1:18 p.m.
I'm not really interested in criticizing the 6.0 release, but I'm interested in
learning from it and doing better. I think all future releases of Teiid have to have some
tooling that works with it. If it's the previous version of Designer that supports it
with the exception of new features X and Y, then I think that's acceptable
alternative. If it's the forthcoming alternative tooling, that's fine too. I
think it greatly undermines the value of a the release, and is the best way to make a
patch release a probability, but it's an improvement. But I think every release has
to have tooling that works on the day we release or none of them will matter to users.
Can we agree to that?
IMHO, I think we should break the releases apart only when we have the modularity allows
us to do it in a verifiable way. Until we have APIs and SPIs that aren't in flux, how
will we know and communicate to the community which version of our tools work with which
version of Teiid? This seems to me like more work in the long run if we do it without the
technical underpinnings already in place. Let's complete the modularization and then
cut the cord.
Can you expand on what you mean by "the independent nature of Teiid
development"?
~jd
----- "Steven Hawkins" <shawkins(a)redhat.com> wrote:
What seems to be disturbing everyone is what the definition of a
community release is. Ultimately that's up to the community, so kudos
to us for having this discussion. Also, I don't take John's comment
below as inflammatory, rather many of you heard me say the same prior
to open sourcing that notion of a Teiid only initial release was
immaterial from the perspective of most end users (but not connector
developers).
In a perfect world with respect to initial community releases, they
would have been done at the edge of a productization cycle. Our
timing was instead to go open source (or at least make public
announcements) in the middle of a development cycle. The 6.0.0
release of Teiid from an product perspective would be a milestone, but
was done for the community for two reasons. First it was considered
desirable to have a release close to the project announcement date.
The other reason, which is more important, was to establish the
independent nature of Teiid development. We may not all agree with
those points, but there it stands.
To sum up the best parts of this discussion:
Does the Teiid 6.0.0 release matter to most users - Not really.
Will there be synchronization between future Teiid and Teiid Designer
releases - Not really.
Will we be in the habit of patch releases - Not really.
Do we need modularize our software more - Yes. A better separation of
Teiid Designer's usage of Teiid's code, separating out projects for
"large" connectors, soap services, etc.
All need to be explored. We've made the jump from 1 code base to 3,
we can do more if needed.
----- Original Message -----
From: "John Verhaeg" <jverhaeg(a)redhat.com>
To: "John Doyle" <jdoyle(a)redhat.com>
Cc: teiid-designer-dev(a)lists.jboss.org, teiid-dev(a)lists.jboss.org,
"Van Halbert" <vhalbert(a)redhat.com>, "Steven Hawkins"
<shawkins(a)redhat.com>
Sent: Thursday, April 23, 2009 10:36:19 AM GMT -06:00 US/Canada
Central
Subject: Re: [teiid-designer-dev] Re: [teiid-dev] 6.1 Release
On Apr 23, 2009, at 10:29 AM, John Doyle wrote:
>
> I agree with this as long as the predicate is that there is a
> version of tooling (Designer or other) that we have validated to
> work with the Teiid release (excepting the new features in the Teiid
> release as you described above).
>
Agreed. That's why I'm pretty much considering Teiid as a whole as
unreleased, regardless of what version the Teiid Server team is
working on. The difficult part is figuring out a way to convey this
effectively to the community. Backing off a little on what I said
before, it certainly would have been ideal in my mind if the very
first version of Teiid wasn't released until Designer was ready, but
that ship has sailed.
--
John Doyle
Senior Software Engineer
JBoss, a division of Red Hat
Office: 978.392.3916
Email: jdoyle(a)redhat.com