Hi Stefano, On Dienstag, 17. März 2009, Stefano Zacchiroli wrote: > Not a big deal. The PTS has always had maps between binary and source > packages. So if you prefer to produce result ordered per binary > packages that is fine too; we'll load them, associate them to binary > packages and link to the individual binary failures. The only > potential problem is for source packages with a lot of binary packages > which will all be failing. That will potentially imply a lot of links > for the PTS, which is not something we would like. Right. I think I'll bite the bullet and write a(nother) Sources file parser... (rather, I'll extend packagesdb.py to handle that. Thats also why I guess reusing code will not really be useful.) > If, in your experience, it is rare to have such situations we can > ignore the problem and live with the sporadic link issues. If not I > would personally prefer a per-source report and link to it from the > PTS. I'm not sure how often this will happen, but I think the source centric view makes most sense. > > But then "latest" should be replaced with "$version". > This is fine *if* you plan to have piuparts run often "enough". Not > having them for, say, months and having for the same period a warning > which was fixed several uploads ago will become annoying for > maintainers. Actually, I (now) think it will become $source/$distro, where $distro is either squeeze or sid. Or maybe not: > Well, I'd say first of all to tag them appropriately in the report, > just to avoid loosing information. In the beginning we can ignore the > distinction and just "sum" them up in the total number of failures, > delegating to the target report page a more detailed explanation. But > nothing inhibits to show them in some way, e.g., with a tooltip on the > link target. I'll know for sure once I have see the first reports per source package, but I think it's better to have a page per source per distro. It also saves me from dealing with renames and removals and such. I'll ping you (here) once those reports are ready. regards, Holger BTW, I've added detection of false positives due to network failures today \o/
Attachment:
signature.asc
Description: This is a digitally signed message part.