Frederic Peters finished making crash.gnome.org look like a GNOME site. Andreas Nilsson helped by changing the icon to one made by Sebastian Kraft (which Fer still hasn’t committed in a Bug-Buddy version.. hint hint hint). Further, the site now is only accessible via port 80. The port 5000 stuff is gone/firewalled. Nice to get people helping out!
I asked Frederic if he could change the other GNOME sites (except Bugzilla and the soon obsolete www.gnome.org) to use the new style layout as well. Anyone is free to help out with this. Just look for SVN modules ending with ‘-web’.
For mail.gnome.org, Frederic changed the main page and added the CSS + images. After that I copied his stuff and changed the archive index page as well (created by a script in a sysadmin-only SVN module). Until you click a few times, mail.gnome.org looks much better. Note that mailman layout is difficult to change, I only want to concentrate on the archives.
Screenshot:
Unfortunately, mail.gnome.org wasn’t updated automatically from SVN. There is a script to handle most of it in a sysadmin module. So I enabled this for mail.gnome.org. Immediately I got an error message via the mail… forgot that the mail.gnome.org machine (menubar) is RHEL3. The script assumes at least Python 2.4. A lot of sysadmin scripts cannot be improved just because of that one machine still being RHEL3. As it is our mail machine, any downtime will have a noticeable effect. However, I really am annoyed by the RHEL3 + too many custom RPM packages and general hackiness of the setup.
That is why I plan to upgrade the machine on October 20 to RHEL5. This is probably the most uncertain upgrade out of all machines hosted at Red Hat. Many of the packages used by that machine are custom (Postfix, mailman, mhonarc, amavis.. basically everything that should run on the machine is not standard RHEL3). Since the upgrade of the other machines to RHEL5 I haven’t seen much progress towards avoiding anything other than a ‘upgrade and see what happens’.
It might be worth using the VMWare stuff to pull the machine into a VMWare image and then try upgrading the image to see what breaks?
IMHO before upgrading the system and replacing a mess of custom installs over RHEL3 with a mess of custom installs over RHEL5 you should check everything you need is available in EPEL5, and if not make EPEL5 move in the right direction (either by contributing packages or motivating others to do it).
Otherwise you’ll just be continuing down the vicious circle of custom installs