I have respun the webrev,
http://cr.opensolaris.org/~padraig/ips-11034-v4/, so that it applies to
the gate.
Padraig
On 10/05/09 07:56, Padraig O'Briain wrote:
On 10/02/09 21:54, [email protected] wrote:
On Tue, Sep 29, 2009 at 07:23:15AM +0100, Padraig O'Briain wrote:
Sorry about that.
Webrev is http://cr.opensolaris.org/~padraig/ips-11034-v3/.
Thanks for the URL. I'm sorry it took me so long to respond. I somehow
missed this message.
This change looks good.
It took me a moment to figure out what the code on 2785 was doing, and
why __do_exit always had to return true. Given the interface for
timeout_add(), that makes sense. I was curious how packagemanager
reduces the number of active threads once you set exiting = True. Is it
that threads that may be running will see this value set and return
from the functions that check exiting? Are there ever any situations
where you'll have a thread that's asleep?
We do not have a way for threads to see the exiting value. At present
we just wait for threads to terminate and when they all have we
terminate the application.
There is a problem with a thread not terminating and
http://defect.opensolaris.org/bz/show_bug.cgi?id=11574 is logged for this.
I would also like to cancel the API operations which are running in a
thread but that is another day's work.
Padraig
Thanks,
-j
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
