RE: [nebula-dev] CTableTree

["Peter Centgraf" <peter@xxxxxxxxxxxx>]

1. I was not aware that Jeremy was continuing work on CDateTime separately,
and I'd love to get back in sync with that effort.  I must admit that I have
not been a very active maintainer for CDateTime -- I've fixed a few small
issues and kept up with applying patches to CVS, but that's about all.  My
early enthusiasm didn't amount to much.  Nearly all of the code is still
Jeremy's original, so I have no doubt that what he has will be preferable to
what is in CVS right now.  Honestly, I was hoping to have a design
discussion with him (you) anyway.  Perhaps we should setup a time to talk on
the phone or Skype.

2. I sympathize with Jeremy's statements about the cost/benefit tradeoff of
working within Nebula vs. SourceForge.  There is a kind of organizational
inertia imposed by working within the Eclipse Foundation that is as much
psychological as it is real.  When I first signed up as a committer, I
thought that there would be well-established procedures for doing things
like automated builds, tracking IP for contributions, etc.  I didn't want to
rock the boat too much by changing things to suit my own preferences.  I
realize now that this is only true for projects that have been around a long
time, like platform or data tools.  Younger projects like Nebula are
basically independent, so we still have to figure this stuff out for
ourselves.  For example, there's no policy to stop us from requiring Java 5
for our users, except that the committers thought it was a bad idea.  I
think we can overcome this by simply making some choices and acting on them.

3. I agree with Jeremy that there is a large potential for benefit from
joining forces within Nebula, but this hasn't materialized yet.  Here are
some things I think we could do to make Nebula more valuable:

a. Provide guidance (documentation) on how to build and distribute
ready-to-deploy OSGi bundles for new widgets using the PDE build
infrastructure.  The docs from the PDE team have improved, so I think this
won't be too hard.  This would be a "best practices" guide for small
components targeted for SWT- and RCP-based apps.

b. Setup automated builds using a. for current widgets.

c. Provide access to builds using an update site.

d. Start some design discussion about low-level widget implementation
choices.  For example, a comparison of the typed and untyped listener
mechanisms of SWT for building Composite widgets, approaches for managing
both a visual containment hierarchy and a class inheritance hierarchy in a
complex widget, and tips for more extensible designs for custom widgets.
This is the sort of thing you can't get from working in isolation.

Fellow committers, what value do you expect and hope for by working under
Nebula?

--
Peter