[cvsnt] Re: Branch merging - this seems wrong...

Michael Wojcik Michael.Wojcik at microfocus.com
Wed Jun 7 16:48:14 BST 2006


> From: cvsnt-bounces at cvsnt.org 
> [mailto:cvsnt-bounces at cvsnt.org] On Behalf Of Tony Hoyle
> Sent: Wednesday, 07 June, 2006 10:38
> 
> OTOH most SCM's don't have the branch/merge capability that CVS has 
> (hence the whole reserved/locking paradigm that still sticks around so

> much - they can't merge so you have to stop other people working on
the 
> files).

While I'm generally in agreement with you in this thread, I don't think
this is accurate.  Certainly RCS and its descendants and clones (such as
PVCS) all offer branching and merging.  RCS used (but did not enforce) a
locking paradigm because Walter Tichy was trying to build a better SCCS,
and SCCS used a locking paradigm.

But it's certainly possible to use RCS with a merge paradigm (concurrent
updates) rather than a lock paradigm (serialized updates).  Of course
that's just what CVS originally was: RCS without bothering to lock
files, and merging when more than one set of changes was applied to a
given revision.

In other words, it wasn't the introduction of branch/merge capability
that changed the paradigm from locking and exclusive updates to lockless
concurrent ones.  That capability already existed; what changed was that
people realized that merging was sufficiently robust and conflicts (and
hence manual intervention) sufficiently rare to make concurrent updates
feasible.

-- 
Michael Wojcik
Principal Software Systems Developer, Micro Focus



More information about the cvsnt mailing list