Re: list what's in the NEW queue?
On Friday 04 February 2005 14:14, Brian M. Carlson wrote:
> On Fri, 2005-02-04 at 13:26 +0100, Frederik Dannemare wrote:
> > On Friday 04 February 2005 02:30, Steve Langasek wrote:
> > > On Thu, Feb 03, 2005 at 04:05:19PM +0100, Wouter Verhelst wrote:
> > > > Op do, 03-02-2005 te 15:44 +0100, schreef Frederik Dannemare:
> > > > > > which
> > > > > > requires no imprimatur from the DPL, before you start
> > > > > > throwing packages that have never even been tested by their
> > > > > > maintainer at us faster than we already get them.
> > > > >
> > > > > I see your point if it is really the case that uploads are
> > > > > being done without proper testing from the maintainer
> > > > > himself/herself.
> > > >
> > > > His point is still valid even if all maintainers do proper
> > > > testing. You can't be expected as a maintainer to be able to
> > > > test /every/ possible or impossible situation in which a
> > > > package could be used. And then I'm not even talking about
> > > > packages that should conflict with eachother but don't, because
> > > > the maintainer of the new package didn't know that a file in
> > > > his package happens to have the same name as a different file
> > > > in a completely unrelated package...
> > >
> > > What I know is that every time an ftpmaster processes a batch of
> > > NEW packages, a percentage of them wind up in testing with
> > > serious bugs for failing to declare build-dependencies, and then
> > > the release team has to track these bugs.
> > >
> > > Since the testing scripts have no way to distinguish an
> > > architecture-specific package from a broken binary that only
> > > builds on the maintainer's system, the only strategies I can
> > > think of off-hand that would be effective at reducing this
> > > problem are to disallow all binary uploads from maintainers,
> > [ snip ]
> > Yes, much better to have everything built by the buildd in a clean
> > env, IMO. This would be on my wishlist for post-sarge. This topic
> > was also discussed (for other reasons, though; security concerns, I
> > think it was) last summer.
> I think this is an awful idea. This means that developers will no
> longer test their packages before uploading,
[ snip ]
I surely hope they would still do so. Another option could simply be to
proceed with the current way of uploading - but then let the buildd
rebuild the uploaded binary. Or is that somehow not feasible?
As of right now it is troublesome to build e.g. gl stuff as a maintainer
if you are using the nvidia drivers on your system. I'm sure there are
many, many other scenarios to choose from.
Frederik Dannemare | mailto:email@example.com
http://frederik.dannemare.net | http://www.linuxworlddomination.dk