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

Re: status of jackd? (bug #318098)



David Nusinow wrote:
On Tue, Aug 09, 2005 at 10:45:16PM +0200, Kurt Roeckx wrote:

On Tue, Aug 09, 2005 at 10:28:58AM -0400, David Nusinow wrote:

Where would you like us to do our work? This is exactly what unstable is
*for*. It lets us break things while they're in development in order to
push the distro as a whole forward.

No, that's what experimental is for.  If you upload something to
unstable, it should be ready to migrate to testing in a short
period.  And it would be best that you could "prove" that it's
ready to go to testing before you upload it to unstable.


This doesn't really work that way in reality, because you're just pushing
things up yet another level. At some point non-developers will use
experimental the way they use unstable now.

Let's give a concrete example that I have some experience with: X.org. I
uploaded it to unstable with the best of my own knowledge that it was ready
to ship. It turned out not to be, breaking things due to port issues and a
mistaken belief that it would have to deal with the C++ transition. I broke
a hell of a lot of things with that upload that I didn't know I would
break. I had unofficial packages that I advertised on planet debian well
beforehand, and I got a lot of testing out of them (got a few important bug
fixes in the first upload to unstable as a result) and I *still* broke a
lot of things when I uploaded to unstable.
And I *still* haven't been able to upload a completely fixed package that
can migrate to testing (waiting on arm now...). It's been over a month that
I've been working on that.

The point is, things break and we need a place to break them. Whether you
name it unstable or experimental makes no real difference.

well, sometime bugs get all the way to stable, no software is without bugs. What I was talking about is that 'unstable' is pretty much only usable desktop so it shouldn't be broken on purpose. Or the release cycle should be <1 year. Or something. Maybe the solution for (desktop) users is to use some actively maintained sort of unstable meant for end users - e.g. ubuntu.

BTW I think it makes a lot of sense to use experimental for most of the initial testing and only release to unstable when it looks like the package is a release candidate. Yes, some problems might only be uncovered in unstable but that's what the unstable is for:-)

Summary of my point: given the imporance of unstable the attitude "who cares, it's only unstable" doesn't seem to make sense (see the jackd problem (bug #318098), jackd which is crucial for audio processing is uninstallable for several weeks and it's not even explained what's going on).

	erik



Reply to: