On Wed, Aug 10, 2005 at 08:54:34PM -0400, Jason Chu wrote:
> Judd develops pacman.  He accepts patches from people when he sees fit.
> He'll solicit some questions from the developers now and again, but that's
> about it.
> 
> It would be nice if pacman was more open, but Judd doesn't want to do that
> until libpacman is out.  I guess that's when it'll become a full project,
> with a mailing list and everything else.

Yes.  And for the record, nobody wants to see libpacman ready more than
I do.  Trust me.  But this is summertime, and I am away on vacation for
another week or so.  And I'm trying my best to ignore any non-urgent
emails and general work until I get back home.

> Berkleydb in pacman?  Doubtful.  Judd's not a big fan of its reliability or
> its hash-based nature.  Sqlite, maybe.  Judd has a better list than I.

Aurelien and I have talked about a different backend database.  We
decided to leave the existing structure in place for now -- Aurelien
reports that he's got some significant performance improvements in
libpacman, and that said improvements may be enough to assuage the
people that complain about pacman's running time.  We'll see.

As for features, the initial libpacman will not have a lot of new
features, per se, but it will be completely restructured and more
welcoming to new features in the future.  Well, the CODE is more
welcoming to new features -- as some of you know, I often am not.  ;)

I've spoken of transaction/rollback support in libpacman before.  We
don't have it in place yet, but I'd like to have it in 3.0 if we can
wing it.  But you guys are already leaning pretty hard on us to get it
out the door, so I don't know if we can survive the barrage of "When
will it be out?" questions long enough to implement something like that
for 3.0.

And yes, the libpacman CVS will be public once we have a stable and
documented API.  The reason I don't open it up now is that I don't want
anyone developing a frontend around it before we have something stable.
Aurelien has done almost all of the work so far, and he's very close to
finished, I'd say.  We still have to document the whole thing and then
it will be thrown to the masses for some testing/breaking/fixing.

You've all survived this long without libpacman.  I'm sure you can wait
a while longer.  :)


- J

_______________________________________________
arch mailing list
[email protected]
http://www.archlinux.org/mailman/listinfo/arch

Reply via email to