Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!uakari.primate.wisc.edu!ginosko!uunet!pcrat!rick
From: rick@pcrat.uucp (Rick Richardson)
Newsgroups: comp.sources.d
Subject: Re: RCS -vs- SCCS (was Re: rcs for SysV)
Message-ID: <1989Oct18.033152.2035@pcrat.uucp>
Date: 18 Oct 89 03:31:52 GMT
References: <6552@ficc.uu.net> <77@venice.SEDD.TRW.COM>
Reply-To: rick@pcrat.UUCP (Rick Richardson)
Organization: PC Research, Inc., Tinton Falls, NJ
Lines: 54

In article <77@venice.SEDD.TRW.COM> baur@venice.SEDD.TRW.COM (Steven L. Baur) writes:
>
>Why do you use RCS, and why should I also use it?  I would love to be
>convinced.

If I had my druthers, I'd use none of the above.  Neither is capable
of automating the software configuration management and manufacturing
process without a lot of roll it yourself customization with
shell scripts/functions and a real database to track workspaces, changes,
modification requests, releases and the like.  I'd also dump make and use
fourth generation nmake (which I loved during the 1 1/2 years that I was
a heavy user). * (see below)

However, to answer the question, I switched to RCS because out of
the box it already had a decent user interface and I was tired of
having to reinvent for the Nth time a usable interface to SCCS.
I also believed at the time I switched that it was capable of
handling binary files (for icons, fonts, etc).  It turned out that
this was wrong.  You can check binaries in, and check them out under
RCS, but you cannot delta them.  This is one better than SCCS, so
I'm still happy.
I also like the fact that the ID keywords ($Id$, $Log$) are preserved
and *expanded* even in a checked out for modification copy.

The icing on the cake, though, was that RCS has symbolic revision
numbers, so you can freeze a release, call it R1_2, and later
check out the whole release by asking for it by symbolic name,
no matter what the individual file revision numbers are.  SCCS
has no equivalent unless you tack on a real database management
system, or use the contrivance of forcing revision numbers
for all files to some common number.

I was always aware that the supposed performance advantage of RCS
was an urban legend.  It was ease of use that drove me to give up
seven years of SCCS experience and switch to RCS.

* What I'd like to do is write an MR, either bug or feature,
and promise it fixed in some release number.  Then, when I go
to make the release, all the MRs are checked to see that they are
included in some "commentary" before it lets me make the release.
Additionally, MR's not committed to a release, but fixed nevertheless,
will be updated automatically "fixed in release NNN".  So when I'm
done with a release, I have a piece of software and release notes
describing exactly what was changed, keyed to the MRs, and nothing
fell thru the cracks.  Neither RCS nor SCCS is this integrated.
They are mearly low level tools that I suppose you could build
such a system around.

-Rick
-- 
Rick Richardson |       Looking for FAX software for UNIX/386 ??????     mention
PC Research,Inc.|                  WE'RE SHIPPING			 your
uunet!pcrat!rick|    Ask about FaxiX - UNIX Facsimile System (tm)        FAX #
(201) 389-8963  | Or JetRoff - troff postprocessor for the HP {Laser,Desk}Jet