Re: [Evolution-hackers] Deadlock when accessing an in-proc address book (fwd)

[Michael Meeks <michael ximian com>]

Hi Chris,

On Thu, 2003-11-27 at 21:12, Chris Toshok wrote:
> Oh this is interesting - I hadn't thought about what would happen with
> the POA threading policies in the inproc case.  I'm guessing they're
> being ignored and everything is handled on a single thread.
...
> Michael, is there some way we can make the POA policies affect in-proc
> calls as well (if they don't already?)

	In-proc calls should obey the POA threading policies too; cf.
ORBit2/src/orb/poa/poa.c (ORBit_c_stub_invoke) -> (get_c_method) ->
(ORBit_poa_allow_cross_thread_call).

	This should do nice cross-thread / thread instantiating calls etc. via.
the ORBit_small_invoke_stub_n method - at least, I believe so; it's
possible that this area needs work - we have few regression tests here.
I have seen (unrelated) stack traces that suggest it works fine though.

	OTOH. if you have a policy like 'thread per poa / per object' and
you're calling a CORBA call from the thread handling that poa/object
then you'll stick in that thread (clearly).

	It'd be great to get a full trace of all the threads ( thread apply all
backtrace) so we can see what's going on.

	Regards,

		Michael.

-- 
 michael ximian com  <><, Pseudo Engineer, itinerant idiot