Re: Version issues

Dylan Barrell (dbarrell@opentext.com)
Fri, 23 Apr 1999 11:05:18 +0200


Message-ID: <01BE8D79.2C99AAA0.dbarrell@opentext.com>
From: Dylan Barrell <dbarrell@opentext.com>
To: "'ejw@ics.uci.edu'" <ejw@ics.uci.edu>,
Cc: "Falkenhainer, Brian" <Brian.Falkenhainer@usa.xerox.com>,
Date: Fri, 23 Apr 1999 11:05:18 +0200
Subject: RE: Version issues

The problem with the space for which we are creating a standard is that it 
is a space which is dominated by lots of niche products. i.e. there are 
different sets of requirements which are considered "minimal" depending on 
the end users (SCCS, PDM, DocMan, Content Managament etc.). There are 
different configuration requirements for a source code control system such 
as MKS Source Integrity as there are for a general web site tool such as 
Site Server. It is arguable whether both products should have some 
configuration support in order to be useful to their users.

Confining my comments purely to the goals document I think it is desirable 
to have orthogonality of support of various "areas" (for lack of a better 
word). The areas in question are

Versioned resources
Configurations
Activities
Workspaces

And within the individual areas (most) there should be 2 levels of support 
- "minimal" and "advanced".

Nobody can argue that minimal versioned resource support does not include 
branches and parallel development. At the same time it must be possible to 
manipulate these versioned resources without any support for activities. It 
should be possible to have workspaces without support for either activities 
or configurations and there is definitely value in configurations without 
activities or workspaces. Minimal configuration support certainly doesn't 
include automatic inheritance and conflict queues and one could go on.

The only absolute requirement is that the minimum requirements for 
versioned resources are met as most of the other areas (with the exception 
of activities) are meaningless without versioned resources.

Cheers
Dylan


On Thursday, April 22, 1999 6:40 PM, Jim Whitehead [SMTP:ejw@ics.uci.edu] 
wrote:
>
> > I would like to see the levels split even more than that. I believe 
that
> > some of the configuration management functionality can be considered
> > "minimal" in order to be useful and the rest should be optional.
> >
> > Has this forum had a detailed discussion on the levels of the 
versioning
> > protocol? If not I think the time has come.
>
> So, the problem with just adding another level is that each level
> exponentially adds to the possible client/server interactions.  With two
> levels, you need to consider level 1 client vs level 1 server, level 1
> client vs. level 2 server, level 2 client vs level 1 server, and level 2
> client vs level 2 server.  With three levels, there are now 9 possible
> permutations, not just four.
>
> The other difficulty is developing a configuration management layer where
> the abstractions are so easily separable that a third layer can easily be
> defined.
>
> Did you have any thoughts on what features should be in this minimal
> configuration management layer, and which features should be excluded? 
 Or
> was this just a general concern that the design might be getting too
> complex?
>
> - Jim