[Scooby] Re: [Cosmo] A couple big ideas: cosmo release and server project merge

[Brian Moseley]

On 2/2/06, Ted Leung <twl at osafoundation.org> wrote:

> The rationale that I understood (before today) for having two
> projects includes:
> 1. Scooby could sit atop any CalDAV server not just Cosmo

this might be true someday - we hope it to be true someday - but it's
not in the short term. and the more we think about it, the more
synergy there will be in the long term between the two. i think there
will always be a useful subset of functionality when scooby runs
against something other than cosmo, but we can do a lot more wonderful
things when we know the two are being run together.

> 2. People who are looking for a CalDAV server might  be interested in
> Cosmo but not Scooby

a unified server project doesn't imply that one would be forced to
download and use both apps. as subprojects, both cosmo and scooby
would continue to have their own deployment artifacts (scooby.war and
cosmo.war) and supporting documentation.

> Arguments that I heard (today) in favor of having one project:
> 1. We want ease of installation - that kind of implies one project,
> although I don't think it requires it.

you're right, but if ease of installation was the only criteria, then
what we have today in snarf would suffice.

> 2. There are some pieces of code that might be in the margin in
> between Cosmo and Scooby, which might have to be a separate project.
> Which I think is what the background idea behind Snarf was.  But
> Snarf as originally presented was mostly about build configurations,
> at least until bcm and Bobby started bringing up these other
> management components.

we've never talked explicitly about what might be in an integrated
server bundle, but i've always had ideas along the lines of a unified
management dashboard and jmx schema (or whatever they call it). and
the way i see jackrabbit evolving, it's possible that we'll eventually
need an app server plugin or two as well. there's a lot of room for
ideas.

> Are we going to merge product planning as well as merging the code?
> If we just have one project, does it make sense to keep the two
> pieces on the same release schedule?  If not, then doesn't that mean
> we have two projects?   One problem that I see with merging the two
> projects is that integrated projects take longer to ship, since you
> have to wait for everybody to be ready.  I'd hate to see the pace of
> Cosmo development/releases slow down (which I think negatively
> impacts Chandler) in order to keep in sync with Scooby.

as i mentioned in my previous message, i think we'd make a new release
of the integrated server whenever one of the subprojects has a
release. no need to wait around for any other subproject, just include
the most recent stable version.

and fwiw, slowing down cosmo development is not in my playbook :) if
anything, i want to release cosmo much more frequently. i don't see
the project unification as having any impact on that.