ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Adam Murdoch" <adammurdoch...@yahoo.com>
Subject RE: [VOTE] Ant2 codebase adoption process
Date Wed, 23 Jan 2002 01:44:42 GMT


> -----Original Message-----
> From: costinm@covalent.net [mailto:costinm@covalent.net]
> Sent: Wednesday, 23 January 2002 3:41 AM
> To: Ant Developers List
> Subject: Re: [VOTE] Ant2 codebase adoption process
>
>
> On Wed, 23 Jan 2002, Conor MacNeill wrote:
>
> > The obvious candidates. IMHO are myrmidon and mutant. Other candidates
> > may emerge in the discussion period. I do not believe we should include
> > Ant 1.x as an option but again let us discuss.
>
> I think Ant1.x should be an option ( because that's what I would vote
> for).
>
> I'm not an active ant commiter curently, but if my vote still counts
> it'll be -1 on any candidate that doesn't have a clear backward
> compatibility plan built into the design - i.e it should be able to run
> most existing ant tasks and provide wrappers between the new APIs and
> the old ones ( like xalan compat package for example ).
>
>
> > [+1] Timetable - vote on or shortly after the 6th Feb
> >
> > [-1] Codebase adoption is by majority approval
>
> I don't think the vote should be on a codebase, but broken by
> features/patterns ( allowing to select by vote what's best in
> the 2 codbases ). This would also require the 2 codebases to provide a
> clear list of 'what's different from ant1 and why' - especially in
> build.xml semantic and the task interface.
>

This is an excellent idea.

We would choose the common features from each proposal, and put them
together in a single source tree.  This tree would serve as a reference
point for ant 2 work, and would contain features that ant-dev have reached
consensus on (to some degree).

Proposals would still be spun off this reference tree.  However, they stop
being proposals like "here's how I reckon Ant 2 should work", and start
being feature-based.  Such as, a proposal for a VFS, or how the classloader
hierachy should work, or an extension to the typelib descriptor.

These smaller, better focused, proposals would be far easier to evaluate and
roll back into the reference tree.  This way, Ant 2 moves forward, people
have a reference point to work against, yet are still free to experiment.

To kick start the whole process, it would be excellent if the owners of
mutant and myrmidon were willing to be involved.  That would mean helping
identify and extract common features, and using code from the reference tree
as it becomes available.  This would be an iterative process:

* Pick a feature.

* Model it with an interface (or file format, or whatever - something
shareable).

* Extract an implementation from one or both of the proposals into the
reference tree.

* Refactor the proposals to use the interface, and optionally, the
implementation from the reference tree.

* Repeat.

Peter, Conor, What do you reckon?  It would be a shame for the two proposals
to sail off in their own directions.


Adam


--
To unsubscribe, e-mail:   <mailto:ant-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>


Mime
View raw message