Re: New GIOP timeout patch - please review

[Michael Meeks <michael meeks novell com>]

Hi Jules,

On Fri, 2007-06-22 at 14:18 +0200, Jules Colding wrote:
> I think that I need to crawl down into glib and take a look... My
> current test patch is below as well as my modifications to the echo
> client/server. The echo server must be executed as:

	;-) we're attaching to the wrong main context:

/echo-client --GIOPTimeoutMSEC=30 `cat /tmp/ior` 1
** Message: 0
Adding timeout for 30 milliseconds
Timeout function invoked
** Message: 00
** Message: 1
Printing exception : IDL:omg.org/CORBA/TIMEOUT:1.0
** Message: 2
** Message: 3
** Message: A
** Message: B
** Message: C
** Message: C

	I attach a patch that works for me [ I was missing timeout.h FWIW ]...

	Things to fix:

	* lifecycle, we can't just add timeouts without removing them
	  again if our connection is closed.

	* adding a timeout is quite expensive CPU-wise; we might do 
	  better by simply extending linc-source.c (link_source_prepare)
	  to handle the timeouts & manage how long each connection has
	  been waiting itself: to avoid adding many more sources. 
	  [ though not sure how to get notified of that timeout - check
	  the glib/gmain.c source I guess ].

	HTH,

		Michael.

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