Last night, I had dinner with William Yu,
Miguel Paraz, and
Jijo Sevilla. Jijo organized the
get-together.

While waiting for Migs and William, Jijo and I chatted about his new
IT consultancy. He described the fat client Debian-based system he had
in mind, with computers automatically drawing updated packages from a
central repository. Jijo wanted to know what I thought of the project.

I told him to check out FAI, a Debian package for doing
fully-automated installs. I also told him about configure packages to
help propagate configurations. He could set up a network-booting
system, with CD backups in case the network went down. He was
delighted – these tidbits would make his job a whole lot easier.

He asked me because he knew I watched the debian-user list and so
might know of similar problems and solutions.

Here’s how I keep an idea of all Debian packages in my head:

– I spent some time reading the package summaries of all the packages

available in Debian’s unstable+experimental archives. This was part
of choosing new packages to install, so it was fun.

– I use aptitude’s “forget” command to mark all of those packages as old.

– Whenever I update my package lists, aptitude marks packages I

haven’t seen as “new.”

– I review those and then use “forget” again to mark all of them as old.

That’s how I keep track of all existing packages available for i386. I
still use apt-cache search and aptitude’s search functions
extensively, but at least I know what to look for.

I also watch debian-devel for Intent to Package (ITP) announcements. I
used to keep track of freshmeat.net, but found it too far from my
normal workload, and not useful enough.

Looks like that’s my Geek Power – I’m a search engine. ;)