Re: Client notification wierdiness
Re: Client notification wierdiness
- Subject: Re: Client notification wierdiness
- From: Doug Wyatt <email@hidden>
- Date: Wed, 1 May 2002 10:19:51 -0700
On Tuesday, April 30, 2002, at 06:05 , Kurt Revis wrote:
I've noted this many times on the list, but I'll say it again: your
notification proc needs to be reentrant. If you call a CoreMIDI
function* from your notification proc, CoreMIDI could call into your
notification proc again while doing its own work. If you don't know
this is happening, the results can be pretty baffling. Could this be
your problem?
I'll just note that this was a bug, fixed for Jaguar -- when waiting for
the Mach message reply from the server, the client's runloop was waiting
in a mode where other runloop sources could fire, including the
notification callback. This breaks the contract of the API being
synchronous as you'd expect it to be (except for I/O, which is done
completely differently).
Doug
--
Doug Wyatt
work: email@hidden (CoreAudio)
personal: email@hidden
http://www.sonosphere.com
"When one does not love too much, one does not love enough."
-- Blaise Pascal
_______________________________________________
coreaudio-api mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/coreaudio-api
Do not post admin requests to the list. They will be ignored.