Re: simple way to register timeout functions in ORBit's-mainloop ?
- From: Michael Meeks <michael ximian com>
- To: Michael Haubenwallner <michael haubenwallner salomon at>
- Cc: orbit <orbit-list gnome org>
- Subject: Re: simple way to register timeout functions in ORBit's-mainloop ?
- Date: 12 Jun 2003 13:33:35 +0100
Hi Michael,
On Wed, 2003-06-11 at 13:42, Michael Haubenwallner wrote:
> The client calls different servants again and again within one
> transaction, and i have to time out this transaction while no servant
> call is active, to avoid an open transaction if the client dies before
> closing it using another servant.
OK - I think I understand you slightly better.
> I already _have_ installed a timeout function, but with the overhead of
> using g_timeout_source_new/g_source_set_callback/linc_main_get_loop/
> g_main_loop_get_context/g_source_attach/g_source_unref, as found in
> g_timeout_add_full, and it works well.
>
> Just this overhead is what i want to avoid when i say "simple way".
Ok :-) you see - the problem is - that this is not a particularly easy
thing to support while already supporting the complexity of the
threading in the ORB as well.
> And when i read some last messages about hiding the linc-stuff behind
> ORBit, i assume i will run into problem using linc_main_get_loop with
> newer releases of ORBit2.
That is correct - sorry.
> If you intend to hide the glib-stuff too, the "simple way"
> (sth. like ORBit_timeout_add) would be required.
Is it really necessary to connect to the linc loop instead of just the
glib mainloop ? - I though you said that it was not blocking in outgoing
calls ? in which case surely it's sitting in the glib mainloop waiting ?
Regards,
Michael.
--
michael@ximian.com <><, Pseudo Engineer, itinerant idiot
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]