incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Niclas Hedhman <>
Subject Re: [policy] incubating projects and maven repositories v1.0
Date Mon, 28 Aug 2006 04:44:55 GMT
On Monday 28 August 2006 11:31, Jason van Zyl wrote:
> > Could that report be made part of the release:prepare and the  
> > release manager
> > had to explicitly approve it??
> How are we supposed to enforce that? And what if they are not using  
> Maven? Say using either the Maven Ant Tasks, or Ivy, or just an http  
> get to get artifacts from a repository.

We are probably misunderstanding each other...
The question that came up was about transitive dependencies, which the user do 
not necessarily check for, and could end up being dependent on incubating 
projects against his/her will. Something that can't happen for snapshots 
(unless you bypass Maven's intended behaviours) 

You said, that one can check the full set of dependencies from a report 
generated by Maven2.

I said, if that report could be output during the release:prepare phase, and 
that if the release:prepare phase would require the release manager to 
approve the use of that dependency tree, then we put the responsibility in 
the hands of the Maven2 user.

You then start talking about 'enforcement'... And I am lost. Enforcing what? 
If the report can be generated, then either your statement above isn't valid, 
or the report is not capable of reporting the dependencies, in which case the 
original statement is not accurate.

I suspect that you are trying to find problems with non-Maven systems, but 
that can always happen and not the issue at hand. BuildSystemAbc could pull 
down all kinds of stuff for the users, including snapshots, pirated software 
and virii. IMHO, Maven repositories exist mainly to support Maven and 
Maven-compatible(!) build systems.

Your suggestions in the original mail is very good, and *I* don't have any 
opinion about whether a separate Incubating repository is needed or not. Both 
arguments for and against sound reasonable.



To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message