Schedule for Wednesday's FESCo Meeting (2013-08-14)

[Adam Williamson]

On Thu, 2013-08-15 at 16:12 -0400, "Jóhann B. Guðmundsson" wrote:
> On 08/15/2013 03:47 PM, Kevin Fenzi wrote:
> > On Thu, 15 Aug 2013 15:36:53 -0400
> > "Jóhann B. Guðmundsson" <johannbg at gmail.com> wrote:
> >
> >> Interesting since they did not do that when I joined QA what 5 or 6
> >> years ago so again can you refer me to that discussion.
> > It's always been a test case/critera that I remember...
> >
> > http://fedoraproject.org/wiki/QA/Meetings/20070117#Fedora_7
> >
> > Shows upgrade test cases were there in Fedora 7 and one of the things QA
> > was testing and ensuring.
> 
> We tested for it to a limited extent ( via yum ) but we never officially 
> supported it.
> 
> We always stayed away from opening that pandora box and it was not until 
> we found out that someone had stamped upgrades to be "officially" 
> supported that we actually properly defined what should be tested, added 
> the criteria for it and made it release blocking.
> 
> And the discussion around who "officially" stamped it and why is what 
> I'm looking for ( not that it has been technically possible for number 
> of years ) and I'm pretty sure it was neither Jeremy,Will,Chris or Seth 
> that pushed for that "official upgrade" stamp when they introduced 
> pre-upgrade once they had finish writing it, since all four knew the 
> ramification for us in QA by doing so.
> 
> I can tell you that fedup blindly inherited the "offically upgrade 
> tool/support" from pre-upgrade by fesco decision, while Will was still 
> scratching his head designing/writing it and Tim being the only one that 
> was properly testing what Will threw over the wall.
> 
> To many including me that seemed like an odd decision making instead for 
> example simply not "officially" support upgrades ( thus not making it 
> release blocking ) until that tool had been written.

Just to make something clear, since at least one person was confused by
what Johann was saying: I've not been around long enough to comment on
the history here, but as things stand, there is a requirement for a
clean stock upgrade case using the 'recommended' upgrade mechanism(s) to
work in the release criteria, and QA does test this as part of release
validation.

Johann is discussing the history that led to this being the case and
questioning how exactly we ever came to "support" upgrades in this way
in the first place, which I don't have any idea about, it's before my
time. Just wanted to make clear that as things stand right now,
upgrading via fedup is "supported" in that it's required by the
validation process to work to the extent described above.

I personally like to try and use the word "recommended", as in, if
you're going to do an upgrade, fedup is the recommended way to do it.
The term "supported" is a bit problematic for Fedora in general as it's
not like we have a phone line and we don't give out refunds if it
breaks. It's also problematic in the specific case of upgrades, because
there are seventy squillion potential upgrade scenarios and no way we
can possibly test them all. Even with the testing we do, it's almost
inevitable that *some* upgrade attempts will turn out badly.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net