[10:29] hello [10:30] scott is not there... [10:45] Give it two to three hours. [13:01] astraljava, thanks [13:38] No worries, but he has exactly 8 minutes to show up, otherwise I wasn't of much use. :) [16:34] ttoine: What's on your mind? [16:55] zequence, just would like to know if he has some news from twitter [17:06] ttoine: Ah [22:06] zequence, I have a hack that lets jack start every time... [22:10] In the qjackctl setup->options->execute script on startup put : [22:10] pasuspender sleep 5 & [22:45] Probably even 1 or 2 would work instead of 5. [22:47] Anyway, For testing I have audacious streaming through pulse to hw:1 (the same IF as jack uses) [22:49] I tell qjackctl to start Jack and there is a quick stop start of audio and pulse switches from direct to the device to jack_sink. [22:49] zequence, I am not sure, But I think pasuspender uses dbus to tell pulse to release the device. [22:50] That would be the same as what Jackdbus does on start up. [22:51] The only difference I can see is that there is more time from PA giving up the device to jack trying to grab it. [23:15] len-1304: pasuspender will make pulse let go of the card, so that enables jack to grab it. Then, when pulse starts again, jack already has the card, right? [23:17] A hack, yes. But I guess we could consider adding it, if there's no fix [23:20] Well my guess is that pasuspender uses dbus to tell pulse to let go of the devices. [23:21] Just the same as jackdbus does. [23:21] The only difference is the amount of time involved. [23:22] I am not that good with c++ which is what jack2 is [23:22] But it should not be hard to come up with a patch for jack2 to fix it [23:23] add a half second delay [23:24] Anyway, I am posting this to the list and a copy to David H, who may at least know if my guess about how pasusupender works is right [23:24] That way I would know if I should bug against PA or jack. [23:26] zequence, BTW my 12.04 install seems to have stopped working too. So something has been updated that causes it. [23:26] Not pulse because it still uses the old version of pulse [23:29] I have no idea about how dbus works, but the way I'd imagine the thing should work: 1. jackdbus tells pulse to give up the card 2. pulse releases it 3. pulse tells jack it's now release 4. jack takes it [23:30] I don't think that happens [23:30] At least I'm imagining that jackdbus is either getting a false response, or ignoring it [23:31] I think dbus just returns an ack that the message got through [23:31] Not that it has been acted on. [23:34] If jack is not able to monitor when pulse let go of the card, it doesn't know when it's ok to grab it [23:34] You'd want some kind of return when the card was released [23:34] I think it assumes an ack from dbus means go. [23:35] an ack that the message was sent is not the same as that the process finished [23:35] Yes. [23:35] I agree [23:35] For some reason it does work when PA is not playing something though [23:36] Not always for me. [23:36] I have had the same problem after a fresh login [23:36] Well, I can only guess, since I don't know anything about it. Does sound very undeterministic anyhow [23:37] Like, it'll work, if pulse was not too busy [23:37] or soemthing [23:37] Once David answers I will put a bug in. [23:38] I think I will download the src of both and see if I can find anything. [23:39] len-1304: You have any experience in searching in code? I'm thinking I need to get a bit more professional with that. Use tools [23:39] I have done a bit, but I am more used to c than c++ [23:40] len-1304: You'll find the reserve code easily in the dbus folder of jackd2. The two files authored by Poettering [23:40] OK thanks.