Branko Čibej wrote: > > By the way, if the answer turns out to be "no", then we can do the > cryptoapi changes that were discussed in another thread for 1.4; and I > was made aware of a patch to make pools friendlier to long-lived > multithreaded applications which would be a nice-to-have (conceptually) > in the 1.x series.
One makes sense, but the second doesn't. If we are adding functions, great, but I didn't think that was the scope. The conceptual change I'd like to work towards for 1.5.0 is an extention of the max free pool setting. Take this one step further to track max fragmentation of a specific pool before the pool is reaped and reallocated as a single contiguous block.
