On Fri, 2012-02-24 at 15:39 +0800, Feng Wei wrote: > Hi David, > I'm appreciated for your comments. UCM really has a long way to go. > > 2012/2/24 David Henningsson <david.henningsson at canonical.com>: > > On 02/21/2012 04:34 AM, Feng Wei wrote: > >> > >> Hi Arun, > >> I'm not clear what should I do to upstream patches. I tested them on > >> ubuntu, so that I must follow what David had done in port structure. > >> In my original mind, I will first upstream to ubuntu, then pulseaudio > >> community. > >> My current patches are maintained in bzr according to ubuntu, I want > >> them to be merged in ubuntu branch. > > > > > > Hi Feng, > > > > Sorry I haven't responded earlier, but I'm not sure what to do about these > > patches either. As I see it there are at least three problems that need to > > be resolved: > > > > * The competing implementation problem: We've had multiple implementations > > posted to the PulseAudio mailinglist, one by Janos and Jaska, one by > > Alejandro and Margarita (probably discontinued?), and one by yourself. It > > would be great if the UCM community could give us a hint on why we should > > choose one over another. > Exactly. Maybe one reason is my patch is the only one to implement the > agreed concepts mapping between PA and alsa-lib. > > The version from Margarita was prior to the ELC-E BOF and is discontinued. However it was the starting point for the other two patches in that it did interface with alsa-lib as intended. I have not seen the patch by Janos and Jaska but I have seen the patch Feng. I could review the latest patches wrt to alsa-lib/ucm but the PA parts would need checked by someone else. > > * The patches are based on an older version of PulseAudio - the one that > > uses the input devices for jack detection. Ubuntu 12.04, as well as > > PulseAudio 2.0, will release with the new kcontrol jack detection interface > > [1]. In essence, your patches do not apply, and should we consider these for > > Ubuntu 12.04 and/or PulseAudio 2.0 we're in quite a hurry. Perhaps it's even > > too late, I don't know. > It's really too late. I'll follow the new jack detection method and > update my patches tho. > > Fwiw, we could still upstream UCM support without waiting for jack support. It was originally done together as it was easy for testing UCM device changes. > > * Verification and testing is difficult, for a variety of reasons (this > > point is not a real blocker like the other two, just a little cumbersome): > > 1) Requires special hardware. I could probably get hold of some hardware > > if that was the only thing keeping me from reviewing it though. > Understandable. A pandaboard is probably easiest to use for testing. David, I could look into getting one shipped to you for testing/review if this helps ? > > 2) Requires special configuration files, which are usually kept private by > > companies. I assume that the files you have used for imx53 and omap are > > public though (where can they be found)? > I hold the configs for linaro release at > https://code.launchpad.net/~b34248/+junk/alsa-lib-1.0.24.1 I'll push to public alsa git today for Panda. I think they are quite stable now. > > 3) I *was* going to complain about the lack of UCM documentation, but it > > seems like running "make doc" in alsa-lib 1.0.25 does create a UCM page > > (called "case interface"), and while it does not look perfect everywhere it > > seems like most things can actually be figured out. > > Regards Liam
