Let me know if anything looks bad (or good!) here; topics needing more time (or less time) etc. I've mostly just laid up the topic suggestions in random order. Also; if you're not attending the full day and depending on something to start at a specific time, let us know, otherwise we reserve the right to e g start with a topic early if the previous topic finished early. Being a topic leader doesn't mean you have to bring slides. It's more about you saying a few sentences in the beginning, knowing what you want the outcome of the session to be, and trying to steer up the conversation if we would get side tracked. Feedback welcome. ============= 09.30 Conference start Reserved time to organize chairs, make sure we have equipment etc 09.45 PulseAudio - vision and mission What do we want PulseAudio to be? What are we working for? How important is it for us to be the standard sound server in desktop/laptop as well as the embedded distributions? Other important use cases for PulseAudio? Are some distributions more important than others? Can we unify our visions for where we see PulseAudio in a few years time? Topic leader: David Henningsson 10.30 Is there any way we can increase the patch review capacity? The pile of unreviewed patches keeps growing. I don't have much ideas for improving this beyond "just work harder", but maybe someone else does... Topic leader: Tanu Kaskinen 11.00 Coffee break 11.15 Routing priority list infrastructure A year ago or even more, Colin Guthrie had some ideas on how to improve the routing infrastructure in PulseAudio, but it is still not (fully) implemented. Until it's there, we're hacking around the current limitations with e g modules such as module-switch-on-port-available, have rerouting problems at suspend/resume, etc. Also, Janos Kovacs and Jaska Uimonen has started implementing some kind of priority/policy infrastructure, which we will also be discussing. Topic leaders: Janos Kovacs, Jaska Uimonen 12.15 Improving low latency behaviour Discussion about how we can make PulseAudio perform better in low latency scenarios, such as VOIP or gaming. In particular, the issue of latency increasing over time (as occasional underruns occur) should be somehow resolved. 12.45 Lunch 13.45 Combining system-wide and per-user modes I think the hardware should be accessed by a system-wide pulseaudio instance. This would finally solve the issue that root/mpd/whatever can't have sound while someone else is logged in. The drawbacks of the system-wide mode can be avoided by having also per-user instances taking care of the per-user stuff. Topic leader: Tanu Kaskinen 14.30 Devices with dynamic capabilities (e g HDMI) HDMI in particular can change its PCM capabilities, i e, maybe you first have one monitor connected that supports 5.1 LPCM, and you then unplug and plug in another one that only supports stereo. How do we update/reprobe/etc PulseAudio accordingly to support this? And how does it relate to the just pushed patches that can add profiles/ports dynamically? Topic leader: David Henningsson 15.15 Improving surround sound behaviour More and more laptops have built in subwoofers; how can we handle/enable this in the best way? Also; can we use jack detection to determine whether a certain profile should be available or not? Topic leader: Arun Raghavan, David Henningsson 16.00 Coffee break 16.15 Separate configuration for hw adaptation, core and policy In the non-desktop world there is often a need to ship different configuration files for different hardware, and also different UIs may need different pulseaudio policy configuration (I think this latter point potentially applies also to the desktop world). This can be solved with some packaging tricks, but I would like to make things easier by having the separation also in upstream. Topic leader: Tanu Kaskinen 17.00 Better drain/underrun reporting behaviour We more than once has got reports that 1) drains take seconds too long to complete 2) underruns are reported before all samples sent have been actually played back, and sometimes a audio glitch can be avoided even when an underrun has been reported to the client. Can we get almost sample accuracy on when we report back that underrun/drain has occurred? Topic leader: David Henningsson 17.30 End of conference -- David Henningsson, Canonical Ltd. https://launchpad.net/~diwic
