diff options
| -rw-r--r-- | nncmpp.c | 25 |
1 files changed, 18 insertions, 7 deletions
@@ -612,7 +612,7 @@ static struct app_context struct str_map playback_info; ///< Current song info struct poller_timer elapsed_event; ///< Seconds elapsed event - int64_t elapsed_since; ///< Time of the last tick + int64_t elapsed_since; ///< Last tick ts or last elapsed time bool elapsed_poll; ///< Poll MPD for the elapsed time? // TODO: initialize these to -1 @@ -719,7 +719,7 @@ static struct config_schema g_config_settings[] = .comment = "Whether to actively poll MPD for the elapsed time", .type = CONFIG_ITEM_BOOLEAN, .on_change = on_poll_elapsed_time_changed, - .default_ = "off" }, + .default_ = "on" }, {} }; @@ -3424,20 +3424,31 @@ mpd_update_playback_state (void) strv_free (&fields); poller_timer_reset (&g.elapsed_event); - if (g.state == PLAYER_PLAYING) + if (g.state != PLAYER_PLAYING) + g.elapsed_since = -1; + else { int until_next = 1000 - msec_past_second; - // We could make use of "until_next", however this might create - // an intensive busy loop when playback stalls (typically because of - // some network issues). Half a second will work reasonably well. + int elapsed_msec = g.song_elapsed * 1000 + msec_past_second; if (g.elapsed_poll) - until_next = 500; + { + // We may receive an earlier time, this seems to compensate it well; + // in any case MPD polling can't achieve any significant stability + until_next += 100; + // When playback stalls, avoid busy looping with the server + if (elapsed_msec == g.elapsed_since) + until_next = MAX (until_next, 500); + } // Set a timer for when the next round second of playback happens poller_timer_set (&g.elapsed_event, until_next); // Remember when the last round second was, relative to monotonic time g.elapsed_since = clock_msec (CLOCK_BEST) - msec_past_second; + + // In polling mode, we're interested in progress rather than stability + if (g.elapsed_poll) + g.elapsed_since = elapsed_msec; } // The server sends -1 when nothing is being played right now |