On Tue, Aug 19, 2003 at 12:13:58AM +1000, Anthony Towns wrote: > On Mon, Aug 18, 2003 at 01:06:20PM +0100, Colin Watson wrote: > > Hm, yeah. So the brave new world will be that bugs won't be listed as > > fully closed in unstable until all architectures are in sync? > > Point. Having a 'arch=source' option might be a better default then? If > it's possible, I'd like an 'arch=*' or 'arch=all' option, though.
'arch=source' is there for support for src=foo queries, but probably won't quite do what you want because it deals with versions of source packages. I could invent a couple of new pseudo-architectures for this easily enough though: let's make 'arch=default' (or nothing) show the bugs corresponding to the current source package associated with whatever you're asking for, and 'arch=*' show the bugs open across the board. > > I think, then, we let the 'found' bugginess state win, then 'fixed', > > then 'absent', so when listing bugs across architectures the buggiest > > one wins. I've got untested code for that now. > > I've no idea what you mean here. Right now, pkgreport.cgi decides where to display a bug based on precisely one version of a package. If it's deciding based on multiple versions, then it needs to be able to resolve the different answers it gets. So, I'm suggesting that if you have: 0.1-1 s390: bug wasn't even reported here 1.0-1 mips,mipsel: bug was found here, but not fixed 1.0-2 alpha,arm,i386: bug was fixed here ... then you want to report the bug as 'found', since that's the most conservative (buggiest) option, etc. > > Eventually yes, although it's convenient for debugging at the moment. > > Actually, maybe have it default to yes but allow "showotherverbugs=no"? > > I rather like having them there to make sure bugs don't get lost, TBH. > > Yeah, make the default "yes" now, consider changing it later maybe. I > think we'll definitely need to change it when we start mixing in archived > bugs -- listing all the bugs still open in stable would get ugly fast, IMO. Oh, yes, you're right. And slow, too, at least if we start indexing version information so that we don't have to read all the .status files. -- Colin Watson [EMAIL PROTECTED]
