newsAdvanced Gtk+ Sequencer - News: Revising AgsThread and AgsMainLoop

Latest News
The new dispatcher: ags_channel_recursive_run_stage() posted by jkraehemann, Sun 18 Mar 2018 05:37:45 PM UTC - 0 replies
GSequencer development insights posted by jkraehemann, Sun 04 Mar 2018 02:15:03 PM UTC - 0 replies
GSequencer v2.0.0 - forecast posted by jkraehemann, Tue 06 Feb 2018 03:04:57 PM UTC - 0 replies
Conservative multi-threaded setup posted by jkraehemann, Wed 17 Jan 2018 04:22:14 AM UTC - 0 replies
Goals of GSequencer 1.4.x posted by jkraehemann, Fri 05 Jan 2018 01:43:07 AM UTC - 0 replies
[Submit News]
[55 news in archive]

Revising AgsThread and AgsMainLoop

Item posted by Joël Krähemann <jkraehemann> on Wed 17 Aug 2016 05:59:48 PM UTC.

I just have added some prototypes to have grained control over program flow. Since GMainContext might have undetermined amount of time wasting on poll(). Advanced Gtk+ Sequencer suffers throughput problems.

Audio output gets delayed and distorted because of that. The way to try to solve it is by suspend AgsGuiThread. AgsGuiThread does something like following and polls in place:

main_context = g_main_context_default();



Now, following functions and signals has been added:

  • ags_main_loop_interrupt()
  • ags_main_loop_monitor()
  • ags_thread_interrupted()

AgsThread::suspend and AgsThread::resume exists for quite a long time but have been forgotten. They get improved, yet. Thus AGS_THREAD_INTERRUPTED and AGS_THREAD_MONITORING sync flags has been introduced. In order to get most out of nanosleep() ags_main_loop_monitor() might alter time_spent. After callbacked by ags_thread_interrupted() from ags_thread_resume().

So this is what basically happens within AgsGuiThread.

/* avoid exceeding time_cycle */

/* soundcard has to start new playback */

/* AgsGuiThread gets notified about next tic and is allowed to recover as */
/* thread resumes just at the next cycle. */



No messages in Revising AgsThread and AgsMainLoop


Back to the top

Powered by Savane 3.1-cleanup1