FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

[Ralf Corsepius]

On 03/01/2010 11:27 PM, Adam Williamson wrote:
> On Mon, 2010-03-01 at 08:07 +0100, Ralf Corsepius wrote:
>
>>> So yeah, I agree it's not a perfect system - detailed suggestions for
>>> improving it would be welcome, I'm sure.
>>
>> Alternatives:
>>
>> * Abandon it (I don't think this would change anything wrt. to QA in Fedora)
>
> Um. Hard to put this tactfully. You're completely wrong.
Feel free to disagree.

As I wrote before, "voting on quality" is the wrong approach and a flaw 
of the system.

> Being actually
> *in* QA I have quite a lot of experience with this, and the Bodhi system
> has already stopped lots of broken updates being shipped.
Then your experiences with karma is different than mine.

May-be you experienced a negative comment to let the maintainer revisit 
a package? This has nothing to do with the karma votes. It's a 
side-effect of the testers utilizing bodhi as communication medium.

> Throwing it
> out would be insane.
Well, I disagree again. I consider the current karma stuff to be flawed 
and keeping it to be silly - The majority of packages don't receive any 
votes and if they are being ignored.

To me the karma voting systems is a means rel-eng utilizes to *cheat 
themself*.

>> * In cases an update is trying to address a particular bug in BZ,
>> replace let people comment in bugzilla.
>
> You mean, let people check a box to have their comment from Bugzilla put
> into Bodhi?
Yes, I had something along these lines in mind.

The idea behind it:
* individual reporters have been facing a bug and BZ'ed it.
* maintainer releases a new package release which is supposed to address 
reporter's bug.

Now let all feedback go into reporter's BZ.

Nowadays, in such kind of situations, a maintainer often adds a "Should 
be fix in release x.y.rpm, please try" comment and a adds a "BUGFIX, 
BZ#NR" to bodhi.

>> All the voting/karma stuff does is to let rel-eng believe to be dealing
>> with bad updates, while it actually doesn't cope with the problems it is
>> trying to address, it's the wrong tool.
>
> As I said, I just disagree. I have seen many cases where updates that
> otherwise would have gone out and caused real pain to real people have
> been caught by Bodhi. The fact that some weren't caught by Bodhi doesn't
> mean it's useless.
Again, I feel what you might have seen are a very small subset of 
isolated cases ... in the overwhelming majority of cases, this karma 
stuff is not being utilized at all ... and even if, it is being ignored.

>>> I think it's pretty easy to make a case
>>> that Bodhi has had a significant positive impact on the overall quality
>>> of the updates that have fully utilized it.
>
>> Well, the only positive impact bodhi had on me was bodhi implementing a
>> more or less usable web-frontend, where Fedora had nothing in place
>> before. This doesn't mean it is a good system and even less does this
>> mean this system is perfect or bug-free.
>
> Didn't I just get done saying it's not perfect or bug-free, but that
> doesn't mean the sensible answer is to burn it down?
I didn't say "burn it down"!!

Bluntly put, I am saying "karma is dysfunctional/conceptionally 
inapplicable", the "web GUI leaves much to be desired", "there is no 
usable CLI" and "koji/bodji" integration sucks" ... but this is 
off-topic, here.

Ralf