Re: repository metadata was: bouncycastle in central

[Brett Porter <br...@apache.org>]

On 01/08/2008, at 3:16 AM, Benjamin Bentmann wrote:

> John Casey wrote:
>
>> To me, all of this points to a dire need to separate dependency  
>> metadata from the POM that all of these derivative artifacts shares.
>
> I could imagine this would also ease long-term interoperability of  
> different Maven versions with the repository. Imagine the day when  
> the POM evolves to the next model version which to my knowledge will  
> prevent any Maven 2.0.x from reading such a POM, failing the build.  
> Those projects that employ Maven 2.1+ will populate the repo with  
> these new POMs. It would be quite frustrating for a Maven 2.0.x user  
> that he cannot use the artifacts built by Maven 2.1+ as dependencies  
> just because the POM format has changed. If the artifact metadata is  
> separated from the POM, the POM could more freely evolve without the  
> risk of breaking consumers of a dependency as long as only the  
> artifact metadata model is compatible.

++1

I mentioned this is one of the challenges in changing the pom format  
earlier this year (http://markmail.org/message/sbouq623fdlujmzt).

I also recall talking about this and my desire for a "repository  
interchange format" when we met at JavaOne 2007 and Kenney raised the  
question of classifiers. Now that the work on artifact is being picked  
up it's a good time to see this added.

One POM, one main artifact, many derivatives is the right way to  
describe the build, but there are plenty of cases to be more specific  
in the information stored about each individual deployed artifact for  
consumption purposes and to not need the tools to understand the whole  
POM construction mechanism.

Cheers,
Brett

--
Brett Porter
brett@apache.org
http://blogs.exist.com/bporter/


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org