Hmm I see. That could be a bit of a problem for my application given
that aolserver is highly threaded. I’d have to make a pool of
interpreters and supervise access to them to ensure only one thread is
using each interpreter at once…
I’m also starting to look at the work going into the byte compilers -
if I could have a pool of compilers, cache the bytecode and then only
fire up a vm for each request that might be ok.
Any timeline for thread safety?