Last night, I wrote code to use the full promotion set all the time and optimized that path a bit (there was a very simple change that eliminated 60% of the run-time of the program). I'm going to look at the queuing and hybrid approaches this evening (I have some prototype code but it's too buggy to test). My guess is that the hybrid approach will be a significant win, since for many of the steps in a typical search, only a handful of promotions will be emitted between when the step is enqueued and when it's examined.
Daniel _______________________________________________ Aptitude-devel mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/aptitude-devel
