My blog posts seem to be running in the same vein.  More test cleaning up.  Today I tracked down tests that leak groups, version tags and queued emails.

After doing that, I thought I'd try prove -j, which runs tests in parallel.  With 255 files, 12K tests, the change in test time with using -j 2 was roughly 50%.

There was one test that failed by running in parallel, AdSpace/Ad.t, but it was caused by two tests creating AdSpaces with the same name.  A quick change fixed that right up.

I'm sure that on other systems, that there would be other failures, but we'll work them out.  Cutting 3-4 minutes out of the test time will be worth it.