From: Andrea A. <aa...@op...> - 2008-08-24 08:32:01
|
Hi, I'm looking into a scalability problem in our current shapefile datastore implementation that prevents fully using the whole CPU(s) once the number of threads hitting the shapefile is > 1. I only had a cursory look, so I may have misunderstood, but it seems the current locking mechanism tells apart reads and writes api wise (ShpFiles class, acquireRead/acquireWrite methods) but internally uses a basic ReentrantLock instead of a ReadWriteLock. The name of the variable (Lock readWriteLock = new ReentrantLock()) seems to suggest the usage of a ReadWriteLock was attempted... what went wrong? Cheers Andrea |