Hello,
On Tue, Nov 30, 2004 at 10:26:53PM -0800, Steve Langasek wrote:
> On Tue, Nov 30, 2004 at 10:46:18AM -0600, Robin Verduijn wrote:
> > On Mon, Nov 29, 2004 at 11:10:17PM -0800, Steve Langasek wrote:
> > > If you are not a buildd maintainer, you should not be uploading binary
> > > packages to the archive that you have not tested. This is triply true of
> > > the testing-proposed-updates queue, which gets practically zero real-world
> > > testing by users before being committed to testing.
>
> > So there is nothing I can (or should) do to help kvirc get out of t-p-u
> > into testing? Will it start being picked up by autobuildds at some
> > point and still be able to make it into testing? The conclusion I draw
> > from your email is that t-p-u never really gets the testing it should
> > since packages in t-p-u will always get in through the backdoor, so to
> > speak.
>
> t-p-u gets minimal user testing; this means the burden is on uploaders and
> the release team to ensure the correctness of the packages added this way.
>
> It is generally assumed that the buildd environments will upload correct
> (which is to say, consistent) packages. No such guarantees are possible
> with hand-built packages; instead, we trust that hand-built packages have
> been tested by the uploader. If this is *not* the case, there are no other
> safeguards against broken binary packages making it to testing by this route.
> So if anything, uploading untested binaries to t-p-u is likely to mean more
> work for the maintainer and the release team, in order to fix the problems a
> broken package would cause.
This is a real problem. A while ago I wanted to use a package[1] just to
find out some files were missing. The reason was the maitainer had
asked to get it build manually which somehow got those files missing.
The fix was a trivial rebuild, but still this can be easily avoided by
sticking to autobuild packages unless you can test them yourself.
Greetings
Helge
[1] This is a real case and if you happen to need more info, just mail
me in private.
--
Helge Kreutzmann, Dipl.-Phys. Helge.Kreutzmann@itp.uni-hannover.de
gpg signed mail preferred
64bit GNU powered http://www.itp.uni-hannover.de/~kreutzm
Help keep free software "libre": http://www.freepatents.org/
Attachment:
pgp49Oqffmrzk.pgp
Description: PGP signature