A few comments on the resulting debdiff; none of these are issues that you've introduced, but as long as we're merging (sorry, I know that was my idea :), I think they should be addressed.
This change is not documented in the current changelog entry; it looks like it was dropped somewhere along the line. The original changelog description is this:
base-files (3.1.0ubuntu2) hoary; urgency=low
* debian/postinst: Do not install /usr/local and subdirectories with "staff"
group writeability. This group is essentially root-equivalent, but there
are cases where somebody can become any user but root (like NFS).
Could you summarize this for the current changelog entry as one of the 'remaining changes'?
This change:
@@ -70,6 +71,7 @@
if [ ! -L /var/mail ] && [ ! -d /var/mail ]; then
ln -sf spool/mail /var/mail
fi
+ install_from_default /usr/share/base-files/networks /etc/networks
fi
preserve_directory floppy
forces the creation of /etc/networks on /every/ upgrade of the package. Since anyone who has upgraded to feisty or later will have had this file created already, we shouldn't need to keep this delta. I think it should be dropped.
+if [ "$2" = 4.0.1ubuntu3 ]; then
+ if [ -f /etc/motd ] && [ ! -L /etc/motd ] && [ -L /etc/motd.old ]; then
+ rm /etc/motd
+ mv /etc/motd.old /etc/motd
+ fi
+fi
This is also transitional code, for a hardy pre-release version of base-files, that I think should be dropped.
I unfortunately can't find any reason documented in the changelog for this change; and glibc 2.4-1ubuntu3 was superseded by a later debian sync, so changelog archaeology there also gives me no ideas. I think this code should *probably* be dropped on the grounds that this is a transitional depends only, but it's difficult for me to say for sure.
Could you provide an updated debdiff addressing these points? On the last point in particular, I'm probably willing to go along with whatever you decide is best here.
Hi Thierry,
A few comments on the resulting debdiff; none of these are issues that you've introduced, but as long as we're merging (sorry, I know that was my idea :), I think they should be addressed.
--- base-files- 4.0.4/debian/ postinst 2008-03-11 11:02:27.000000000 +0100 4.0.4ubuntu1/ debian/ postinst 2008-05-19 00:44:17.000000000 +0
+++ base-files-
200
@@ -5,8 +5,8 @@
if [ ! -d $1 ]; then
mkdir -p $1
fi
- chown root:staff $1 2> /dev/null || true
- chmod 2775 $1 2> /dev/null || true
+ #chown root:staff $1 2> /dev/null || true
+ #chmod 2775 $1 2> /dev/null || true
}
install_ from_default( ) {
This change is not documented in the current changelog entry; it looks like it was dropped somewhere along the line. The original changelog description is this:
base-files (3.1.0ubuntu2) hoary; urgency=low
* debian/postinst: Do not install /usr/local and subdirectories with "staff"
group writeability. This group is essentially root-equivalent, but there
are cases where somebody can become any user but root (like NFS).
Could you summarize this for the current changelog entry as one of the 'remaining changes'?
This change:
@@ -70,6 +71,7 @@ from_default /usr/share/ base-files/ networks /etc/networks
if [ ! -L /var/mail ] && [ ! -d /var/mail ]; then
ln -sf spool/mail /var/mail
fi
+ install_
fi
preserve_directory floppy
forces the creation of /etc/networks on /every/ upgrade of the package. Since anyone who has upgraded to feisty or later will have had this file created already, we shouldn't need to keep this delta. I think it should be dropped.
+if [ "$2" = 4.0.1ubuntu3 ]; then
+ if [ -f /etc/motd ] && [ ! -L /etc/motd ] && [ -L /etc/motd.old ]; then
+ rm /etc/motd
+ mv /etc/motd.old /etc/motd
+ fi
+fi
This is also transitional code, for a hardy pre-release version of base-files, that I think should be dropped.
+ifeq ($(shell dpkg-architecture -qDEB_BUILD_ ARCH),amd64) /Depends: libc6 \(>= 2.4-1ubuntu3\),/' \ tmp/DEBIAN/ control
+ sed -i -e 's/^Depends:
+ debian/
+endif
I unfortunately can't find any reason documented in the changelog for this change; and glibc 2.4-1ubuntu3 was superseded by a later debian sync, so changelog archaeology there also gives me no ideas. I think this code should *probably* be dropped on the grounds that this is a transitional depends only, but it's difficult for me to say for sure.
Could you provide an updated debdiff addressing these points? On the last point in particular, I'm probably willing to go along with whatever you decide is best here.