johnm@nvuxd.UUCP (J. Montgomery) (12/13/84)
<eat me> I would appreciate comments on the relative merits of RCS and SCCS. I DO NOT intend to start a raging debate on a subject that I'm sure has been ventilated before, so please reply to me, rather than the net. Thanks in advance! -- John Montgomery Bell Communications Research ...{allegra,ihnp4}!nvuxd!johnm
lcc.barry@UCLA-LOCUS.ARPA (12/18/84)
In addition to the efficiency of RCS, it has one other advantage: the ability to attach symbolic tags to revisions lets you tie together corresponding revisions to different modules. Since the modules may have been modified independantly, the revisions may have different numbers (delta #s in sccs). Forcing a new major level each time a change affects more than one module in a system just isn't practicable. Another thing I liked about RCS was that it was easier to use and hid its control files in a separate directory (if you provide the directory). Since then I've seen shell scripts to do the same for sccs--it's just that the nice interface and file hiding is already there. RCS's locking scheme also seemed nicer to me, though I can't quite say why (not using an extra ".p" file? a specific command/flag for breaking a lock (rcs -u)?). On the other hand, sccs is available on most unix(TM) systems. RCS is available only by arrangement with Purdue University. It's not a difficult arrangement to make, but the last copy of RCS I saw explicitly forbade redistribution as part of a commercial system. If you plan to sell a system with source and the same tools you used to build/maintain it, you CAN'T use RCS. That's why I'm currently stuck(?) with using sccs. barry ucla-cs!lcc!barry lcc!barry@ucla-cs.ARPA