ipc pickle

[Janez Jere]

> 
> # results:
> 0.518367439587 (your)
> 0.228796436055 (mine)
> -------------------------------------------------------
> 
> But, your version seems reasonably fast already :). My "Full-Blown
> RPCServer" that uses dedicated threads for function calls (so that one
> RPC call can issue another RPC call while not blocking other threads
> making their RPC calls etc) is *a lot* slower...
> 
> 
As you noticed my code is not a real word example, just a test of small
serialization part of evolving IPC mechanism, current performance is not
too bad, so I could use the code. But every overhead is annoying, and
you just showed how to cut half of the overhead. And is is always a nice
time when losing half of the overhead :)

Janez