Smail3.1.29.1 #57)
id m0tpIgH-000gXUC; Wed, 21 Feb 96 10:50 MST
Message-Id: <m0tpIgH-000gXUC@fishpond>
From: fnf (Fred Fish)
Subject: Re: GCC 2.7.2 patches
To: kiskra@ernie.icslab.agh.edu.pl (Kamil Iskra)
Date: Wed, 21 Feb 1996 10:50:20 -0700 (MST)
Cc: ade-gcc@ninemoons.com
In-Reply-To: <Pine.SUN.3.91.960221161537.3291A-100000@ernie> from "Kamil Iskra" at Feb 21, 96 04:27:53 pm
X-Mailer: ELM [version 2.4 PL23]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 1910
Sender: owner-ade-gcc@ninemoons.com
Precedence: bulk
> Size of "diffs" surprised me a bit - 400 KB in comparison to less than 100
> KB in previous distribution.
>
> It seems that some very strange things got integrated into this patch, or
> rather into Fred's source tree.
>
> One of them is "gcc.info", which is very, very big and is the main culprit
> of the size of the patch.
> Second one is (if I remember correctly) c-parse.c.
These are probably artifacts of the way the rules are written in the Makefile
and the timestamps on the files in my source tree, which is checked out from
CVS.
My guess is that the c-parse.c file appeared to be out of date at some point
with respect to the c-parse.y file (or whatever the yacc source file name is),
and the Makefile caused c-parse.c to be regenerated. Because of a bug in the
Makefile, the c-parse.c file is regenerated into the source directory when it
should actually be going into the build directory.
This also points out a more subtle problem. Typically FSF distributions
include automatically generated files so that people who try to build them
won't have to worry about having bison, texinfo, etc. This leads to problems
like this one.
I'm considering removing these automatically generated files from both the
baseline distributions (which already have to be repacked anyway) and the
CVS source base, and then making sure that the Makefiles are modified
if necessary to ensure that they get generated into the build tree.
The other problem at work here is that the diffs are generated automatically
from both the baseline source archive and the amiga source archive, so when
these files appear in the baseline archive and the source archive, and the
the one in the source archive has been regenerated, you get big extraneous
diffs like this. Removing them from both archives would fix the problem,
as would coming up with a complicated scheme to omit certain files from
the diffs.
-Fred