| 00:02.45 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 00:22.58 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 00:31.51 | gtjoseph | file: a sorcery object type can't exist in 2 different wizards in the same sorcery instance, correct? I.E. endpoint A in config and endpoint B in memory. |
| 00:35.45 | file | define exist |
| 00:36.07 | gtjoseph | endpoint thinks therefore it is. |
| 00:36.50 | file | it is possible - you can create a hierarchy but the first one to return the endpoint or persist or whatever wins |
| 00:36.53 | gtjoseph | an object type can only have 1 wizard associated at a time correct? |
| 00:37.21 | gtjoseph | that's what I want but I can't figure out how to do it with a config and memory wizard. |
| 00:37.45 | gtjoseph | because when you call ast_sorcery_create you can't specify which wizard to use. |
| 00:38.04 | gtjoseph | i'm sure it works if the wizard is an external one with a backing store |
| 00:38.12 | file | correct - they are given the object in the order they are configured |
| 00:38.54 | gtjoseph | yeah, oh well. You're not going to like my solution then. :) |
| 00:39.23 | file | I don't know what you're working on |
| 00:39.33 | gtjoseph | res_pisip_config_wizard. |
| 00:40.36 | gtjoseph | You can't call ast_sorcery_create on objects created on the fly because the config provider doesn't write to anything. |
| 00:40.43 | file | correct |
| 00:41.16 | gtjoseph | <PROTECTED> |
| 00:41.32 | file | erm |
| 00:41.43 | gtjoseph | it works great because I don't want the obkects persisted anyway |
| 00:42.06 | gtjoseph | i said you wouldn't like it. :) |
| 00:44.53 | gtjoseph | https://gist.github.com/gtjoseph/a29c850763eed1dcf44a |
| 00:46.04 | file | yeahhhhhh |
| 00:47.01 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 00:48.30 | file | heh |
| 00:48.53 | gtjoseph | got an alternative |
| 00:48.55 | gtjoseph | ? |
| 00:49.08 | file | not off the top of my head |
| 00:49.44 | file | what you've done is a hack to get what you need, need to figure out exactly what you need and how it would work in the grand scheme of things |
| 00:50.30 | gtjoseph | yep. i was trying to put the created objects in a memory wizard, hence the first question. |
| 00:51.53 | gtjoseph | maybe a tewak to sorcery itself so tou can reset the default wizard for future creates. |
| 00:52.41 | file | meh - programatically a defined hierarchy that a user has defined is not... great... |
| 00:52.54 | file | what you're really doing is wanting a wizard in there that you control |
| 00:53.52 | file | what would be useful is some manipulation of that (such as adding a wizard in) - then you could do your own that provides all of this |
| 00:54.49 | gtjoseph | the ability to search multiple wizards for an object already exists, correct? |
| 00:54.56 | file | yes |
| 00:55.09 | file | the ability to add your wizard, via code, does not |
| 00:55.16 | gtjoseph | so I just need to tell the instance that endpoints can also come from the memory wizard, correct? |
| 00:55.20 | file | has to currently be defined within the sorcery.conf |
| 00:55.39 | gtjoseph | that's OK. |
| 00:56.00 | gtjoseph | now I just have to get the objects into the memory wizard. |
| 00:56.22 | file | if it's defined in sorcery.conf it'll fall through to the memory wizard |
| 00:56.29 | file | since the config wizard can't do that |
| 00:56.52 | gtjoseph | yep, but I have to get them into the memory wizard somehow, right |
| 00:56.54 | file | but you can't directly access that wizard - which means if something else in between can create.. boom |
| 00:57.13 | gtjoseph | didn;t understand ^^ |
| 00:57.22 | file | I think ultimately you need your own wizard which is read-only like the config one |
| 00:58.24 | gtjoseph | the memory wizard is exactly what I want. |
| 00:58.27 | file | you can only create against a sorcery instance |
| 00:58.33 | file | you can not explicitly specify a wizard |
| 00:58.43 | file | and really, you shouldn't have knowledge of what wizards are in use 'nor should you care |
| 00:59.33 | file | you care because you are using the memory wizard as an in-memory store specifically for your implementation |
| 00:59.54 | gtjoseph | what'sit for if not that? |
| 01:00.29 | file | it's for consumers and users of sorcery, it's not for stuff that actually provides the objects |
| 01:01.20 | gtjoseph | presumably someone's putting objects in it. |
| 01:02.00 | file | sure - but they don't care where it goes |
| 01:02.09 | file | I want to persist an endpoint |
| 01:02.14 | file | that's all they care about |
| 01:02.50 | *** join/#asterisk-dev anthm (~anthm@freeswitch/developer/anthm) |
| 01:03.09 | file | you don't want to persist it - you want to put it in temporary storage for lookup |
| 01:03.40 | gtjoseph | exactly, i don't want to persist my objects either |
| 01:03.55 | file | sorcery doesn't guarantee that won't happen |
| 01:04.59 | gtjoseph | hmmm. I'll write something up for the -dev list. |
| 01:06.05 | file | example: I have a realtime entry configured, then config wizard, then memory. If your code created an endpoint - it would go into realtime. |
| 01:06.42 | gtjoseph | that's where i was saying that a tweak to sorcery could let you reset the default or specify the exact wizard to create in. |
| 01:07.06 | file | but a normal user of sorcery should not know or need to specify that |
| 01:08.12 | file | you're basically wanting a direct API into the memory wizard |
| 01:08.27 | gtjoseph | true, kinda. |
| 01:08.43 | gtjoseph | and I've never been accused of being 'normal' |
| 01:09.04 | file | to manipulate a specific memory wizard instance to what you need |
| 01:11.30 | gtjoseph | well, i suppose I could create a wizard that pulls 'pjsip_wizard' objects from pjsiop.conf and creates the objects instead of having the res_pjsip_wizard module push the objects into the memory wizard |
| 01:11.35 | file | so - what if you were able to programmatically attach an in-memory wizard to the sorcery instance, give it your own handle, and have direct access to it? |
| 01:11.42 | file | yes - that would work fine |
| 01:12.35 | gtjoseph | i'll follow that trail then. |
| 01:13.04 | file | if this becomes a thing I'm inclined to poke at what I just said |
| 01:13.08 | gtjoseph | thanks! |
| 01:13.27 | file | to reduce code duplication potentially |
| 01:13.32 | file | *shrug* |
| 01:13.36 | file | this is me in a half asleep state |
| 01:13.38 | gtjoseph | Give me a day to get something working. |
| 01:13.44 | gtjoseph | then we can chat again. |
| 01:14.02 | file | yup yup |
| 01:20.26 | *** join/#asterisk-dev Dovid (~Dovid@ool-4a584809.dyn.optonline.net) |
| 01:34.52 | *** join/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 01:34.52 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 01:36.27 | mjordan | gtjoseph: ping! |
| 01:39.38 | gtjoseph | mjordan: pong |
| 01:39.44 | gtjoseph | what did I do? |
| 01:41.23 | file | it's a trap |
| 01:42.10 | gtjoseph | the only times he pings me is if I broke something |
| 01:43.17 | file | I don't think you broke anything |
| 01:45.31 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 01:45.45 | file | apparently you broke mjordan |
| 01:45.52 | gtjoseph | :) |
| 02:00.12 | *** join/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 02:00.12 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 02:00.25 | mjordan | yay VPNs |
| 02:00.49 | gtjoseph | mjordan: were you looking for me? |
| 02:00.57 | mjordan | hm. I'm guessing you didn't get my messages. |
| 02:01.11 | file | correct |
| 02:01.11 | mjordan | https://bamboo.asterisk.org/bamboo/download/AST-ATRUNKUNIT-C632UT/build_logs/AST-ATRUNKUNIT-C632UT-804.log |
| 02:01.15 | gtjoseph | eh, no. |
| 02:01.15 | mjordan | sorry :-( |
| 02:01.23 | mjordan | That's what I get for turning on a VPN |
| 02:01.35 | mjordan | unit test crash due to - I think - the semi-colon escaping changes |
| 02:01.45 | gtjoseph | i checked bamboo a while ago... |
| 02:02.06 | gtjoseph | checking again |
| 02:02.12 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 02:02.42 | file | mjordan, also ummm I looked at that issue you just filed... and the event is in the debug? |
| 02:03.13 | mjordan | just one of them is. |
| 02:03.16 | mjordan | there should be two. |
| 02:03.35 | mjordan | you leave the first bridge, enter the second, then don't get another ChannelLeftBridge prior to the StasisEnd |
| 02:03.45 | file | ah |
| 02:04.17 | mjordan | er |
| 02:04.32 | mjordan | yes. |
| 02:04.35 | mjordan | the StasisEnd is premature. |
| 02:04.45 | mjordan | goes back to update the description |
| 02:04.50 | *** part/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 02:05.17 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 02:06.35 | *** join/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 02:06.36 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 02:06.40 | mjordan | fixed |
| 02:07.00 | mjordan | that's much less scary at least. |
| 02:07.23 | file | I'm not... surprised |
| 02:07.55 | mjordan | me neither. |
| 02:10.05 | file | the dog is making it extremely clear that it's time for bed |
| 02:10.06 | file | runs off |
| 02:12.49 | gtjoseph | mjordan: damn, DO_CRASH is turned on for the tests isn't it? |
| 02:12.57 | mjordan | yup |
| 02:13.10 | gtjoseph | ok, simple fix. give me 10 minutes. |
| 02:14.44 | mjordan | np, and sorry for bugging you during the evening |
| 02:15.12 | gtjoseph | no worries. still relatively early for me. besides I just say the bamboo notice show up. :) |
| 02:15.18 | gtjoseph | saw |
| 02:28.11 | mjordan | it is kind of hard to filter out the good bamboo notices from the spam. We need to work on that. |
| 02:39.01 | *** join/#asterisk-dev babak (uid19622@gateway/web/irccloud.com/x-ylhgoszogcpjtblh) |
| 02:41.40 | gtjoseph | mjordan: all fixed 11->trunk |
| 02:55.45 | mjordan | gtjoseph: thanks! |
| 02:55.57 | gtjoseph | no prob |
| 03:06.26 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 03:13.54 | *** join/#asterisk-dev terryw (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 03:17.52 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 03:31.50 | mjordan | file: you around? |
| 03:44.23 | *** join/#asterisk-dev Dovid (~Dovid@ool-2f113961.dyn.optonline.net) |
| 03:57.58 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 04:59.09 | *** join/#asterisk-dev tuxd00d (~tuxd00d@ip70-162-151-19.ph.ph.cox.net) |
| 05:48.10 | *** join/#asterisk-dev Ta^3 (~tacvbo@187.189.162.201) |
| 06:12.16 | *** join/#asterisk-dev [TK]D-Fender (~chatzilla@64.235.216.2) |
| 06:58.09 | *** join/#asterisk-dev oej (~Adium@h87-96-134-129.dynamic.se.alltele.net) |
| 06:58.10 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 07:30.50 | *** join/#asterisk-dev hehol (~hehol@2001:1438:1009:200:d5ae:8941:e8f0:9a21) |
| 08:21.08 | *** join/#asterisk-dev oej (~Adium@h87-96-222-88.dynamic.se.alltele.net) |
| 08:21.08 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 09:29.59 | skrusty | morning |
| 11:14.39 | *** join/#asterisk-dev bulkorok (~Benjamin@gw1.pinguin.ag) |
| 11:15.39 | *** join/#asterisk-dev tzafrir (~tzafrir@local.xorcom.com) |
| 11:15.40 | *** mode/#asterisk-dev [+o tzafrir] by ChanServ |
| 11:18.47 | *** join/#asterisk-dev oej (~Adium@h87-96-222-88.dynamic.se.alltele.net) |
| 11:18.47 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 12:35.39 | *** join/#asterisk-dev oej (~Adium@h87-96-222-87.dynamic.se.alltele.net) |
| 12:35.39 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 12:39.55 | *** join/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 12:39.55 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 12:40.04 | file | mjordan, you rang last night? |
| 13:07.53 | *** join/#asterisk-dev bulkorok (~Benjamin@89-96-153-211.ip13.fastwebnet.it) |
| 13:07.53 | *** join/#asterisk-dev [TK]D-Fender (~chatzilla@216-191-106-163.dedicated.allstream.net) |
| 13:12.51 | mjordan | file: yup, sorry about that |
| 13:13.01 | mjordan | file: I think I know what is going on with the SIP directmedia fax tests |
| 13:13.25 | mjordan | when we determine the glue type we need from the participants, it goes out and determines the type of bridge it wants, not what type of bridge it has right now |
| 13:14.05 | mjordan | In the case of T.38, that ends up being kind of a bad thing when stopping a bridge. Prior to getting the T.38 re-INVITE, we were actually in a directmedia bridge. The T.38 re-INVITE flips things to local. |
| 13:14.32 | mjordan | When we stop the bridge, we ask the channels, and they reply back with 'local' - which is correct - but wrong from the perspective that we need to still issue a re-INVITE for the audio to come back to Asterisk |
| 13:14.56 | mjordan | I think this oddly used to work simply due to quirks |
| 13:15.14 | mjordan | the act of updating the peer during a bridge stop *will* cause the T.38 code in chan_sip to trigger the right re-INVITE |
| 13:15.33 | mjordan | but we have to tell it still to do it, and currently we don't due to the glue calculation deciding we don't need to |
| 13:16.10 | mjordan | I'm contemplating keeping things like they are, but at least noting the "previous" state of the bridge in a pvt in bridge_native_rtp |
| 13:16.40 | mjordan | or |
| 13:16.45 | mjordan | since that might get confusing |
| 13:16.59 | mjordan | it might be worth always telling the peers to go update themselves when stopping the bridge. |
| 13:17.05 | mjordan | I'm not sure how harmful that really would be |
| 13:18.21 | skrusty | can you dial a sip uri using ARI? endpoint: {"SIP/someuri@somedomain.com"} for example? |
| 13:18.55 | mjordan | skrusty: yes. It's just a dial string. |
| 13:18.59 | skrusty | mjordan: oh btw - the async agi issue was my bad! :) |
| 13:19.04 | mjordan | np |
| 13:19.35 | skrusty | hmm, ok - just trying it now but i was getting an error |
| 13:19.44 | skrusty | im using chan_sip, i assume that makes no diff? |
| 13:20.00 | mjordan | It shouldn't. |
| 13:20.11 | mjordan | I'd make sure you can Dial() whatever you're trying to use first |
| 13:20.17 | mjordan | if that works, it *should* work in ARI. |
| 13:21.01 | skrusty | hmm, good thinking |
| 13:23.34 | *** part/#asterisk-dev bulkorok (~Benjamin@89-96-153-211.ip13.fastwebnet.it) |
| 13:25.15 | skrusty | <PROTECTED> |
| 13:25.39 | skrusty | only sending channelid and endpoint, so will do as you suggested |
| 13:26.49 | mjordan | that's not a valid request |
| 13:27.46 | mjordan | Hm. Our documentation could be a bit better here. |
| 13:27.58 | skrusty | go on? :) |
| 13:28.02 | mjordan | skrusty: you're basically originating something, which means you have to provide where it goes |
| 13:28.09 | skrusty | oh |
| 13:28.22 | skrusty | i thought i could just create a channel then a bridge etc |
| 13:28.22 | mjordan | you have to provide either the Stasis application for it to drop into, or a Context/Exten/Priority tuple if you want the dialplan |
| 13:28.30 | mjordan | skrusty: sure, but where does the channel go? |
| 13:28.53 | mjordan | skrusty: if you want the channel, tell it to go into your Stasis application. Then you can do as you just said. |
| 13:29.01 | mjordan | but we also allow for people to throw things out into the dialplan |
| 13:29.06 | mjordan | https://wiki.asterisk.org/wiki/display/AST/Asterisk+13+Channels+REST+API#Asterisk13ChannelsRESTAPI-originateWithId |
| 13:29.29 | mjordan | add an app argument (the name of your Stasis application), and optionally your app-args |
| 13:29.35 | mjordan | appArgs |
| 13:29.45 | skrusty | ah. |
| 13:30.19 | *** topic/#asterisk-dev by mjordan -> Asterisk Development Discussion -=- http://www.asterisk.org/developers -=- Tier 2 and 3.14159265 support is in #asterisk -=- Asterisk 13.0.0 now released. Do the release dance! -=- Follow on Twitter at @AsteriskDev |
| 13:30.36 | *** part/#asterisk-dev mjordan (~mjordan@75.76.55.191) |
| 13:37.14 | *** join/#asterisk-dev WIMPy (~wimpy@e183095026.adsl.alicedsl.de) |
| 13:48.24 | *** join/#asterisk-dev Dovid (~Dovid@static-173-63-105-210.nwrknj.fios.verizon.net) |
| 13:54.08 | file | reads up |
| 13:54.30 | file | ah |
| 13:54.32 | file | curse you T.38 |
| 13:55.43 | leifmadsen | file: you love it |
| 13:55.50 | file | leifmadsen, no |
| 13:55.55 | file | leifmadsen, T.38 cost me dinner at PF Chang's |
| 13:56.05 | leifmadsen | fuck you T.38 |
| 14:09.33 | *** join/#asterisk-dev newtonr (~newtonr@173-17-135-67.client.mchsi.com) |
| 14:09.33 | *** mode/#asterisk-dev [+o newtonr] by ChanServ |
| 14:20.38 | *** join/#asterisk-dev infobot (ibot@rikers.org) |
| 14:20.38 | *** topic/#asterisk-dev is Asterisk Development Discussion -=- http://www.asterisk.org/developers -=- Tier 2 and 3.14159265 support is in #asterisk -=- Asterisk 13.0.0 now released. Do the release dance! -=- Follow on Twitter at @AsteriskDev |
| 14:35.03 | malcolmd | file: how's that, the lost dinner? |
| 14:35.16 | file | malcolmd, I SHALL EXPLAIN IN IM |
| 14:35.22 | malcolmd | very good |
| 14:35.45 | WIMPy | You pre-ordered via vax? |
| 14:36.04 | skrusty | is trying to get his head round ARI Originate |
| 14:46.45 | file | skrusty, have any questions/thoughts? |
| 15:00.52 | *** join/#asterisk-dev mjordan (~mjordan@nat/digium/x-jctqrhidjugpdyhn) |
| 15:00.53 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 15:00.54 | skrusty | well, first time i've used it and as yet not really got it working - trying to replicate dial in my app |
| 15:01.40 | file | you can't reproduce Dial exactly using the current ARI Originate |
| 15:02.12 | file | mjordan, so two thoughts - should T.38 stuff actually cause a bridge re-evaluation? |
| 15:02.20 | file | mjordan, well... that's the first |
| 15:02.27 | skrusty | so, i want to create a new channel, then a bridge and then bridge the original channel and new channel |
| 15:02.47 | file | skrusty, you can do that but originate will only execute your stasis app, or dialplan, upon answer |
| 15:02.50 | skrusty | (or add both channels to the bridgte should i say) |
| 15:03.06 | skrusty | right |
| 15:03.12 | skrusty | yes, i see. hmm |
| 15:03.22 | file | so if you try to add it to the bridge immediately it will fail |
| 15:03.55 | skrusty | :/ |
| 15:04.34 | skrusty | ideally i'd like to just create a new channel with and endpoing and channelid, but mjordan pointed out that doesn't work |
| 15:04.41 | skrusty | which i understand |
| 15:05.54 | file | mjordan, second - I guess a bit of state is what we need in our lives! |
| 15:06.01 | mjordan | file: well, if a bridge that is doing T.38 breaks, you have to tell the channel drivers to re-INVITE |
| 15:06.39 | mjordan | file: I'm wondering if we can't just tell the channel drivers unilaterally "oh, hey, you know that native bridge you were in? Go fix yourselves" |
| 15:06.58 | mjordan | the logic for re-INVITEing based on T.38 already properly lives in the T.38 capable channel drivers |
| 15:07.06 | mjordan | so I don't want that to creep into the native RTP code |
| 15:07.26 | file | mjordan, hum |
| 15:07.32 | mjordan | but I can also see things being wonky even without T.38. If, prior to stopping the native RTP bridge, we were in a remote bridge but - when we stop - the channel drivers evaluate to local, we won't re-INVITE back |
| 15:08.35 | file | I sort of regret not breaking the two up |
| 15:08.55 | mjordan | yeah |
| 15:08.57 | mjordan | :-P |
| 15:09.28 | mjordan | if we maintained a bit of bridge state, we could probably make better choices. Generally, when stopping a bridge, you don't really care what the channels want anymore |
| 15:09.34 | mjordan | if you are in a remote bridge, you have to re-INVITE back. |
| 15:09.47 | mjordan | of course, the weird thing is that in a T.38 re-INVITE, you're in a wacky state |
| 15:10.04 | mjordan | and even though you are in a local p2p bridge, you have to treat it as a remote bridge so that you force a re-invite back |
| 15:10.05 | file | yeah - if the conditions which are used to determine what bridge we should be in within bridge_native_rtp changes while in there... all bets are off without state |
| 15:10.13 | mjordan | hence why the easiest solution may be to just always update the peers in a stop |
| 15:10.19 | file | ew yeah... |
| 15:10.45 | mjordan | I suspect prior to your patch the 'fake bridge' was screwing things up so this worked (that is, we always updated the peer on stop) |
| 15:10.50 | file | that should be fine |
| 15:10.59 | mjordan | I'll test it out :-) |
| 15:11.06 | file | well... I say that |
| 15:11.11 | file | it may be fine for chan_sip |
| 15:11.26 | file | WELL - correcting myself |
| 15:11.48 | file | it's fine no matter what, chan_sip may just be a bit smarter about when it sends re-INVITEs so if you update the peers but nothing has really changed it won't send |
| 15:12.13 | file | aha nope chan_pjsip is also smart |
| 15:12.14 | file | cool |
| 15:14.00 | *** join/#asterisk-dev kharwell (~kharwell@nat/digium/x-xcifacwjzkgnljbr) |
| 15:14.40 | file | I should have known better than to question chan_pjsip |
| 15:15.38 | mjordan | heh |
| 15:15.45 | mjordan | well, that apparently fixed *that* test |
| 15:16.17 | mjordan | the question is how many direct media tests did I just break |
| 15:17.13 | file | every other one! |
| 15:24.38 | leifmadsen | only the important ones? |
| 15:29.00 | mjordan | leifmadsen: which is none. Who uses direct media anyway. Crazy people. |
| 15:29.05 | mjordan | p2p media is a fad |
| 15:29.08 | leifmadsen | mjordan: exactly |
| 15:29.14 | leifmadsen | it all makes sense until you try to actually use it |
| 15:29.47 | mjordan | whatev |
| 15:29.49 | mjordan | no one has NATs |
| 15:29.56 | mjordan | IPv6 or bust |
| 15:30.34 | *** join/#asterisk-dev putnopvut (~putnopvut@asterisk/master-of-queues/mmichelson) |
| 15:30.34 | *** mode/#asterisk-dev [+o putnopvut] by ChanServ |
| 15:33.31 | mjordan | file: don't hug me, I'm scared. The chan_sip tests passed. |
| 15:33.43 | file | mjordan, that's because they love you |
| 15:39.14 | *** join/#asterisk-dev pc-m (~pcm@modemcable094.94-70-69.static.videotron.ca) |
| 15:39.25 | skrusty | my productivity just dropped looking at mjordan's tweets! :// |
| 15:40.20 | mjordan | I closed my twitter tab. |
| 15:41.45 | *** join/#asterisk-dev sruffell (~sruffell@asterisk/the-kernel-guy/sruffell) |
| 15:41.45 | *** mode/#asterisk-dev [+o sruffell] by ChanServ |
| 15:42.13 | *** join/#asterisk-dev bulkorok (~Benjamin@gw1.pinguin.ag) |
| 15:42.23 | file | but Twitter is part of what I do for work! sorta! |
| 15:42.58 | skrusty | :] |
| 15:43.28 | file | tweet tweet tweet |
| 15:44.18 | skrusty | now i have the birdy song in my head.... :| |
| 15:44.36 | jameswf | hopes mjordan will discuss dialplan deprecation in the webinar.... |
| 15:47.59 | mjordan | oh lordy |
| 15:48.04 | mjordan | one can only hope |
| 15:49.05 | coreyfarrell | we shouldn't deprecate dialplan.. just delete it :P |
| 15:49.11 | Qwell | mjordan: We (and obviously, by using the word "we", I mean the entire Asterisk development community) don't like dialplan. |
| 15:49.14 | file | webinar? did I miss something? |
| 15:49.37 | mjordan | file: I got co-opted |
| 15:49.49 | mjordan | although I wasn't aware there was an announcement for it yet |
| 15:49.50 | file | mjordan, not gonna lie - I'm laughing :P |
| 15:49.57 | mjordan | you are correct to laugh at me. |
| 15:50.00 | mjordan | For I do. Not. Learn. |
| 15:51.45 | leifmadsen | mjordan: turns out you're not busy enough |
| 15:52.04 | *** join/#asterisk-dev oej (~Adium@h87-96-222-88.dynamic.se.alltele.net) |
| 15:52.05 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 15:56.45 | jameswf | yeah I got the spam a few minutes ago |
| 15:56.49 | *** join/#asterisk-dev oej (~Adium@h87-96-222-88.dynamic.se.alltele.net) |
| 15:56.51 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 15:57.08 | mjordan | jameswf: on behalf of Digium, you're welcome :-) |
| 15:57.32 | *** join/#asterisk-dev sgriepentrog (~sgriepent@nat/digium/x-yvqibgabcgdgewhj) |
| 15:57.40 | jameswf | I like the spam... I learn about all sorts of stuff early... like 13 stable :) |
| 16:15.26 | *** join/#asterisk-dev sgriepentrog (~sgriepent@nat/digium/x-gmepqebgamvpfxgl) |
| 16:15.50 | *** join/#asterisk-dev bweschke (~bweschke@74-94-3-210-Philadelphia.hfc.comcastbusiness.net) |
| 16:24.27 | *** join/#asterisk-dev babak (uid19622@gateway/web/irccloud.com/x-bpsfkdwgayqhxfsm) |
| 16:28.59 | *** join/#asterisk-dev rmudgett (~rmudgett@nat/digium/x-fbijslhlxguiyykw) |
| 16:51.33 | *** join/#asterisk-dev bulkorok (~Benjamin@gw1.pinguin.ag) |
| 16:53.59 | *** join/#asterisk-dev sruffell (~sruffell@asterisk/the-kernel-guy/sruffell) |
| 16:53.59 | *** mode/#asterisk-dev [+o sruffell] by ChanServ |
| 17:06.33 | *** join/#asterisk-dev oej (~Adium@h87-96-134-129.dynamic.se.alltele.net) |
| 17:06.33 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 17:07.50 | gtjoseph | file: can you test a quick sorcery.conf thing for me? |
| 17:08.02 | file | ummmm |
| 17:08.08 | file | maybe? what? |
| 17:08.44 | gtjoseph | i'm getting errors when I add 'endpoint=config,pjsip.conf,criteria=type=endpoint' to the res_pjsip section |
| 17:09.02 | gtjoseph | Could not create an object of type 'endpoint' with id '1001' from configuration file 'pjsip.conf' |
| 17:09.11 | gtjoseph | even though it creates the object |
| 17:09.17 | gtjoseph | this is stock 12 |
| 17:10.51 | file | all fine here |
| 17:11.04 | file | ah |
| 17:11.05 | file | one sec |
| 17:11.07 | gtjoseph | well, that's weird. |
| 17:11.39 | file | I also see that |
| 17:11.42 | file | strange! |
| 17:12.15 | gtjoseph | ok, at least it's not just me. the other modules like res_pjsip_outbound_registration don't have that problem |
| 17:12.43 | file | ah |
| 17:13.03 | file | weird |
| 17:25.28 | *** join/#asterisk-dev otherwiseguy (~otherwise@69.76.221.33) |
| 17:27.23 | *** join/#asterisk-dev anthm (~anthm@freeswitch/developer/anthm) |
| 17:30.46 | *** join/#asterisk-dev newtonr (~newtonr@nat/digium/x-fbfasmhnlhclvhtv) |
| 17:30.46 | *** mode/#asterisk-dev [+o newtonr] by ChanServ |
| 17:38.16 | *** join/#asterisk-dev tuxd00d (~tuxd00d@wsip-98-191-184-213.ph.ph.cox.net) |
| 18:11.30 | gtjoseph | hey file: all i had to do was add a memory wizard before the config file sorcery wizard in sorcery.conf and my pjsip_config_wizard module writes to it. no need for me to write a new sorcery wizard. |
| 18:12.01 | gtjoseph | whoa too many uses of the words wizard and config. |
| 18:12.12 | file | I know what you meant but I don't think that's viable long term |
| 18:12.51 | gtjoseph | no problem, I can clone memory and expose apis to write directly to it |
| 18:13.15 | file | relying on wizard order and assuming it'll be like that = badness |
| 18:14.17 | gtjoseph | no worries. actually i already did the clone so it's just a few extra lines of code to call it. |
| 18:15.40 | gtjoseph | so to be clear... I'm going to create a new wizard (like memory) which has external apis to add and remote to it's internal container. That'll get added to sorcery.conf in any order. |
| 18:15.50 | gtjoseph | s/remote/remove |
| 18:16.24 | gtjoseph | my pjsip_config_wizard will call those ap[is instead of sorcery_create. |
| 18:23.15 | file | you could also allow a wizard to be created and added to a specific instance, and have access to it specifically to manipulate |
| 19:08.35 | gtjoseph | file: can you elaborate? are you suggesting I could create a new sorcery api ast_sorcery_object_register_wizard(sorcery, object_type, wizard_name) ? that would insert a wizard directly into the type's wizard list? |
| 19:09.02 | gtjoseph | that would save hacinf to screw with sorcery.conf. |
| 19:09.05 | gtjoseph | having |
| 19:09.38 | file | yes. |
| 19:09.47 | gtjoseph | perfect |
| 19:14.00 | gtjoseph | actually that's sorcery_apply_wizard_mapping with a few mods to not allocate the object type. |
| 19:15.11 | file | so - if you return the wizard... |
| 19:15.22 | file | that gives you direct access to it, bypassing the chain of things |
| 19:15.30 | file | and guarantees behavior |
| 19:16.11 | gtjoseph | yep |
| 19:16.45 | gtjoseph | makes sense now. pizza always stimulates my brain cells. |
| 19:33.10 | bweschke | mjordan: ping |
| 19:34.56 | mjordan | bweschke: pong! |
| 19:35.53 | bweschke | hey! so wanted to touch base after our emails about uri based playback support and throw in assistance whatever/wherever I can to help it along |
| 19:37.26 | mjordan | awesome :-) |
| 19:37.38 | mjordan | I haven't had a chance to reply to the e-mails yet, but I will today |
| 19:37.50 | mjordan | the caching stuff is a 'doh' on my part - the ETag is a far better way to handle it |
| 19:38.42 | mjordan | I think we can handle an & in a URI, so long as we don't allow an & as part of a resource |
| 19:39.06 | mjordan | i.e., http://myserver.com/sounds/foo%26bar would not work |
| 19:39.24 | mjordan | since we can't tell the difference between a URI encoded '&' as part of a query parameter and as part of a resource name |
| 19:40.07 | mjordan | given that Google has a TTS engine "IN THE CLOUD" - I think it's worth accepting a URI encoded URL in the Playback args and just not letting someone use '&' in the data |
| 19:41.16 | mjordan | The remote playback filestream feels like something that would be very hard to do, but worth exploring. I might end up punting on using func_curl and writing a specific cURL module specifically for pulling down media files. That would probably be needed for the ETag support anyhow. That would at least give you the 'right' location to insert a 'blob of bytes to frames' to create a streaming download |
| 19:41.30 | file | using the curl stuff directly is not hard |
| 19:41.45 | mjordan | file: it isn't, until you want it to behave differently than func_curl |
| 19:41.57 | file | mjordan, it's still not that hard :P |
| 19:42.09 | mjordan | yeah, but I'm not sure I want to overburden func_curl with stuff and thing |
| 19:42.12 | mjordan | er, things |
| 19:42.32 | file | I meant using libcurl directly |
| 19:42.37 | mjordan | AH |
| 19:42.40 | mjordan | yes, it is super duper easy |
| 19:42.54 | mjordan | the problem is the media cache was going to live in the core |
| 19:43.07 | mjordan | and the core does not require libcurl |
| 19:43.43 | mjordan | so there's some dependency stuff to go figure out - plus libcurl can only be initiated one time (currently handled by res_curl) - so there's some stuff & things to think through. |
| 19:44.00 | mjordan | SO! |
| 19:47.01 | file | you and your details |
| 19:50.28 | mjordan | bweschke: so, given that infodump: what part are you interested in? :-) |
| 19:51.04 | seanbright | so make curl a required dependency |
| 19:51.07 | seanbright | it's 2014 |
| 19:51.09 | mjordan | heh |
| 19:51.14 | mjordan | I'd actually be fine with that for 14 |
| 19:51.17 | bweschke | seanbright: true story |
| 19:51.20 | mjordan | I think that would kill this feature on 13 |
| 19:51.23 | file | meh |
| 19:51.34 | mjordan | ah, and the last part: I agree with you on your idea for pushing media to Asterisk |
| 19:51.48 | mjordan | I think that's going to creep into 'another project' - namely, pushing configuration (OR DATA) to Asterisk |
| 19:52.07 | mjordan | must now write this all using words and things |
| 19:52.17 | file | instead of the words you used above? |
| 19:52.54 | mjordan | better ones at leat |
| 19:52.56 | mjordan | er, least. |
| 19:53.46 | bweschke | how extensible is the File Bucket API right now? |
| 19:54.56 | bweschke | I mean ... what if you wanted to change out the local file system with, say, something like memcached ? or is that just "crazy talk that shouldn't even be discussed"? <pun intended> |
| 19:55.17 | file | I wrote it on the premise that it would be used with the existing file API |
| 19:55.26 | file | it doesn't support streaming |
| 19:56.11 | file | so if the underlying scheme impl puts it in a temp file, then all is well |
| 19:56.30 | skrusty | +1 pushing media to asterisk :) |
| 19:56.38 | leifmadsen | bweschke: you live! |
| 19:57.02 | bweschke | leifmadsen: I do!? |
| 19:57.06 | leifmadsen | apparently |
| 19:59.19 | bweschke | What can I say ... I missed the open source crew WAY too much |
| 19:59.26 | leifmadsen | :) |
| 19:59.29 | file | wait - I do open source? |
| 20:00.53 | bweschke | file: I thought your middle name was Oracle |
| 20:01.44 | coreyfarrell | why not have res_curl register functions with the core? something similar to the timer interfaces.. |
| 20:05.53 | bweschke | mjordan: this is a pretty big change in behavior for Asterisk, and then there's the whole sync/blocking issue with remote media on initial acquisition. I never really saw this as a 13 feature in the first place. |
| 20:07.38 | mjordan | yeah, I was hopeful, but the more I wrote down, the less I thought it was an option |
| 20:07.44 | mjordan | coreyfarrell: that is the other option |
| 20:07.52 | mjordan | make a virtual table and have things register to it |
| 20:08.16 | mjordan | I thought about going that route as well, since I think TTS will want a caching mechanism, but it will have a different scheme then 'http/https' |
| 20:08.23 | mjordan | but I haven't thought about that enough yet |
| 20:08.52 | bweschke | and if the virtual table has a null ptr for the function then just say the feature isn't available instead of insisting on curl being a hard requirement? |
| 20:08.56 | bweschke | that's always an option |
| 20:08.56 | mjordan | yup |
| 20:09.20 | mjordan | I think I'll bite the bullet and go down that road. That also lets you optionally load a "res_http_media_cache" as opposed to having to put that in the core |
| 20:09.31 | mjordan | and then write a generic "media_cache" API that multiple providers can register to |
| 20:09.32 | bweschke | right. |
| 20:09.35 | coreyfarrell | putting libcurl into the core locks us in, registering an interface defined by asterisk leaves the option open to use an alterative client library |
| 20:09.39 | mjordan | we could then write a "res_tts_media_cache", etc. |
| 20:09.49 | mjordan | OTOH, 2014! :-D |
| 20:10.39 | leifmadsen | if you just use a docker container and share a local checkout of media files, then it's only 1 per VM, not 1 per container. Solution solved! :) |
| 20:10.57 | bweschke | I know it's non-trivial but I really think at some point when we start talking about doing this kind of stuff, there's going to have to be some consideration made for streaming and/or chunking of media |
| 20:11.39 | leifmadsen | if you do that, then you're DEFINITELY moving Asterisk towards a media service ;) |
| 20:12.21 | bweschke | providing a way for the media to be there ahead of the time that a ast_openstream call is going to need it is really a stepping stone type of approach to the long term solution |
| 20:12.37 | bweschke | there's still some use cases there are not going to be able to use that approach either |
| 20:14.34 | seanbright | why in god's name should this be pluggable? |
| 20:15.15 | seanbright | sorry, came in late |
| 20:15.25 | seanbright | "putting libcurl into the core locks us in" |
| 20:15.26 | seanbright | who cares? |
| 20:15.36 | bweschke | I think the reasoning was so that it could still be a feature candidate for 13 |
| 20:15.45 | bweschke | that was the only reason I could see |
| 20:16.28 | seanbright | i could see allowing "schemes" to be pluggable |
| 20:16.41 | seanbright | behind that (the network library, the caching code) is just a waste of time |
| 20:17.31 | mjordan | agreed - but once you've made schemes pluggable, having libcurl in or out of Asterisk just kind of doesn't matter |
| 20:17.45 | coreyfarrell | I'd want to allow it to be excluded from a build. I have a system where I would not want to allow it to be used, but the customer has access to modify dialplan |
| 20:17.47 | mjordan | a pluggable scheme means the "HTTP" portion of the media cache can be an external module, so it may as well be |
| 20:18.03 | mjordan | coreyfarrell: Customers. Pshaw. :-P |
| 20:18.42 | seanbright | that's like the edgiest of edge cases |
| 20:18.48 | mjordan | bweschke: in the long run, I think having a streaming media option is the right solution. I worry that trying to do it immediately will get us bottled up in dealing with buffering, and off nominal cases of out-pacing the network in playback, and - of course - media containers |
| 20:19.16 | mjordan | to really do streaming media, we'll need to eventually support containers, since video is the likeliest candidate to need the streaming option |
| 20:19.31 | bweschke | coreyfarrell: I COMPLETELY agree, and is why I'm concerned about the synchronous blocks on initial media acquisition. A lot of people aren't going to get what's going on behind the scenes and the front side impression is going to be that "the developers just threw in bleeding edge stuff that doesn't quite work". |
| 20:20.29 | coreyfarrell | what about doing it the same way as libasteriskssl, I'd be fine with ./configure --without-curl |
| 20:20.51 | seanbright | you don't need an external module to do that |
| 20:21.44 | coreyfarrell | yeah so I don't care about it being a module, I do care about being able to disable building it |
| 20:22.13 | mjordan | it's actually easier to do it as an external module once you've made the schemes pluggable |
| 20:22.53 | mjordan | bweschke: if you're looking to do a part of the project, the streaming media part is the part I'd be looking for help on. Thinking through what that API would look like from the perspective of the rest of Asterisk would be interesting |
| 20:23.00 | bweschke | coreyfarrell: what's the real ask though? don't build it, or don't allow it to be used even though it's built |
| 20:23.43 | seanbright | modules modules everywhere |
| 20:24.22 | seanbright | i don't like the & behavior of Dial and my customers have dialplan access... can i compile that out somehow? |
| 20:25.00 | Marquis42 | seanbright: We're going to remove the dialplan anyway, so it won't be an issue soon. :P |
| 20:25.15 | seanbright | well played |
| 20:25.43 | Marquis42 | :D |
| 20:25.57 | seanbright | i'm unusually annoyed by this discussion |
| 20:26.03 | seanbright | i must be in a bad mood |
| 20:27.33 | coreyfarrell | seanbright: I'm not talking about being able to compile out a small part of something. I'm talking about compiling out a big feature that doesn't have to be required |
| 20:28.14 | seanbright | i understand, but we're talking about a library that is packaged on literally every distribution in existence |
| 20:28.44 | seanbright | if it was a library that you had to download/build/install, i could see the argument |
| 20:28.59 | seanbright | but libsqlite3 is required. libxml2 is required. |
| 20:29.18 | seanbright | hell, i think jansson is required |
| 20:29.20 | mjordan | Asterisk: the blurch of dependencies. |
| 20:29.30 | bweschke | right, which is why I think this is really more a discussion about what features you want to allow to be used in a configuration type of setting vs. changing build parameters or removing libraries that are uber-default/commonplace to make sure it cannot be used |
| 20:29.50 | mjordan | blerch, that is. |
| 20:29.52 | mjordan | http://s3.amazonaws.com/theoatmeal-img/thumbnails/blerch_cosplay.png |
| 20:29.52 | seanbright | asterisk.conf - remote_media = off |
| 20:30.10 | bweschke | seanbright: +1 |
| 20:30.11 | seanbright | well... asterisk does a lot of stuff |
| 20:31.00 | seanbright | chrome has 50 library deps and it is a web browser |
| 20:32.48 | mjordan | is still pretty sure this will end up being a moot point |
| 20:33.19 | seanbright | a moo point |
| 20:33.33 | seanbright | libcurl in core, no pluggable cache modules |
| 20:33.36 | seanbright | those are my votes |
| 20:33.37 | bweschke | probably. I still think it's too big of a change for 13. |
| 20:33.37 | seanbright | god speed |
| 20:34.39 | skrusty | (i'll keep saying it) http schema for media |
| 20:35.35 | seanbright | what do you mean? |
| 20:36.40 | seanbright | skrusty: what does "http schema for media" mean? |
| 20:37.01 | skrusty | i.e. play file using url |
| 20:37.44 | seanbright | that is what this entire discussion is about |
| 20:37.46 | skrusty | like mod_http_cache (i think) fpor freeswitch |
| 20:37.52 | seanbright | unless i misunderstand |
| 20:37.59 | skrusty | oh sorry, i didn't read up :) |
| 20:38.18 | skrusty | well, that's good then |
| 20:40.12 | skrusty | ah, read up, now i see |
| 20:40.27 | skrusty | +1 :) |
| 20:41.04 | seanbright | mjordan: bweschke: we already have streaming media in asterisk - it's AGI + STDERR |
| 20:41.27 | mjordan | lawl |
| 20:41.54 | mjordan | does anyone use that? |
| 20:42.05 | seanbright | i use it extensively |
| 20:42.15 | seanbright | and it must stay supported for infinity |
| 20:42.27 | seanbright | (i don't really use it - but i know it exists) |
| 20:42.32 | skrusty | no good for fastagi |
| 20:43.22 | seanbright | sorry, it's not STDERR, it's 3 apparently |
| 20:49.23 | mjordan | oh, open source |
| 20:49.42 | mjordan | "there's this thing I've hacked together, and oh crap now it's totally running some 10000 person company" |
| 20:50.42 | skrusty | :) |
| 20:50.51 | bweschke | mjordan: you have no idea how much I run across that scenario, for real, on a week to week basis. |
| 20:51.29 | skrusty | that's pretty much me atm, not quite that many people though |
| 20:53.05 | leifmadsen | welcome to every company ever built |
| 20:59.27 | file | silly people |
| 21:03.28 | skrusty | :) |
| 21:03.36 | zerohalo | every. single. one. |
| 21:22.01 | bweschke | mjordan: buffering and re-buffering adds a ton of complexity to this. there's no doubt and I think that's kind of outside my comfort zone of knowledge skills in the first place. What if we took a phased approach to this? eg - Playback of uri resources is retrieved in chunks (curl_easy_setopt(curl, CURLOPT_HTTP_TRANSFER_DECODING, 0) and all the callbacks and infrastructure needed with that) and then if |
| 21:22.01 | bweschke | <PROTECTED> |
| 21:22.40 | mjordan | I think that'd be fine |
| 21:22.58 | mjordan | I think there may still be some quirks with hiding this behind the openstream |
| 21:23.13 | mjordan | er, ast_openstream calls, as they exist today |
| 21:23.57 | mjordan | I think as a first step, merely having something outside of func_curl that uses libcurl that exposes the place where you'd grab those chunks to convert them to frames is a good place to get to |
| 21:25.18 | bweschke | I think that's a more than suitable compromise. At some point we've got to remember that this IS 2014 and Asterisk servers SHOULD be sitting on a fat enough pipe such that transfers are able to keep in front of your typical 8khz playback |
| 21:31.54 | mjordan | and I'm out for a bit - I'll update the wiki page tonight :-) |
| 21:32.01 | *** part/#asterisk-dev mjordan (~mjordan@nat/digium/x-jctqrhidjugpdyhn) |
| 22:00.51 | *** join/#asterisk-dev [TK]D-Fender (~chatzilla@64.235.216.2) |
| 22:45.42 | *** join/#asterisk-dev tzafrir (~tzafrir@bzq-218-28-58.cablep.bezeqint.net) |
| 22:45.42 | *** mode/#asterisk-dev [+o tzafrir] by ChanServ |
| 22:50.47 | *** join/#asterisk-dev Dovid (~Dovid@ool-6bbc3907.dyn.optonline.net) |
| 22:52.05 | *** join/#asterisk-dev asterisk562 (d4b2f61f@gateway/web/freenode/ip.212.178.246.31) |
| 22:53.26 | asterisk562 | banging my head, how to detect channel volume increase, say, 30% |
| 22:55.00 | asterisk562 | I mean channel volume on sip channel :) |
| 23:04.21 | asterisk562 | anybody? |
| 23:05.09 | *** join/#asterisk-dev otherwiseguy (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 23:35.22 | *** join/#asterisk-dev terryw (~otherwise@23-228-175-118.mci.googlefiber.net) |
| 23:54.24 | *** part/#asterisk-dev kharwell (~kharwell@nat/digium/x-xcifacwjzkgnljbr) |
| 23:57.01 | gtjoseph | file: it lives. Barely. |
| 23:57.43 | gtjoseph | each object type still needs to call wizard apply or have sorcery.conf set. |
| 23:58.39 | *** join/#asterisk-dev Dovid (~Dovid@ool-6bbc3907.dyn.optonline.net) |
