Building a tool to make the process automated

[Lois Desplat]

Hi,

Thanks to Joey Hess and his CVS and SVN articles which allowed me to find
this mailing list and give me more information. I am a student in the Google
Summer of Code and I am making a program that allows to put $HOME and /etc
under revision control automatically and provides tools to easily administer
them. As I wrote my SoC abstract, I wanted to use SVN and it seems that is
what most of you use. The primary backend that I will use will be Bazaar-NG
but we can stick with SVN now since I will also implement it with SVN.

My first question is, do you have any general advice? I have gone through
Joey Hess' articles and Scott Scriven's article as well as put my own $HOME
and /etc directories under a VCS (Bazaar-NG).

I found that quite a few applications store information that is not
necessary to put under revision control (firefox's cache files), do you
handle that individually or do you usually not care?

Under SVN, how do you make sure that the repository does not grow too large
over the years? I would assume that after a few years, you might want to
just keep a monthly granularity of your backups so as to reduce the size of
the repository?

Some things that I am going to do, my application involves a daemon that
listens to file movements in the HOME directory. This way a user can just
move files normally and the daemon records the file movement. When it is
ready to auto-commit, it can do bzr mv (in Bazaar, the move operation is a
first class operation) or at least let bzr know that the file moved instead
of being deleted and recreated which helps in keeping the versioning
history.

I do not know how to handle deletion though. When you delete a file, is it
possible to delete all its version information too. I know it is very hard
to not possible to do in Bazaar, how about in SVN?

Thank you very much for the wiki which I am going through right now and your
articles.

Lois Desplat.