Re: oneway/Object_release race

[Michael Meeks <michael ximian com>]

Hi Jules,

On Mon, 2005-03-14 at 13:40 +0100, Jules Colding wrote:
> The problem is that although I am seeing the oneway GIOP data in my
> debug dumps it never hits the wire. Putting a 1 second delay in between
> the method and the release makes the GIOP data being truly send:
> 
> 	BRUTUS_IMAPISession_Destroy(mapi_session, ev); // one-way
> 	sleep(1);
> 	CORBA_Object_release(mapi_session, ev);

	Wow - that's hideous,

	No idea why that is; quite possibly the connection is not keeping a ref
on itself while there is pending data - should be fairly easy to fix I
guess.

	Of course, mostly people don't see that because the connections are
long lived.

	HTH,

		Michael.

-- 
 michael meeks novell com  <><, Pseudo Engineer, itinerant idiot