using RCS to track system changes

[lgu at pobox.com (Larry Underhill)]

On Mon, 2005-12-26 at 20:00 -0500, Tom Metro wrote:
> A CVS working directory, which /etc/ would become, as well as any 
> subdirectories of it that contained config files, would each need a
> CVS 
> directory to hold the CVS meta files. That seems like unnecessary 
> clutter. RCS simply has one history file per file under its control,
> and 
> you have the option of whether to collect them into RCS
> subdirectories.

I also use RCS for control of config files (and various admin scripts
that invariably get written) on my server at home. I agree with Tom that
it is a practice that should be more widely employed. The only place we
seem to differ is that I do setup a RCS subdirectory in places that I
need version control. However, I do it with a symlink that points to
common directory (/var/RCS).

This is mainly done for backup purposes. My backup script always takes
care of /var/rcs, so if I dump a new package into /usr/local/foo and I
modify /usr/local/foo/foo.conf to suit me then a simple symlink
of /usr/local/foo/RCS => /var/RCS will ensure that my changes get backed
up without having to remember to modify my backup script. 

YMMV, but it is helpful and has saved my butt in the past. 

--Larry