Re: Help with core dump

[Bowie Owens <bowie owens csiro au>]

Nick Glencross wrote:

> I shall give that a go in the morning. I have been experimenting with 
> simpler examples, and it looks like as soon as I start submitting the 
> Object pointer to a thread through a queue that things break. I am 
> attaching the example for anyone that would like to see what I've been 
> doing.
>
> What I've done (perhaps incorrectly) is have a method spawn the thread 
> which waits for a task to be submitted (where the task is simply 
> another Object pointer, which is duplicated as I believe that it 
> should be). The thread then simply invokes 'ping' on this Object, 
> which is where it dies.

I've added your code as a test under test/cpp/threading. I've modified 
the names in the IDL to make it a little clearer (to me if no-one else) 
what is going on. I've changed the code from using glibmm to just plain 
old glib. I was worried the problem might be in the mm code but it 
isn't. More importanly I don't want to make orbitcpp depend on glibmm 
(at least not for this test).

I think problem lies in there being a method call in a different thread 
from the run loop. Because if the ping is done in the main thread there 
is no problem. I'm going to post this problem on the orbit-list and see 
if anyone there has some ideas.

--
Bowie Owens

CSIRO Mathematical & Information Sciences
phone  : +61 3 9545 8055
fax    : +61 3 9545 8080
mobile : 0425 729 875
email  : Bowie Owens csiro au