Hello Andreas,
> Well, I'm a bit concerned about > > https://lists.debian.org/debian-qa/2013/09/msg00001.html > > The thing is that if we manage to get new teams involved it might > increase the acceptance if the way from editing a tasks file to the > creation of the metapackage is "short" in terms of not beeing forced to > wait for one day until UDD is updated. It could be discussed with DSA > on IRC where they are more responsive than on mailing lists like > debian-qa. > > Yes I agree, talking on IRC sounds good we can schedule that. > It also could speed up things if we could > > a) make the UDD importer capable to import a single Blend (which could > speed up things drastically for one Blend that just changed its > tasks files > consequently we should create separate log files per Blend which > would make more sense anyway. > b) use some "cache" for the tasks files of the Blends like storing some > md5sum of recently imported tasks files which do not need to be > imported twice > c) try to make the insertion procedure itself more efficient by for > instance: > - check, whether we could speed up the check for a package that > just exists in UDD > - inject all packages in one rush > > This should give some ideas what could be done as a bonus to the GSoC > task. > > These are good ideas, but I am not quite sure if I have the time needed to deploy them before the GSoC deadline (upcoming Monday). So if you also agree for the moment I will sum up all the documentation we have and probably finish up the sgml to xml convertion (it should not take very long). Once I am done with these (and finish some other university obligations this week ) I will then start working on the above ideas :-). Kind regards Emmanouil PS: In case we don't speak again before Friday, have a nice time in your upcoming vacations :-)
