On Tue, Jan 28, 2003 at 11:26:44AM -0500, Jason van Zyl wrote:
> > > My initial reaction is that I'd like to see Turbine stay away
> > > from Plexus. Even if it a technically superior implementation
> > > of the Avalon Framework,
> > 
> > plexus is not an "implementation of Avalon Framework", nor is
> > any of the containers hosted @ avalon. I also don't think it
> > smart to get into any kind of "pissing contest" over technically
> > superior container implementations. Fact of the matter is, there
> > is no "container spec" and hence no "container impl", just an
> > implementation of a somewhat too vaguely described concept.
> 
> Never claimed it was 'superior'. It will certainly be different
> but I never attached the 'superior' moniker to it an any point in
> time.

Sorry, the superior moniker was my fault. I should have specified,
as I was thinking at the time, that I don't know which container is
technically superior, I was just trying to illustrate that
'technical superiority' should not be the determining factor in
using Plexus over an Avalon container.

Actually saying that sounds rather stupid, e.g. why not go with the
best implementation, but (with my naive knowledge of Plexus) I'd put
more stock in the Avalon community.

Course maybe Jason's points are correct, that Fortress is a good
stepping stone to 'SuperContainer' but currently not a good option
for the Turbine/Tambora communities.

I could accept that, except when we start bringing non-Avalon
Framework concepts into the picture (e.g. the ServiceManager.lookup
change and hint of different lifecycles) it gets a little scary.

It sounds like Plexus embraces and extends the Avalon Framework,
which might not be a bad thing, e.g. if we're getting lots of cool
new features, but how does that affect our relationship with the
Avalon community?

E.g. the granduer plans where all Avalon applications run in one
container, their is a single component repository to share amongst
Turbine/James/Cocoon, someone who picks up Turbine can quickly
understand how James/Cocoon operate because their based on the same
container/meta-data/etc.?

With the changed interfaces/lifecycles, what is going to keep
Plexus/Plexusized-Fulcrum/Plexusized-Turbine end up being Stratum?

> How long has Avalon been working to release something? There was
> certainly a business motivation because you can't get squat to build
> without a _great_ deal of difficulty, hardly any of the examples worked
> and everyone is busy. Not just a business need, but any need. 

This is a really good point and, to me, would be the best reason for
going with Plexus. However, I'd just like to make sure that:

- For one, that using Plexus now doesn't lock out the option of
switching to Merlin/SuperContainer in the future (e.g. the
embrace/extend still seems fishy, though I know little about it),
and

- Plexus is brought in and accepted by the community (and ideally
would have it's own community). Turbine is already stretched thin
with people working on Turbine/Torque/Fulcrum, so without a new
influx of developers, adding Plexus to the mix would make things
even more chaotic.

Also note that, and I mean this with all due respect, from my short
time at Turbine, Jason is a revolutionary who likes to pushing out
new code to solve new and interesting problems. This is great and
results in some really cool applications, e.g. Maven and Plexus.

But at the same time, we need the support of our evolutionaries,
e.g. Henning and Martin, or bring in some more evolutionaries (which
I think is a great idea to ease the load on Henning and Martin) to
make sure that after Jason has gone on to the next new project
(which is perfectly fine, I enjoy doing the same thing, only,
unfortunately, to a more frivolous extent, which is an attribute I'm
working on), we have people here in Turbine willing to maintain
Plexus, assuming it is donated.

- Stephen
 

--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to