On Sep 4, 2007, at 5:03 AM, Brett Porter wrote:
- I would like to see changes done entirely within the artifact trunk in a self-contained way, and Maven updated as it stabilises - not a constant chasing of snapshots (or even alphas), especially with the troubles we've seen with the similar plexus changes recently. We should expect to see a stable, tested, documented release of artifact 3.0 as a result of this proposal well in advance of Maven 2.1.
One interesting side-effect of this approach would be that we have no real clients for this library, at least not at first. This is dangerous in a way, because it leaves open the possibility that any APIs designed could be very cumbersome to use, and may not fulfill all of the requirements. In some ways, this is where maven-artifact* fell down in the first place, not having very many clients (we didn't have many plugins using maven-artifact extensively at the time, for instance)...not having a lot of use cases, we failed to provide easy access to POM and/or transitive dependency metadata for users of the library, forcing them to rebuild the associated MavenProject instances to get at this info...which in turn forced us into caching MavenProject and Artifact instances, to make this process less time consuming.
It means we're going to have to be very diligent in harvesting use cases (and integration tests!) from the existing tools using maven- artifact before we make radical changes to the API...then, adapt these use cases as we refactor.
-john --- John Casey Committer and PMC Member, Apache Maven mail: jdcasey at commonjava dot org blog: http://www.ejlife.net/blogs/john
