C. Michael Pilato wrote:
> I spent a few hours thinking about the next generation of our
> filesystem last night, and I'm convinced at this point that to undergo
> some sort of major code unification effort today would be a waste.
> Better to begin thinking about how to properly segregate the business
> logic from the storage format peculiarities in the next generation of
> the filesystem;
>
Oh, yes...
> better still to invest real effort into deciding if
> we've a need for multiple backends at all.
>
Actually, I think we want to keep multiple backends for aesthetic
reasons if nothing else. It will help with maintaining the separation
you mention above.
Of course, we first have to get the separation right.
(And make the BDB backend design sane; I still think that FSFS
outperforms BDB because its storage model and implementation are
designed better, not because a filesystem is "intrinsically" faster --
at least, for larger repositories, one would expect that a proper
database with indexes, etc., would scale better.)
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Nov 3 02:22:19 2005