Re: Making fsfs generate unique transaction names

On 6/13/07, C. Michael Pilato <cmpilato@collab.net> wrote:
> If it's not obvious, I'm
> already bummed that we're using sqlite in the BDB backend, but I think it's
> too late to try to remedy that at this point.

Not really. For what it's worth, the merge tracking folks have done a
good job of abstracting out the mergeinfo storage from the logic that
uses it. Check out include/private/svn_fs_mergeinfo.h: the internal
API is just four non-sqlite-specific functions (and a #define which is
only used externally for hotcopy). I don't think it would be very
hard for a BDB hacker to implement similar functionality using BDB
instead of sqlite and call that from libsvn_fs_base instead of the
libsvn_fs_util functions. (I certainly would expect, say, Google Code
to be be doing something like this for their backend...) Most of the
active work, and all the high-level mergeinfo logic, are in other
parts of the code.

(Yeah, I know this sounds like "patches welcome", but the point is
that this is well-enough designed so that it shouldn't be much harder
than adding other tables to BDB. Of course, I've never touched BDB so
I have no idea how hard that is...)

--dave

-- 
David Glasser | glasser_at_mit.edu | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org