seth vidal wrote:
On Mon, 2007-04-30 at 15:00 +0200, Florian Festi wrote:
Jeremy Katz wrote:
On Mon, 2007-04-30 at 12:24 +0200, Florian Festi wrote:
RpmSack.whatProvides and .whatRequires build package objects inside but only
return the pkg tuples. This is really annoying as most places need to
reconstruct these objects later.
While (at times annoyingly :-) inconsistent, changing this breaks API
compatibility and thus I'd really rather not do it, at least not at this
time.
We could go the (still ugly) route of adding a aspo flag and only do it
when that's passed; that would allow new callers to get the benefit
while not breaking older callers. And if we wanted to keep things
consistent, we could implement the same flag on the other sacks, just
with the different default
Could look like the attached patch...
I'm not keen on messing with depsolving this late in the game. I realize
it is a minimal variation but its still worrisome. How about this:
1. we branch for 3.2
2. we can work on your sqlitesack and rpmsack patches for HEAD w/o worry
of breaking too much.
Sounds good for me. That patch wasn't intended to go into F7. My empty PRCO
patch may be worth a consideration as it has a high speedup/LOCs. To be
really sure it could be modified to use real flags somewhere else or an
empty tuple (to support the list interface) instead of None.
I am currently working on a mail to the list with much further suggestions
to finally get rid that performance issues. So a new branch might be a good
idea.
Florian
_______________________________________________
Yum-devel mailing list
[email protected]
https://lists.dulug.duke.edu/mailman/listinfo/yum-devel
