Re: Lenya and Maven

On 7/16/2010 3:34 AM, Andreas Hartmann wrote:
> Am 16.07.10 09:55, schrieb Vik Tara:
>> On 07/16/2010 08:53 AM, Vik Tara wrote:
>>>>
>>>> WDYT ?
>>>
>>>
>>> Makes sense. We could do something very quick and dirty :), as maven
>>> can just call ant tasks.
>>>
>>> So as a first step perhaps just make lenya build from maven by calling
>>> the exsiting ant tasks - and then replace them one by one?
>>>
>>> This way everything will continue to work ok.
>>
>> A pause for thought though...
>>
>> Should we do this right now - or should we do it in the future lenya
>> version.
>>
>> We may spend a fair bit of time doing this now - time which we could
>> spend making new modules etc.
>
> I totally agree. Fundamental changes to the build process of a stable 
> app might be not such a good idea. There are many risks of breaking 
> things, and you introduce a barrier for people who are familiar with 
> the current build process.
>
>> What are the benefits of doing it now?
>
> * Use Maven's dependency mechanism for our module dependencies
> * Get rid of our custom dependency management
> * Reduce the size of the repository
>
> But I agree that we should rather spend our limited resources on code 
> that creates a direct benefit for users.
>
> -- Andreas
>
>
I'm a big fan of maven, but I'm with Andreas on this one. I don't think 
that a switch to Maven on the current branch (2.1.x) is worth the risks. 
As a community, I think we have more urgent needs than switching the 
current version to maven. Here's my list:

1) Fix editor problem (some tasks are remaining)
2) Better documentation (specifically surrounding editor problems)
3) Release 2.1.0
4) Close tickets (over 200 open, some for years, looks very bad)
5) Design V3
6) Implement V3

At the meeting, there was strong consensus that v3 is going to be 
completely different. It will be built from the ground up on JCR and 
Cocoon probably won't be our primary engine. V3 will be built from the 
ground up to use maven and probably OSGi. The issues surrounding 
converting the existing project to maven won't be the same with a fresh 
project.

I think we need to do 1 & 2 above quickly, which would result in 3. We 
should also be working towards 5, so we can do 6. It would be so nice to 
go to Apache Con in November with a design for v3 and to be able to 
encourage people to work on the project.

The 2.2 branch is there to do Cocoon 2.2 and maven. Experimenting there, 
learning there is a good thing and can be a useful set of skills and 
experiences for the future. I am -1 on switching 2.1.x to maven.