George,
I realize my previous message was rather a rant around the
multithreading problem. It's been perhaps more useful to myself in order
to recollect all the pieces of the puzzle.
The bottom line of the message is simple though: yes, we could try work
on the multithreading MPM.
The critical points I can make out are:
- we must be certain Rivet_InitHandler or Rivet_ChildInit are the
calls where thread private Tcl interpreter can be created (sorry for
having mistaken the names in my previous message).
- we have to understand which is the thread exit handler (is
Rivet_ChildExit playing that role?) to clean up a thread's private data.
- global data must be made thread safe protecting them or thread
private
-- Massimo
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
