Then again, it might also be faster for you the old way (I wasn't able to measure a difference on high-level tests).
-Mike
A small FYI: I noticed that I accidently commited a change several
months ago that reordered doc ids before retrieval. It is now fixed.
I doubt that it would really affect anyone, but it did add a few
allocations, so for doc-retrieval-heavy applications using trunk, you
may want to sync.
