[Mesa-dev] stupid development model causing more problems than its worth

[Zack Rusin]

On Thursday 29 April 2010 20:58:04 Dave Airlie wrote:
> On Fri, Apr 30, 2010 at 10:42 AM, Jose Fonseca <jfonseca at vmware.com> wrote:
> > We discussed the pros & cons of this several times, there was never
> > consensus, and there is always room for improvements, but I refuse to
> > discuss them in such terms, where you paint this absurd image of vmware
> > bullying everybody around.
> 
> So far the only people who have come out *for* this system are from
> the vmware group, apologies if calling the group by the company name
> irks people, I suppose I could just name everyone individually, and I
> was more meaning people who work on the vmware drivers irrespective,
> since they seem to have the ongoing development in stable branch
> methodology that no other project people work on uses. For one none of
> the older driver contributors seem to be following this model, and
> these driver have been around a lot longer. I'm mainly calling you
> guys out as you have a very different deployment model that what all
> the other mesa drivers have, since you don't actually ship anything in
> distros or seem to place any priority on it, but anyone who does want
> a useful stable branch has to conform to this wierd and frankly
> unusual development model.
> 
> What set this off this morning, if contributors to radeon for example
> like Maciej have been diligently fixing bugs on master, now with the
> development model I cannot cherrypick all these fixes back into 7.8
> without causing some upset to someone. Like ideally Maciej would care
> about stable branches and the like, but I don't think placing a
> requirement on community and free time contributors to take on the
> overheads that regression testing stable entails vs just regression
> testing their fixes in master as they add features. The thing is fixes
> in stable might actually cause regressions in stable, that you won't
> find with this model, until its too later. Ideally fixes should go
> into master, stew for a week or two, get some more community testing
> and then go into stable. I understand the stable fixes model might
> work if you have a QA team and comprehensive test coverage but for
> most of the community maintained and even company maintained drivers
> that isn't a resource we all have to hand.
> 
> So sorry I've called out vmware, go back and ^vmware^brian + keith.

Dave, whatever points you might have had were trumped by the sissy fit you 
threw on top of them. I'm not sure what social contexts are you living in that 
you expect a big "screw you" to result in "ah, yes, lets work to accommodate 
the screw you" but this isn't one of them.

It all sounds like it based on the premise that big bad VMware (or Brian and 
Keith as you say) is/are out there to screw the poor old Dave. Which is 
especially silly given that the "stupid development model" was already 
discussed.
Either way I'd suggest that you start a new thread without all the drama, so 
that we can discuss this again without all the baggage that you added here.

z