[haiku-commits] Re: haiku: hrev50603 - src/kits/media headers/os/media
- From: Dario Casalinuovo <b.vitruvio@xxxxxxxxx>
- To: haiku-commits@xxxxxxxxxxxxx
- Date: Thu, 20 Oct 2016 13:50:54 +0200
On Thu, Oct 20, 2016 at 9:14 AM, Axel Dörfler <axeld@xxxxxxxxxxxxxxxx>
wrote:
Am 20/10/2016 um 03:22 schrieb b.vitruvio@xxxxxxxxx:
3588e74aef85: BMediaEventLooper: Return to simple lateness computation
* Remove enqueue_time again.
* Lateness seems generally useless. No way to get something useful
experimentally. The commit noise however was useful, the port wait is
behaving more in accordance with the lateness calculus.
* Fixes #12817 and possibly #12594.
Huh? How can lateness be useless?
The media_kit attempt to solve things by manipulating latency/lateness
values but for what I see it doesn't work.
At the time my test was to add a "snooze(100)" in different parts of the
chain depending on the situation I wanted to simulate :
* Before the enqueue
* After the thread is suspended to wait
* Before the buffer is sent
Trying to give a more meaningful value results always in ever increasing
latency or cracking sound.
And how can you compute lateness without enqueue time?
I'm not saying the enqueue_time doesn't have it's meaning but alone isn't
going to solve anything. As said various times, perhaps it could be solved
by changing completely how latency is handled through nodes, having a
thread activation log and looking at the whole chain behavior once every
buffer reach the final consumer. In any case I might be wrong so feel free
to adjust my errors.
Saluti,
Dario
Other related posts:
