On 04.02.2015 12:14, David Henningsson wrote: > > > On 2015-02-04 11:30, Georg Chini wrote: >> On 04.02.2015 09:40, David Henningsson wrote: >>> >>> >>> On 2015-02-03 15:04, Tanu Kaskinen wrote: >>>> On Mon, 2015-02-02 at 16:49 +0100, Georg Chini wrote: >>>>> I think the release notes for 6.0 should be revised to account for >>>>> that. >>>> >>>> Indeed. Before I update the notes, though, I want to get a decision on >>>> whether we will release with your patch to use both backends >>>> side-by-side (probably implies another rc before final release), or >>>> will >>>> we postpone that to the next release. Either is fine by me, but I'd >>>> vote >>>> for releasing with your patch. >>> >>> Hmm. I'm not totally sure about the differences between HF and AG so >>> take this with a grain of salt, but... >> >> The HF role is used when you connect a phone to the computer, in this >> case the phone is the >> initiator of the connection, your microphone and speakers are on the >> computer side. >> The AG role is used when you connect a headset to the computer, then the >> computer initiates the >> audio connection and microphone and speaker are on the headset side. >> >>> >>> It seems to me that it's not extremely unlikely that either of us will >>> step into the other domain in the future, i e, bluez 5 might implement >>> AG audio or PulseAudio might implement a native support for HF. >>> >>> Also, if the AG plugin of Bluez 5 supports RFCOMM/AT commands then >>> we're already partially overlapping, because that's what we use to >>> set/get headset volume and mic gain. If we enable both backends, will >>> that then send AT commands from both backends when we try to set the >>> volume, or...? >> >> No, this will not happen. Only on connection will take place. Usually >> the headset tries to connect to >> UUID 0x111f (HFP profile) first, if that is not available it tries UUID >> 0x1112 (HSP profile). Ofono >> registers UUID 0x111f with bluetooth while pulse registers 0x1112. So >> normally when ofono is >> running (and you have an active modem in your system) the headset will >> connect HFP to ofono and >> pulse will only see the a2dp profile. You can disable the AG role in >> ofono though with a command >> line switch if you prefer to have HSP connected to pulse. > > So while many headsets support both HFP and HSP, there are probably > also some headsets which only support one or the other, right? In > which case it would make sense for both ofono and pulseaudio to > support both HFP and HSP. So; maybe my argument about ofono and > pulseaudio potentially overlapping in the future isn't about AG and HF > then, but rather about HFP and HSP? HSP is the older version, that means there are some headsets around which only support HSP. But as far as I know every headset which supports HFP will also support HSP because HSP is a subset of HFP. And yes, there is a potential overlap, so maybe your idea of just adding a switch "both" is the best way to handle it. I'll send a patch later today, then you can decide which of the patches you prefer. > >>> Hence, instead of removing all backend switching code, maybe we should >>> instead add a switching mode "both" which does what you say. Or >>> potentially replace "auto" with "both", if "auto" now makes no sense. >>> >> >> Fine for me, I can send another patch to implement it if you like. >> >>> Finally, I remember Arun had a strong preference for not enabling the >>> ofono backend by default, so Arun, could you elaborate upon whether >>> that still makes sense given this new information? >>> >>> >> >
