Re: [aether-dev] Plugin mechanism for the ConflictResolver?

[Benjamin Bentmann <bentmann@xxxxxxxxxxxx>]

Hervé BOUTEMY wrote:

> this is a question for Aether :)
> [...]
> Le samedi 18 octobre 2014 11:23:28 Christofer Dutz a écrit :
> > While looking for the answer to your question, I noticed that the
> > ConflictResolver seems to be using the JavaScopeDeriver which is of type
> > ScopeDeriver to implement the scope derivation. It seems the
> > JavaScopeDerives is the only implementation of ScopeDeriver. It seems that
> > this is all part of Aether. So let me rephrase my question:
> >
> > Would it be possible to add additional ScopeResolvers to Aether?

I don't quite understand how this is "a question for Aether". There's an 
extension point in Aether called ScopeDeriver and it's the 
responsibility/power of the application to drop in whatever 
implementation suits its specific needs when setting up the 
RepositorySystemSession.

> > Le samedi 18 octobre 2014 08:52:34 Christofer Dutz a écrit :
> > > Im currently struggling with some changes introduced with Maven 3.1
> > > explicitly regarding the ConflictResolver in conjunction with custom
> > > scopes. I know that these are generally not supported, but they had
> > > been working nicely for years. With Maven 3.1 it seems the scope of
> > > transitive dependencies of custom scope artifacts are all flattened to
> > > runtime.

It would be nice to share a runnable and self-contained example project 
demonstrating the issue (like the one attached I used for investigation).

AFAICT, the behavior seen with Maven 3.0.x is due to a bug in handling 
dependency cycles, Maven 3.1.x+ fixed this, bringing the 
derived/inherited scope back in line with what Maven 2.x produces. So 
when custom scopes "had been working nicely for years", I'm curious to 
know how exactly Flexmojos worked with Maven 2.x.


Benjamin

Attachment: flex-test.zip
Description: Zip archive