yum error

[Miloslav Trmac]

On Tue, May 17, 2005 at 06:26:58AM -0400, Ivan Gyurdiev wrote:
> > Why should it be? The point of rawhide and test releases is development,
> > so what justifies slowing down what's pushed to rawhide?
> 
> I disagree - the point of test releases is not development - it's
> stabilization.
But rawhide is _not_ a stable source of updates for the test release.
A test release is a _single point in time_, with no scheduled updates.
The development repo is enabled in test releases to allow further testing
if you want to (e.g. to test bugs you reported about specific packages
have been fixed).

> The point of rawhide may be development, but broken
> dependencies are unnecessary, and scare away the beta testers.
I think they actually _protect_ the beta testers (from software not
working because the dependencies changed).  If dependencies are
necessary, broken dependencies are necessary too.

> A package which introduces dependency damage will not be tested,
> until the dependency damage is fixed - so why is it even released
> until that point?
The package _has to be built_ before the "damage" can be fixed by updating
other packages.  Rawhide is just a collection of built packages.  Nobody
looks at the daily compose to decide whether to "release rawhide" or whether
it is "not good enough".  Rawhide just _happens_.

Test and final releases are "released" and should contain no broken
dependencies.  During the freeze for FC 4 rawhide will more and more
resemble a final release, but that's not caused by what rawhide is,
or what rawhide should be; it is caused purely by what the final release
should be.
	Mirek