You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@buildr.apache.org by To...@KE4KUG.net on 2007/11/28 01:21:35 UTC

problem with how poms are located

It appears to me that there is a bug in how an artifact's POM is  
located (maven coordinates).
Using the Maven central repo's testng jars as an example, when an  
artifact has a classifier, the classifier shouldn't be used to locate  
the corresponding POM.
So,
org.testng:testng:jar:jdk15:5.1
but,
org.testng:testng:pom:5.1

I'm seeing a case where (while trying to add transitive dependencies  
to a 'runtime' tarball) buildr is apparently trying to download  
org.testng:testng:pom:jdk15:5.1 and this causes an error since there  
is no such POM in the central repo.

Thanks,
Tommy

Re: problem with how poms are located

Posted by Assaf Arkin <ar...@intalio.com>.
On 11/28/07, Gilles Scokart <gs...@gmail.com> wrote:
>
> Ho, you are also leveraging the maven repository.  Great!
> Did you plan to also make dependency resolution?


There's an experimental dependency resolution, you can do transitive(spec)
and get a list of transitive dependencies.   The idea is to extend this
further to handle them in more cases (e.g. compile.with), add exclusion
patterns, etc.

We are currently doing similar exercice in ivy.  You can maybe found
> interresting lessons learned by searching our mailing list and our jira.
> You might also find some interresting discussion in the repository mailing
> list.


Thanks for the tip.  I'm subscribing to the mailing list, perhaps we can
share tips with each other.

Assaf


I don't have the full list of problems we have met (or we still have), but
> here are a few out of my head :
> - Classifier managment
> - Dependencies herited from parent pom
> - DependencyManagment
> - Relocation
> - Finding the source artefact
> - Handling of profiles
> - Handling of variables
>
>
>
> Gilles
>
>
> 2007/11/28, Tommy@ke4kug.net <To...@ke4kug.net>:
> >
> > It appears to me that there is a bug in how an artifact's POM is
> > located (maven coordinates).
> > Using the Maven central repo's testng jars as an example, when an
> > artifact has a classifier, the classifier shouldn't be used to locate
> > the corresponding POM.
> > So,
> > org.testng:testng:jar:jdk15:5.1
> > but,
> > org.testng:testng:pom:5.1
> >
> > I'm seeing a case where (while trying to add transitive dependencies
> > to a 'runtime' tarball) buildr is apparently trying to download
> > org.testng:testng:pom:jdk15:5.1 and this causes an error since there
> > is no such POM in the central repo.
> >
> > Thanks,
> > Tommy
> >
>

Re: problem with how poms are located

Posted by Gilles Scokart <gs...@gmail.com>.
Ho, you are also leveraging the maven repository.  Great!
Did you plan to also make dependency resolution?

We are currently doing similar exercice in ivy.  You can maybe found
interresting lessons learned by searching our mailing list and our jira.
You might also find some interresting discussion in the repository mailing
list.

I don't have the full list of problems we have met (or we still have), but
here are a few out of my head :
- Classifier managment
- Dependencies herited from parent pom
- DependencyManagment
- Relocation
- Finding the source artefact
- Handling of profiles
- Handling of variables



Gilles


2007/11/28, Tommy@ke4kug.net <To...@ke4kug.net>:
>
> It appears to me that there is a bug in how an artifact's POM is
> located (maven coordinates).
> Using the Maven central repo's testng jars as an example, when an
> artifact has a classifier, the classifier shouldn't be used to locate
> the corresponding POM.
> So,
> org.testng:testng:jar:jdk15:5.1
> but,
> org.testng:testng:pom:5.1
>
> I'm seeing a case where (while trying to add transitive dependencies
> to a 'runtime' tarball) buildr is apparently trying to download
> org.testng:testng:pom:jdk15:5.1 and this causes an error since there
> is no such POM in the central repo.
>
> Thanks,
> Tommy
>

Re: problem with how poms are located

Posted by Assaf Arkin <ar...@intalio.com>.
Indeed.  I just fixed this in the trunk.

Assaf

On 11/27/07, Tommy@ke4kug.net <To...@ke4kug.net> wrote:
>
> It appears to me that there is a bug in how an artifact's POM is
> located (maven coordinates).
> Using the Maven central repo's testng jars as an example, when an
> artifact has a classifier, the classifier shouldn't be used to locate
> the corresponding POM.
> So,
> org.testng:testng:jar:jdk15:5.1
> but,
> org.testng:testng:pom:5.1
>
> I'm seeing a case where (while trying to add transitive dependencies
> to a 'runtime' tarball) buildr is apparently trying to download
> org.testng:testng:pom:jdk15:5.1 and this causes an error since there
> is no such POM in the central repo.
>
> Thanks,
> Tommy
>



-- 
CTO, Intalio
http://www.intalio.com