On Mon, 2015-04-20 at 11:21 +0200, David Henningsson wrote: > > On 2015-04-17 22:26, Tanu Kaskinen wrote: > > If two ports have different default volumes, and module-device-restore > > doesn't have an opinion on the port volume, we should change the > > sink/source volume to the new port's default volume when switching > > ports, instead of keeping the volume at whatever it happened to be > > when the old port was active. > > > > This introduces pa_sink/source_port_changed_hook_data, because I > > wanted to call pa_sink/source_set_volume() only once in > > pa_sink/source_set_port(), instead of calling it also in > > module-device-restore's hook callback. The extended hook data allows > > module-device-restore to communicate to the core that the volume > > should be set to a non-default value. > > Now that pa_device_port has a default volume, then module-device-restore > should just set it for all ports at the card_new + PA_HOOK_EARLY > callback. There is no need for module-device-restore to do anything at > port_changed anymore; and thus no need to add the > pa_*_port_changed_hook_data struct either, unless I'm missing something. module-device-restore should still restore the saved port volume on port changes, because the port default volume is a different thing than the saved port volume. I could make another patch set after this one that adds volume to ports. That would make the default_volume field redundant, and as you explain, the pa_sink_port_changed_hook_data struct would become redundant too. Port volume restoring should then be moved from module-device-restore to module-card-restore. module-device-restore would restore volumes only for sinks without ports. IIRC, we agreed that ports should actually have per-profile volume, so I should add support for multiple volumes to ports? > There's also the point that now volume and mute are handled differently, > which I'm not really happy with. I can add mute and save_mute fields to pa_sink_port_changed_hook_data to make it symmetric. -- Tanu
