On Thu, 2015-10-29 at 14:11 +0100, David Henningsson wrote: > > On 2015-10-29 12:49, Tanu Kaskinen wrote: > > On Tue, 2015-10-27 at 16:09 +0100, David Henningsson wrote: > > > Changes since v1: > > > Â Â Â * Rebased to master (thanks Tanu) > > > Â Â Â * Added pa_device_port_set_preferred_profile > > > Â Â Â * pa_card_set_profile now updates preferred profile, instead > > > Â Â Â Â Â of module-card-restore > > > Â Â Â Â Â (to make Tanu happy - if it was up to me, I probably would have kept > > > Â Â Â Â Â it as it was) > > > > If the preferred_profile semantics are module-dependent, then the > > variable doesn't belong in the core. If you want to make module-card- > > restore or some other module the "owner" of the semantics that are > > implemented in this patch set, I'm fine with that, but then the > > variable needs to move to some non-core API. > > preferred_profile means "the profile that this port prefers", so there's > no module dependency in the semantics. > > But exactly how to determine what profile a port prefers, that's > something that could benefit from being module dependent. You're right, the core isn't really any more suitable place to set the preferred profile than module-card-restore. The problem with module- card-restore is that it should only be about making the card state persistent, so this kind of policy goes beyond its mandate, and the problem with the core is that the core shouldn't really decide this kind of policy. Creating another policy module just for this is probably overkill. If you want, you can move the code back to module- card-restore. I don't really have a preference any more. --Â Tanu
