I have to mostly agree with Hawkins on this - these projects should
not be tied to each other's release schedules, even from the
beginning. However, the reality is treating Teiid 6.0.0 as "released"
is a misnomer until Designer 6.0.0 is released (which we should be
very close to doing). Teiid is pretty much useless without at least
*some* version of Designer being available. We want to change that
fact in the future, but that's how things sit today. In other words,
for all practical purposes - and for this very first release only - we
need to treat the Teiid solution as a whole - both Teiid Server and
Designer - as *unreleased* until Designer is released. Once Designer
is out there, it shouldn't matter how often either project releases.
Certainly the Designer release schedule in particular will remain
important to the user for the short term, since Designer is, for now,
the only way to produce the metadata and transformations that the
server consumes. Another way of putting this is, for the near term,
Teiid releases will be irrelevant to customers until a supporting
version of Designer is released or unless they don't depend upon
tooling in the first place. I definitely believe that while the
server relies on the Designer as its sole source of input, the project
needs to be diligent in documenting in its release notes what features
are available and accessible, either due to their support by Designer
or their independence from tooling, and those that are "available" but
not accessible until a supporting version of Designer is released. If
we don't do this, we'll just be creating our FUD within our own
community.
Regarding the connector testing issues John originally referred to, if
there really is a Designer dependency, then that connector should
probably should even be advertised as available until a supporting
version of the Designer is available. My disclaimer is I don't really
understand the particular issue with the connector, so I might be way
off here...
I also want to comment on the potential for a 6.0.1 release of Teiid -
if we do this at all, it should be the last time we every do it.
Again, since this is the first release of both products in the open
source world, this first time might be an exception, but in general we
shouldn't be releasing patches in projects, i.e., there should never
be any third segment to our release versions. If ever, for instance,
there should be a bug in Teiid that from the Designer point of view
would be considered a blocker, then Designer would need to wait for
the next release of Teiid that contains a fix for that bug before
Designer could be released.
BTW, the "aggressive" schedule for Designer 6.1.0 is a bit of a lark.
That date was set a long time ago, and there's pretty much no chance
of us coming anywhere close to it, especially considering the impact
of some of the changes we're planning towards removing the unwanted
dependencies between Teiid and Teiid Designer.
On Apr 23, 2009, at 8:54 AM, Steven Hawkins wrote:
Everyone here is attempting to build a thriving community. It's
just that there are differing opinions about how to do it. The
major issue is whether Teiid "releases" should be mothballed prior
to Designer availability. So far there is no agreement on that.
Van's assertion that Teiid or the Server isn't valuable without a
Designer release is really is really the same line of thought.
The Teiid lead's perspective is that we need to have defined Teiid
releases - in terms of connector api, server api, functionality,
etc. - so that the Designer, connector, or server developers have
something stable to target. To that end a 6.0.0 release of Teiid
does not impede the 6.0 release of Designer, rather it is
essential. Whether it's "usable" on its own only matters to how we
promote its use. Furthermore all changes made based upon Designer
integration problems have been made to both 6.0.1 and 6.1.0. So we
will have the ability to provide a release version of Teiid that is
completely compatible with the Designer near its release. Finally
the Teiid developers have been working constantly with the Designer
developers to hammer out all release issues.
I should have also been more particular about the phase "forward
looking". I mean to say the 7 release will have alternative
metadata mechanisms.