Ted Husted wrote:
>
> Rather than re-invent the wheel, I'd like to start with the Taglibs
> model.
There is no re-invention. I thought all was pretty clear.
>
> Depending on the circumstances, they may propose the tablig to the list
> first and then open a whiteboard directory, or, if the code is solid,
> open a directory first and then start the discussion.
Right - but we have the sandbox - a place to do exactly that - go
work/play/experiment w/o any approval or oversight of the
jakarta-commons committers. And it keeps the mainCVS 'clean' of
experimental stuff.
> I believe whether a Commons committer chooses to open a directory in the
> Sandbox or the main CVS should be a judgment call. Craig felt the
> BeanUtil was solid, and started up in the main CVS without the
> super-majority mentioned in the charter. But since he is already a
> Commons committer, I'm good with that.
1) I thought craig had enough. If not, then it shouldn't be there :)
2) I don't think it's a 'judgement call' as theres nothing to judge.
There is *no* advantage to putting it in a whiteboard directory in the
main CVS, is there? I can think of a few disadvantages :
* confusion about what are supported components to a user.
* makes the jakarta-commons snapshot larger for no benficial reason.
> My thinking on the life cycle is:
>
> (0) Someone has itch to write a package.
Go!
> (1) They bounce it off the Commons list to see how other people feel.
I don't think this is necessary - a great idea, because of the great
feedback and participation it might attract, but not necessary.
> (2) If the feelings are good, they start a whiteboard directory in the
> main CVS (Commons Committer) (e.g. BeanUtil), or in the sandbox CVS
> (other Jakarta Committer) (e.g. Cactus).
No - I think that we all should respect main CVS for what it is - the
place for 'approved' components that are in serious development. They
don't have to be release-ready. Just approved, serious and active.
I believe our only responsibility in the Commons-wide voting is to
decide if a proposal fits our charter, has plausable probability for
success, etc. After that, we get out of the way. If a
Commons-committer wants a role in an approved component, they should
actively participate in that component...
> (3) When the package Committers are ready, they propose the package to
> the list for a subproject vote.
Yes - but 'ready' doesn't require release-ready anything. It just means
that the committers proposing have a plausable story, and plausable
solution, and are committed to working on and supporting the proposed
component. I didn't need to see Craig's code for BeanUtils, nor did I
think it had to be release-ready for me to believe that there was a
point to it, and he (we) would work on it.
> (4) Repeat (3) as a majority vote for subsequent stable releases.
Nope. I believe that once approved as a commons component, only the
people actively working on the component have the ability and, dare I
say it for risk of being Big Brother, right to decide that. Since the
barriers for participation in the components is non-existant (i.e.
trivially easy if you know how to use an editor and 'cvs commit') for
Commons committers, there should be nothing stopping an interested
Commons-committer in playing a role in the development and release of a
component - just get involved.
Anyway, that's my US$0.02 Unlike the whole 'Can we have a copy of Ant'
thread, I hope I was clear :)
Time for coffee...
geir
> "Geir Magnusson Jr." wrote:
> >
> > Ted Husted :
> >
> > > I would also say that, as a Commons Committer, he can open whiteboard
> > > directories in the main CVS now and then call for a vote on the initial
> > > release later.
> >
> > Two issues here:
> >
> > I] Regarding 'a vote on the initial release', I thought the vote would
> > be on making the project into a 'Commons component', not on the
> > releasability of that project.
> >
> > II] Regarding random whiteboard directories under the main CVS :
> >
> > Can we try to avoid that if possible, and consider the
> > jakarta-commons-sandbox CVS as the 'whiteboard' area for
> > non-yet-voted-on component source?
> >
> > There are a few reasons I ask :
> >
> > 1) The jakarta-commons 'main CVS' are the components that we have all
> > agreed are the individual entities that make up Commons. It makes up
> > the parts of the Commons project that is supported code with a charter.
> > Having arbitrary stuff under a top-level whiteboard would be confusing.
> > Of course, each component project can whiteboard to their hearts content
> > under their own section of jakarta-commons.
> >
> > 2) Since we have the sandbox that is equivalent in capability and
> > access, keeping the not-yet-approved components there keeps the sandbox
> > complete as the place for proposed and embryonic projects.
> >
> > 3) This will help keep the size of the jakarta-commons main CVS limited
> > to that of the supported components. I am afraid that the monolithic
> > CVS tree may get ungainly over time as Commons grows, so keeping the
> > main CVS limited to the actual supported components may help in this
> > regard.
> >
> > geir
> >
> > --
> > Geir Magnusson Jr. [EMAIL PROTECTED]
> > Developing for the web? See http://jakarta.apache.org/velocity/
--
Geir Magnusson Jr. [EMAIL PROTECTED]
Developing for the web? See http://jakarta.apache.org/velocity/