(reposting my email to you here, since I never got a reply)
I was wondering if you could help us get this feature ready for introduction into Maven 2.1.0. We're putting together a release plan for it right now at:
As you can see, things have changed WRT the versioning in Maven's future. The former 2.1 code - on trunk - has been renamed to 3.0-SNAPSHOT since it contains extensive changes, and we've decided to introduce some intermediary releases that will incorporate less risky new features such as parallel downloads that don't fit on the 2.0.x codeline appropriately. We're also going to be putting 2.0.x in end-of-life mode very soon.
I've heard good things about your patch for parallel downloads, but since I'm trying to push to make sure we have adequate documentation and tests for everything new that goes into Maven, we're still coming up short on this feature.
Can you provide us with a small design document just laying out the problem you solved, including scoping (for instance discussion about groupId limitations, and maybe any known issues on CI systems where multiple Maven processes may be sharing a single local repository with large numbers of parallel downloads happening simultaneously) and a short blurb describing your solution.
Also, we need to create some tests that we can use to guard against regressions in the future. Since you know this code the best, it'd be very helpful if you could help us design some test cases that might verify key pieces of the code to make sure nothing is broken (for instance, nothing that might introduce threading problems or file-locking issues).
If you can provide some of this information, it will allow us to put this feature into 2.1.0. Otherwise, I'm not sure whether we have the resource-hours available to grok the code, reconstruct the design decisions you took, and come up with adequate tests.