Hi Brett,

On 06/06/07, Brett Porter <[EMAIL PROTECTED]> wrote:
My recollection would be that some artifacts are filtered out but
their transitive dependencies still needed to be taken into
consideration for version calculations, so this could be the reason
for this.

If an artifact is filtered out then why would it's transitive
dependencies impact upon the resolved artifacts?  The filtering
performed by DefaultArtifactCollector.filterTrail appears to snip the
branch at any filtered artifact.

I think it's probably better to approach this from the angle of the
problem you are trying to solve - is there a reason you think it
should behave differently to how it does at present?

This occurred when trying to add filtering to the
maven-dependency-tree component.  I'd like to specify a scope filter
to the tree builder, which would in turn be passed to the artifact
collector.  I would then expect filtering to take place prior to
resolution listener methods being invoked, otherwise the filter is a
little bit pointless ;)

> Whilst everyone's busily replying to this, here's another
> ResolutionListener question: are the omitForNearer and omitForCycle
> methods intended to be invoked instead of, or after, the
> includeArtifact method?

the answer is "yes" :)

they should be invoked instead of it for the current node if it
should be omitted. But by the same token, the artifact may have been
previously included and are later omitted.

For a given node, it's "instead of".

That makes sense thanks.  I'm just restructuring some of the
maven-dependency-tree code and wanted to make sure I was testing it
correctly.

Just a quick check - are you working on fixing up issues in 2.0.x or
making other improvements and features like the conflict resolution?

The areas I'm working on at the moment are:
- release poms, to allow reproducible builds with version ranges
(MRELEASE-177, MRELEASE-240, MNG-3023 and our pending discussion)
- making version ranges usable (MNG-2994, MNG-2988)
- highest-wins conflict resolution (MNG-612)
- dependency diagnostic tools (MDEP-94 and more features to come)

Given the "Artifact changes" thread, it appears there are a few
people working on, or interested in working on, that area of the code
but that we haven't really found a unified direction yet.

Is this the general refactor of artifact resolving proposed by Jason
for 2.1?  That does sound interesting and seems the way to go,
although my focus is the above issues for 2.0.x since we cannot
justify moving to 2.1 until it's stable.

Cheers,

Mark

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to