F20 System Wide Change: ARM as primary Architecture

[Jiri Eischmann]

"Jóhann B. Guðmundsson" píše v Čt 11. 07. 2013 v 09:17 +0000:
> On 07/10/2013 09:28 PM, Adam Williamson wrote:
> 
> > On Wed, 2013-07-10 at 13:56 +0000, "Jóhann B. Guðmundsson" wrote:
> > > > On 07/10/2013 12:36 AM, Bill Nottingham wrote:
> > > > 
> > > > > > Plus, in relation to this - the llvmpipe issue brings up that one of
> > > > > > the 'release blocking desktops' *does not work*. This would, by definition,
> > > > > > block the release unless we intend to have different criteria for ARM as a
> > > > > > primary arch.
> > > > 
> > > > Then we should remove the default label and "release blocking
> > > > desktop(s)" entirely concept with it. 
> > > > 
> > > > It's far outdated anyway and relic from the past.  
> > > > 
> > > > Each sub-community ( be it spins be it various arch ) should need to
> > > > provide the necessary QA/Releng resources from their sub-community
> > > > ( if no such thing the relevant party needs to build one ) while we QA
> > > > and Releng focus our available resources on the components that
> > > > everyone in the whole distribution use and provide the necessary sub
> > > > community with the assistance in relation to QA and Releng.
> > I'm afraid I can't agree. I like the simplicity of the model you're
> > proposing, but from a practical point of view, there is still a commonly
> > held perception that there is a 'product' called Fedora which is
> > basically composed of what you get if you go to get.fedoraproject.org,
> > download one of the things we push at you there, and install it.
> > Practically speaking, I believe we have to QA that 'thing called Fedora'
> > as a whole. I don't think your model quite matches what people perceive
> > Fedora to be.
> 
> What's your definition of what people perceive Fedora to be?
>  
> Regardless of peoples perception or what you think they are there
> still would be products however we QA would take care of the installer
> + base/core os bits and sub-communities that build upon that base/core
> OS like Gnome takes care of QA their own spins. 
> 
> Who better are to QA their own spin then the people that a) use it b)
> create it c) release it?

Have you ever reached to these sub-communities and asked them if they
are interested in something like this? Because I'm not sure whether
they're and forcing something like this upon them is not the best idea.
First I think keeping the QA know-how in one team actually has a lot of
benefits. 
Second I don't think that the sub-communities or spins, if you will,
have enough manpower and expertize to run their own QA teams.
The current team of Fedora testers is rather small and dividing them
into several more teams would lead to a subcritical number of testers to
run a functional QA.
Third I don't agree with the opinion that people who create the software
are the best people to test it. Quite the opposite. Having a QA team
independent on spins is actually a very good idea. I have witnessed a
lot of situations where developers were too lazy/busy/... to fix serious
problems and it was pertinacity of our independent QA team that forced
them to fix the problems although it sometimes took an escalation to
FESCo. I highly doubt that spin's QA teams would develop into such
independent entities.

Jiri