[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: "testing" improvements

En réponse à Anthony Towns <aj@azure.humbug.org.au>:

> On Thu, Feb 27, 2003 at 03:32:59PM -0500, Matt Zimmerman wrote:
> > The way I see it, it doesn't matter one bit whether anyone even
> _looks_ at
> > testing until we start to prepare for a release.  Until that time, all
> of
> > the "testing" happens in unstable, and "testing" is misnamed.
> And around we go again. The "testing" that is meant to be happening on
> testing is user-level testing, not "is this package completely fucked
> over". That means things like:
> 	* can we install the system?
> 	* can we upgrade smoothly from previous releases?
> 	* does the behaviour of the system match the docs?
> Unfortunately, we've regressed into being unable to maintain unstable
> at minimal levels of quality, so the relevance of any of the above is
> somewhat distant.

So you admit that people would be encouraged to use testing,
so "testing" is not misnamed at all?

I agree with about people the responsibility of people that
have to fix bugs.

But don't you think our dependencies are too strict. In a lot
of cases, most packages work well with dependencies from
testing but they are built against those from unstable.
Don't you see any way to improve this? (versioned symbols
in libs would be ideal, but it is a problem with

> All these discussions remind me a lot of phrases like "fiddling while
> Rome burns". Seriously: we've got well over 700 RC bugs in unstable;
> we've got an installer that barely works at best; and we're spending
> all our time trying to find excuses _not_ to work on these things?
> Here's a thought experiment for y'all. First, you notice that glibc
> is holding lots of things in testing up. Second, you notice it has two
> outstanding RC bug reports: #181493 about Sun RPC code being
> potentially
> non-free, and #181494 about the GNU FDL as used in glibc being
> non-free.
> Third, you notice not much has been done about these bugs. The
> question
> is: is your instinctive response:
> 	(a) Post to Debian lists / websites about how horrible all this is,
> 	    and how someone (else) should do something about it.
> 	(b) Try to convince everyone (or at least me) that those bugs
> 	    aren't that important and shouldn't hold back glibc from
> 	    testing.

Didn't you mentioned some day that it would be possible to rebuild
packages against testing?

Jérôme Marant <jerome@marant.org>


Reply to: