Re: another approach to what current does

On Tue, Aug 13, 2002 at 05:01:26PM -0400, Hunter Matthews wrote:
> On Tue, 2002-08-13 at 16:47, Jeff Johnson wrote:
> > On Tue, Aug 13, 2002 at 04:31:03PM -0400, Hunter Matthews wrote:
> > > On Tue, 2002-08-13 at 16:18, Michael Redinger wrote:
> > > > -----BEGIN PGP SIGNED MESSAGE-----
> > > > Hash: SHA1
> 
> > > More significantly in my mind is rpm 4.1. How/where did the api (esp the
> > > python api) change.
> > > 
> > 
> > Try <[email protected]> :-)
> > 
> 
> Documentation would be nice.

Doco assumes stability, and there has been little development
(and as corollary, !development == stability++) in rpm-python
for quite a while now.

FWIW, most of what I've added is documented, and is closely
parallel to the rpmlib API doco, which is fairly complete
these days.

What's still insane is having a C programmer design python bindings,
but that can't be helped. I'm getting there as fast as I can.

For some datapoints:
	1) up2date took only a couple of days to get running.
	2) yum took an afternoon to get (mostly) functional
and I hardly know python at all.

What is gonna be painful is changing the paradigm, as both headers
and rpmdb objects are coming out of the rpmlib API as much as I can
manage, in order to reduce the memory footprint and increase performance.

There are, of course, lighter data structures and equivalent methods
to replace the Good Old Stuff. For starters, you're gonna want
a transaction set object with nearly global scope.

73 de Jeff

-- 
Jeff Johnson	ARS N3NPQ
[email protected] ([email protected])
Chapel Hill, NC