KDE 3.5

[Jeff Spaleta]

On 12/23/05, Thomas M Steenholdt <tmus at tmus.dk> wrote:
> I believe it's been mentioned on this list somewhere before, that
> updates-testing is probably not getting the attention (it should?) from
> a wide variety of testers. My feeling is that updates-testing is
> *mostly* being used by people who have a bugzilla for productX and needs
> to test if their bug has gone away and stuff like this.
> This may or may not be intentional (or even true), but that's what I
> seem to recall from previous discussions.

I think you are wrong. The yum 2.4.1 to updates-testing definitely got
noticed by several testers, and the bugs in the new caching feature
were caught and fixed before it got pushed to updates-released. People
are using updates-testing and it would have been perfectly suitable to
spin up kde packages in updates-testing.  And i think its absolutely
absurb to make any claim about how the repo is being used. We have no
stats on the usage pattern of updates-testing.. nor do we have ANY
policy with regard to having testers register that they have tested a
certain package. We have an unknown number of people testing an
unknown percentage of the packages. We can't draw any conclusions from
that.
The fact that we don't have metrics on the testing pattern, is not an
excuse to avoid providing the testing packages. Suck it up and provide
the test packages.

If people are really concerned about the amount of testing being done,
then make it a policy that every testing update annoucement has a bug
report referenced in the annoucement and have every tester provided a
"i tested it" comment.

Where updates-testing falls down is when hardware specific
functionality needs testing.
Coverage over a full spectrum of hardware isn't going to happen.
Kernels in updates-testing are never going to get enough hardware
coverage to catch every problem before the kernel moves to
updates-released.  I have run every single updates-testing kernel that
has come over the wire.. and for my hardware and my configuration i've
seldom noticed a problem.  You can not expect updates-testing to get
full hardware coverage, and you certaintly can't expect people to
screw around as much trying different hardware configurations as
compared to someone running rawhide.

But for application layer software and most libraries beneath them...
updates-testing packages can be useful and will get tested. You don't
need comprehensive hardware coverage to catch problems with kde. You
just need users who do everyday user things, in kde.

> If it had been in updates-testing, would you have taken the time to
> seriously and thoroughly testing it?

If the existence of the kde 3.5 testing packages were made known in
the forums and in fedora-list, then I'm pretty sure rabid kde users
would show up and chew on the packages.

-jef