On Wed, 2011-03-02 at 00:20 +0000, Adam D. Barratt wrote: > On Tue, 2011-03-01 at 23:38 +0000, Adam D. Barratt wrote: > > Self-cluestick: b2 already has a get_full_tree method which iterates > > reverse-dependency lists, so we should just use that. Updated patch > > coming to a b2 tree near you soon. > > Actually, it looks like get_full_tree() is hugely overkill in this case. > Purely building the complete list of affected packages when considering > debianutils takes several minutes (I have to admit I got bored and > cancelled the run after five minutes of no obvious activity) whereas > with my admittedly less elegant patch the whole run takes less than > that.
... because my patch isn't doing what it says on the tin in the general case. :-( Making the necessary changes to correctly loop over all the reverse dependencies isn't particularly involved, but does take quite a bit longer than b1 when checking debianutils; my initial testing last night suggests it's in the order of a few minutes per architecture, producing a combined list of approximately 11,000 cumulative reverse dependencies each time. Regards, Adam -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org