cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject Re: [Vote] Build infrastructure
Date Mon, 29 Sep 2003 16:14:06 GMT

On Monday, Sep 29, 2003, at 14:29 Europe/Rome, Geoff Howard wrote:

> Jeff Turner wrote:
>> On Mon, Sep 29, 2003 at 12:50:57PM +0200, Stefano Mazzocchi wrote:
>> ...
>>> At the same time, the Maven vs. Centipede debate is a human one, 
>>> technology is something that can easily be changed, personal 
>>> feelings aren't. There is friction between the people behind Maven 
>>> and the people behind Centipede.
>> I think it goes deeper than people friction.  Maven tries to "embrace 
>> and
>> extend" Ant/Gump/Velocity/JJAR/Avalon/The World, whereas Centipede 
>> tried
>> to work with them.
>
> I know none of the people involved in either, have no opinion 
> technically or philosophically about either, and consider myself 
> completely neutral in this aspect of the debate.  In short, I am 
> Switzerland.  So,
>
>> So personally (everyone gets 2c to spend, right?), I don't think 
>> Cocoon
>> should bet on Maven (philosophical reasons) or Centipede (technical
>> reasons).  Stick with Ant.  Ant 1.6 has this <import> feature 
>> (similar to
>> XSLT import), that is going to make complex build systems like 
>> Cocoon's
>> *much* easier to manage.  Best feature since Ant was invented.
>> ...
>
> I believe this is the best way to go for the following reasons:
> - We have so much new stuff to do the cost/benefit really has to be 
> extraordinary to justify the move.
> - I think the perceived need diminishes _significantly_ with real 
> blocks.
>   a) The blocks build will not be a part of the cocoon build.
>   b) Retrieving blocks will not be a part of the build, but of the 
> deploy process.
> - I think we need to keep the familiarity of ant available for the 
> _blocks_ build process, as that's the one most users may get involved 
> in.  How does Maven or Centipede's market share compare to Ant? (of 
> course no one has numbers on this, but would anyone argue that it's 
> even close?)
> - The core cocoon build system will be less important after blocks as 
> we can resume binary distributions.  If at that point we still see a 
> need, fine.
>
> This really is my VHO.
>
>>> We care about Forrest, we care about Gump. If Maven does good things 
>>> but lacks a few, we should use it *exactly* for that: so that we can 
>>> improve it, build synergies, instead of wasting energies in 
>>> progressing a competition.
>
> And I'll play ball if we decide to go that way too.  However, I see 
> the same argument in favor of ant.  Why waste energies progressing a 
> competition to it?  If it's broken, let's fix it.  If 1.6 is much 
> better, let's help move it along.

Amen, brother.

let's work incrementally. The less changes, the less likely new bugs 
are introduced, the less time we spend time arguing instead of moving 
along.

Let's just move the exising build system along on cocoon-2.2 and start 
from there incrementally, all right?

--
Stefano.


Mime
View raw message