.cvsignore audit

[Chipzz <chipzz ULYSSIS Org>]

OK, so I'm building GNOME 2 from cvs for a while now.

Since I noticed make clean doesn't do it job very well in all cases,
and make distclean doesn't either, I made a little script to reverse my
tree to the state it was in just after I checked it out. This script is
included here:

#!/bin/bash
for i in `find . -name .cvsignore` ; do
(
 echo $i | sed 's:/.cvsignore::'
 cd `echo $i | sed 's:/.cvsignore::'`
 rm -fRv `cat .cvsignore`
)
done

I noticed that after running this script on my whole tree some of the
modules were rendered inbuildable. So I checked out a clean copy of the
whole tree, copied it to a new location, and ran the above script on
the second tree, followed by a diff -ur between the two trees. The re-
sults are included below, and were in some cases astounding. I have
grouped the results together according to what kind they are.

All modules in this list are HEAD, except gtk+ and glib which are the
stable branches from CVS.

I don't know what the policy is about removing files from CVS. If our
policy is not to do this, please disregard my statements about this.

| Only in src2/at-spi: COPYING
| Only in src2/gnome-desktop: COPYING
| Only in src2/gnome-panel: COPYING
| Only in src2/gnome-session: COPYING
| Only in src2/metacity: COPYING
This file is installed by automake. Don't know if automake replaces this
file when it's allready there. We have 2 possibilities here:

a) automake doesn't replace this file when it exists, and we have our
   version. If we have our own version, then why is it in .cvsignore?
b) automake replaces it regardless. So it's generated and shouldn't be
   in cvs anyway.

Also note that allmost all modules have this file in CVS.

| Only in src2/eel: po
| Only in src2/yelp: po
No comment ;)

| Only in src2/gnome-applets: macros
| Only in src2/gnome-control-center: macros
Not sure, but I think it's generated and shouldn't be in CVS. Please
correct me if I'm wrong. (It's possible that this fir may have to be in
CVS if we have some custom m4 files in there - not sure how much this
occurs thou).

| Only in src2/gnome-control-center/capplets/background: background.desktop.in
| Only in src2/gnumeric: gnumeric.desktop.in
These are definately wrong.

| Only in src2/gnome-applets/gweather/docs/C: gweather
| Only in src2/gnomeicu/doc/C: gnomeicu
| Only in src2/gnome-utils/gdialog/doc/C: gdialog
| Only in src2/gnome-utils/gdialog/help/C: gdialog
These appear to be harmless and the dirs should be removed from CVS

| Only in src2/gal/po: POTFILES.skip
Not sure, but I think this file shouldn't be in .cvsignore.

| Only in src2/gconf: ABOUT-NLS
No idea at all.

| Only in src2/gconf-editor/src: gconf-marshal.c
| Only in src2/gconf-editor/src: gconf-marshal.h
These files are generated from src2/gconf-editor/src/gconf-marshal.list
and shouldn't be in CVS.

| Only in src2/gnomeicu/src: .xvpics
Again, I think this dir should not be in CVS.

| Only in src2/gnumeric/doc/es: func.defs
| Only in src2/gnumeric/doc/ja: func-list.sgml
| Only in src2/gnumeric/doc/ja: func.defs
| Only in src2/gnumeric/doc/ja: functions.sgml
| Only in src2/gnumeric/doc/ru: func.defs
Not sure about these.

| Only in src2/gnumeric/plugins/dif: plugin.xml.in
| Only in src2/gnumeric/plugins/sc: plugin.xml.in
| Only in src2/gnumeric/plugins/sylk: plugin.xml.in
These look wrong to. In the .cvsignore file there's this line which is
the culprit:

*.in

Which should be Makefile.in I guess (Makefile is allready in there).

| Only in src2/gtk+/po: Makefile.in.in
I think this file shouldn't be in CVS.

| Only in src2/gtk+/tests: makefile.msc
| Only in src2/glib: makefile.mingw
| Only in src2/glib: makefile.msc
Not sure about these.

| Only in src2/libbonoboui/samples/compound-doc: bonobo-hello
| Only in src2/libbonoboui/samples/compound-doc: container
Particulary bad because container is in Makefile.am SUBDIRS, which cau-
ses the build to fail.

| Only in src2/libgnomeprintui/doc: html
| Only in src2/libgnomeprintui/doc: sgml
Not sure.

| Only in src2/libgnomeprintui/tests: testprint.c
| Only in src2/libgnomeprintui/tests: testprint4.c
Ok, testprint.c is in the Makefile.am but is commented out. testprint4.c
is not in there and can I think be removed from CVS.
Also none of the other .c files are in the Makefile.am.

| Only in src2/libgnomeui/images: .cvsignore
How did this get in there in the first place? If the person who made
this change had a local version of this dir and changed the .cvsignore
file, and therefore didn't want his version of .cvsignore in CVS,
shouldn't the cvs binary on his system have ignored this file
(.cvsignore being in .cvsignore)? Looks like a broken cvs binary, or
maybe a bug. Or is this intentional behaviour? (I checked, .cvsignore
is actually in there, not a wildcard or something)

| Only in src2/pan: intl
Not sure, but again, I think it shouldn't be in CVS.
gnome-applets also has: gnome-applets/battstat/intl which shouldn't be
in CVS either if I'm right.

| Only in src2/pan/tests: Makefile.am
| Only in src2/pan/tests/data: Makefile.am
| Only in src2/pan/tests/data/cache: Makefile.am
| Only in src2/pan/tests/data/save: Makefile.am
| Only in src2/pan/tests/data/tmp: Makefile.am
Basically, all of these files suffer from the same problem: they have
'Makefile*' in them.
Also, the 2 last dirs are basically empty and I think shouldn't be in
CVS anyway.

OK, that's it for now. I also noticed that some modules don't have *.o
in their .cvsignore files for source dirs, but I'll get to that later.
I hope I didn't step on any toes and that I contributed something use-
full.

Kind regards,

Chipzz AKA
Jan Van Buggenhout
-- 

--------------------------------------------------------------------------
                  UNIX isn't dead - It just smells funny
                            Chipzz ULYSSIS Org
--------------------------------------------------------------------------