> On 30 Mar 2016, at 18:35, Ken Thomases <k...@codeweavers.com> wrote:
>
> On Mar 30, 2016, at 1:44 AM, Gerriet M. Denkmann <gerr...@mdenkmann.de> wrote:
>>
>> This is done once (OS X 10.10.5):
>> commonSendQueue = dispatch_queue_create( “common socket queue",
>> DISPATCH_QUEUE_SERIAL);
>>
>> Several objects do this quite often:
>>
>> NSDate *dax0 = [ NSDate date ];
>> dispatch_async(commonSendQueue, ^{ @autoreleasepool { [ someOtherObject
>> doSomethingwithDate: dax0]; }});
>>
>> double delay1 = 50 msec
>
> What does this mean? That's not valid code. What’s the "msec" doing in
> there?
Sorry about this confusing pseudo code. The actual code is:
double delay1 = 50 * 1e-3; // 50 msec in units of second
>> dispatch_after(disTim1, commonSendQueue, ^{ @autoreleasepool { [
>> someOtherObject doSomethingwithDate: dax1]; }});
>>
>>
>> someOtherObject does:
>> - (void) doSomethingwithDate: (NSDate *)date
>> {
>> NSTimeInterval inDelay = -date.timeIntervalSinceNow * 1e3;
>> if ( inDelay > 12 ) NSLog(@"%s in-delay %9.3f msec",__FUNCTION__,
>> inDelay);
>> }
>>
>> The in-delay is usually less than a dozen milliseconds.
>>
>> But in the call with dispatch_after (never with dispatch_async) the in-daly
>> is sometimes is 10 seconds - sometimes even 20 or 30 seconds - mostly (but
>> not always) near an integral multiple of 10 seconds.
>>
>> Why this delay of 10 seconds, and how can I find out, what is the cause of
>> it?
>
>> P.S. instead of dispatch_after I tried NSTimer and nanosleep. Same problem.
>
> This may have something to do with App Nap or timer coalescing.
Well, this might explain a shift in a timer doing its stuff by a small amount.
Seems unlikely to explain a delay of several 10s of seconds.
> Is your app otherwise active during those delays?
It is constantly sending messages to someOtherObjects and these
someOtherObjects send back messages.
Each someOtherObject creates (at init) a sequential queue (socketQueue) and two
socket descriptors (socket4FD and socket6FD for ipv4 and ipv6 respectively)
and binds these to some host/port; then does:
dispatch_source_create(DISPATCH_SOURCE_TYPE_WRITE, socket4FD, 0, socketQueue);
same for DISPATCH_SOURCE_TYPE_READ;
then adds dispatch_source_set_event_handlers for each dispatch_source.
Same for socket6FD.
But all activity of these someOtherObjects (which b.t.w. are a modified version
of GCDAsyncUdpSocket) is done on its own sequential queue (socketQueue), which
is ≠ commonSendQueue.
In sending back massages someOtherObject uses the main queue.
So it really should not have anything to do with the delay in commonSendQueue.
I just checked: same problem with 10.11.4.
Kind regards,
Gerriet.
_______________________________________________
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
