| Log Message: |
Compile most of Lucene's core code with a single call to gcj. This
permits more cross-class inlining. In particular, the final method
PriorityQueue.size() can inline, which speeds things 7.5% in my
benchmarks. Note that 'final' declarations *do* sometimes make things
faster in GCJ. So let's not remove them all!
|
lucene/java/trunk/src/gcj/Makefile