governance, fesco, board, etc.

[Ralf Corsepius]

On Tue, 2007-06-12 at 11:10 -0500, Rex Dieter wrote:
> Ralf Corsepius wrote:
> 
> > * bodhi, koji are immature, semi-functional, semi-cooked pieces of SW
> > which still have to prove their longevity, but so far don't do anything
> > but introducing bureaucracy and are almost strangling former FE.
> 
> But "good enough" to get the job done *now*.
I disagree. It's bad enough to kill FE.

> Could you provide constructive suggestions(1) on how to make things better?
Kick this new workflow or at least simplify it. Don't force maintainers
to use web-forms when maintaining packages.

In a nutshell: The actual effect of bodhi and koji on maintainers
condenses to:
* eye-candy
* much more complex workflow/more bureaucracy


> > * rel-eng could have done much better.
> 
> Anything lacking now?
Well, _release_ means a point in time, i.e. most of their mistakes show
effect at one point in time: release time (and install time).

Issues having shown since the release:
1. releasing unsigned packages
2. releasing mal-packaged packages (cf. gprolog*.rpm)
3. ACLs
4. mock still lacking fedora-7-*.cfgs (effects local build on cvs check
outs).
5. koji, bodhi are directly connected to rel-eng enforcing their (RH)
visions violently.
6. Them refusing to acknowledge "Release early/release often". If they
don't believe in it, so be it, but they should not hinder maintainers
who are convinced in it (like me) from applying it, as they do now.
7. The server layout (Everything/Fedora). At least I fail to understand
why there can't be one single repo, containing "all of Fedora" with a
subset of the packages therein packaged as iso's.
...

>   rel-eng agendas, meetings, are open you know.
1) Mon. 19:00 UTC closes me out
2) I am not interested in directly participating in rel-eng.
I am just contributing packages to Fedora and using Fedora as an
ordinary user. As such am a "just victim" of their deeds.

> > * fedora-testing is a nothing but a hoax close blinding yourself about
> > QA - I consider it to be a dead born child. The best I can say about it
> > is it causing delays in updates.
> 
> I strongly disagree.  Tis a good thing updates-testing is optional, 
> you're welcome not to use it.
I already said everything needed on other occasions, but I'll try to
recap:

1. One can only test, if one knows what and how to test for. 
A "testing group" doesn't have any possibility to have the knowledge, HW
nor required infrastructure.

2. Nobody will ever be able to prove "a program's completeness/
correctness". A "testing group" is no exception.
All one can test for is "has a particular bug been fixed". 
This already is the job of a package's maintainer.

3. So far neither the testing repo (for FC7) nor the testing group has
provided any evidence to be functional.

Mal-packaged packages have been released, maintainers continue to push
update candidates through private repos.

4. The impact this kind of "testing" has on the workflow of contributors
is unreasonable.
It renders "cvs commit; make tag build" non-functional, but requires you
to additionally launch a browser and wade through various web-pages.

Can we please have a "make release"?


5. "Test to death" is the closed source way of thinking.
"Release early/release often" and "fix ASAP/release fix ASAP" is the
working principle having shown to be functional in OSS. 
FE and Core had been no exceptions.

Ralf