[CALUG] Basic open source CM

[Bryan Smith]

Berend Tober wrote:
> Definitely not a web app, but can you get much more basic than RCS? It's
> old, er, I mean very-well vetted and works great for revision history
> tracking for the simple case of individual files.

James Ewing Cottrell III wrote:
> Um, NO! Don't use RCS or CVS EVER Again!
> Just Don't.
> JIM

Care to explain that?

First off, he never said CVS.  I agree it's much better to look to
another, more modern, centralized revisioning store with Xdeltas for
binary (e.g., Subversion) or the various, distributed ones that work
in this day'n age of disconnected systems (e.g., GIT), than CVS.

But, back to what he did say, RCS ...

All RCS does is create version (,v) files in the same, or any
pre-created ./RCS sub, directories.  It's a very useful tool for
locally revisioning files.  At most, it causes issues for Extended
Attributes (e.g., POSIX ACLs and SELinux contexts), but it is
otherwise a solid, basic diff tool for text files.

I.e., if you don't have anywhere to setup a service or directory, it
can be used -- e.g., in haste -- to quickly revision files.

Then, later, the RCS version (,v) files can be converted to another
type in another store (like Subversion or GIT).  It's also very easy
to tarball up all RCS version (,v) files on a system.

E.g., someone built a system, you made changes to it, and you want to
capture all those changes for later review or even re-applying,
especially after they blow it away.

In any case, RCS is definitely better than _no_ revision history on a
file.  I've caught so many people making changes to /etc/hosts and
other files by just always using RCS.  While I personally would use
GIT instead, and setup a repo in /srv/git, /var/lib/git or even
${HOME}/git, if necessary, I can understand why people sometimes still
create version (,v) files immediately where and when needed.

-- bjs

P.S.  Is this one of those classic Jimmy Knee-Jerks you have from
time-to-time where you go off on something, and don't tell people why?
 ;)


--
Bryan J Smith  -  http://www.linkedin.com/in/bjsmith
E-mail:  b.j.smith at ieee.org  or  me at bjsmith.me