On 15 Aug 2000, Paul D. Smith wrote:
>
> %% Christian Goetze <[EMAIL PROTECTED]> writes:
>
> >> First, Rational doesn't publish any API or other interface that would
> >> allow GNU make to communicate with MVFS and construct the data required,
> >> much less communicate it to the VOB and view servers. For years I asked
> >> them for something like that; I've given up.
>
> cg> It's really a pity that the free software community is so
> cg> infatuated with CVS. It's a genuine case of "they don't know what
> cg> they're missing", probably combined with a dose of general
> cg> mistrust of SCM. I really think the time has come for some kind of
> cg> open source MVFS...
>
> I think this would be difficult. I'd be very surprised indeed if Atria
> (now Rational) didn't hold patents on at least some of the crucial
> aspects of MVFS.
>
> Obviously there have been opaque filesystems around for a long time, but
> MVFS is much more than just that (although you could make pretty cool
> SCM toolset based on opaque filesystems, too...)
Actually, all things considered, one doesn't really need an MVFS. All one
needs is for the repository to be a database, combined with the ability to
store both the structure and the versions of files and directories and the
ability to generate reliable snapshots. Sure, MVFS is much nicer...
The real quantum leap forward would be to break out of the flat file model
for storing code and into a more structured storage model. I dream of the
day where I can use any names I want and display code any way I choose
without creating merge conflicts - essentially distinguish between the
meaning and the presentation - or "skins" if you choose...
We were so close with lisp machines, and then we lost it...
--
cg
GPL discussion skipped - I just don't know enough about the legalities
involved, and honestly, I don't find GNU make to be worth the effort.
Make is just fundamentally broken.
