readiness of 10.2

[Rick Hillegas <Ri...@Sun.COM>]

I do not believe we have reached consensus on the readiness of 10.2. 
There seem to be two broad groupings of opinion:

1) Some people believe that 10.2 is ready for release. These people 
argue that we should downgrade the urgency of the remaining issues which 
block us from building a release candidate. These people also argue that 
we will get faster feedback by releasing 10.2 than by appealing to the 
user community for more extensive testing.

2) Other people believe that 10.2 is still too green. These people would 
like to see the user community test 10.2 against production applications 
for three weeks in order to shake out instabilities. These people are 
particularly concerned about optimizer-induced instabilities affecting 
the performance and correctness of query plans.

I ask for the community's guidance on how we can reconcile these 
perspectives.

Thanks,
-Rick

Re: readiness of 10.2

["Jean T. Anderson" <jt...@bristowhill.com>]

Rick Hillegas wrote:
> I do not believe we have reached consensus on the readiness of 10.2.
> There seem to be two broad groupings of opinion:

I'm in the minority, perhaps, with a third opinion:

3) 10.2 is ready to produce a release candidate once the issues you
(Rick) posted earlier today are taken care of. But it would still need
to be tested and voted on.

A release candidate without any jdk16 license issues is required for #1
below and is helpful to #2. How many people want to test a snapshot with
known license issues at this point? The problems have been so well
publicized now on derby-user that if I were an end user I'd stay well
away from that snapshot, wouldn't touch it with a ten foot pole.

> 1) Some people believe that 10.2 is ready for release. These people
> argue that we should downgrade the urgency of the remaining issues which
> block us from building a release candidate. These people also argue that
> we will get faster feedback by releasing 10.2 than by appealing to the
> user community for more extensive testing.
> 
> 2) Other people believe that 10.2 is still too green. These people would
> like to see the user community test 10.2 against production applications
> for three weeks in order to shake out instabilities. These people are
> particularly concerned about optimizer-induced instabilities affecting
> the performance and correctness of query plans.

We have to balance what we would *like* to see the community test with
what the community will be *willing* to test. We can ask, but can't require.

 -jean


> I ask for the community's guidance on how we can reconcile these
> perspectives.
> 
> Thanks,
> -Rick

Re: readiness of 10.2

[Daniel John Debrunner <dj...@apache.org>]

Rick Hillegas wrote:
> I do not believe we have reached consensus on the readiness of 10.2.
> There seem to be two broad groupings of opinion:
> 
> 1) Some people believe that 10.2 is ready for release. These people
> argue that we should downgrade the urgency of the remaining issues which
> block us from building a release candidate. These people also argue that
> we will get faster feedback by releasing 10.2 than by appealing to the
> user community for more extensive testing.
> 
> 2) Other people believe that 10.2 is still too green. These people would
> like to see the user community test 10.2 against production applications
> for three weeks in order to shake out instabilities. These people are
> particularly concerned about optimizer-induced instabilities affecting
> the performance and correctness of query plans.
> 
> I ask for the community's guidance on how we can reconcile these
> perspectives.

I think at some point the release manager as the "unquestioned
authority" has to take a stand and say what they as the release manager
want to do. Rick, you can choose either of the opinions* but pick one
and run with it. The community will always have differing opinions, at
some level it's what the ASF is about. It's good to try & reconcile them
but it's not going to happen every time.

Dan.
[*I think Jean's third opinion is really the same as 1 but with more detail]