Monday, 2014-12-01

*** tpb has joined #timvideos00:00
mithrohyades-: ping me when you get up00:38
mithroCarlFK: audio is treated very similarly to video00:38
mithroCarlFK: we turn it into video too provide a visual representation on what the audio is appearing00:39
CarlFKmithro:   well, video has thumbs so you have a clue what it is.00:39
CarlFKah, so like a vu meeter ?00:40
mithroCarlFK: yeah00:40
CarlFKgood enough00:40
mithroCarlFK: like a multi-frequency VU meter00:40
CarlFKI can kinda dream up some way of previewing audio, but I (we?) only ever have one source anyway00:41
CarlFKor, I have multiple souces, (PA + camera mic) that I setup once and leave it that way the whole show00:42
CarlFKon the left side "videosink N" comes from audio.00:46
CarlFKthat diagram doesn't really show support for more than one audio source00:48
mithroCarlFK: well, eventually I hope we can do something a lot more intelligent with audio - but that is a long way of00:48
CarlFKyeah, I am not to worried about it00:48
mithroCarlFK: that is more a space issue :)00:49
*** hyades has joined #timvideos00:49
mithroCarlFK: on the diagram, I'm pretty sure multiple audio feeds are supported00:50
CarlFKkinda - 2 go in, then it is unclear.  but as long as this isn't making you wonder "how does the app do audio.. oh, it doesnt."00:54
mithroCarlFK: yeah - it doesn't work very well at the moment00:59
CarlFKmithro: can it handle 1 audio source and just mux/save it?01:06
mithroCarlFK: what do you mean?01:06
mithroCarlFK: video + audio is totally independent01:07
CarlFKwhat doesn't work very well?01:07
mithroCarlFK: the UI for audio01:07
CarlFKfosdem is showing interest01:09
CarlFKit's in Feb01:09
mithroCarlFK: we'll know after the sprints + LCA if its ready for something like FOSDEM01:09
CarlFKyou still have too high a bar for "ready for fosdem"01:10
CarlFKwouter is starting to make a dvswitch replacment in perl01:10
CarlFKby himself01:10
mithroCarlFK: beat him with a stick01:10
CarlFKhave you merged in the 2 PR I remviewed ?01:11
mithroCarlFK: not yet, but will very shortly01:11
CarlFKhe told me about his perl stuff as I was building them.01:11
CarlFKit was good that I was able to tell him "bugs are getting fixed"01:12
mithroCarlFK: as I mentioned, the work which gets done is proportional to the effort other people put into it01:13
CarlFKIt still doesn't look like a safe bet, so I can understand why wouter is spending his time making his own thing01:15
mithroCarlFK: making his own thing is a significantly less of a safe bet than starting from an existing thing01:15
CarlFKfor various values of things ;)01:16
CarlFKI didn't say it was a good idea, just that I can understand why a human would behave that way01:17
mithroCarlFK: I wish he could get out to the sprints01:19
CarlFKhe is even more busy than me01:19
CarlFKi'm hoping to get some of his time during the sprint01:20
CarlFK1/2 because he will likely contribute, but really just to get him engaged and not write more perl01:21
CarlFKmithro: is there a concept of DVswitch Cut button?01:31
mithroCarlFK: yes, I think it's the "r" button01:32
mithrofrom the docs01:32
mithro <tr><td>r</td><td>01:32
mithro Start a new recording, the *--record _name_* will be used as a template,01:32
mithro e.g. *record 2013-01-23 131139.dat*01:32
CarlFKk - so the file name will be timestamped or something ?01:32
CarlFKah right.  saw that and wondered: does it do the datetime automatically, or do I need to feed it a template ?01:33
mithroCarlFK: the default template has the datetime automatically in it01:39
mithroCarlFK: i think....01:39
CarlFKI'll check that out pretty soon01:39
*** micolous has quit IRC01:42
*** micolous has joined #timvideos01:49
mithrothaytan: ping?01:59
JoelwI think he's still asleep :(02:02
mithroJoelw: how do you know that? :P02:02
JoelwHis webcam02:03
JoelwIt's hard to tell asleep from dead though :(02:03
mithroCarlFK: it sounds like is unrelated to that pull request?02:19
tpbTitle: Replace faac component with voaacenc. by mithro · Pull Request #42 · timvideos/gst-switch · GitHub (at
mithroand I think you mean it needs libgstreamer-plugins-bad1.0-dev02:21
CarlFKmithro: 1/2 related.02:21
CarlFKah, yes.  that's what the -++ diff below shows02:21
CarlFKthat didn't format as hoped02:21
mithroyou need to wrap it in triple backticks02:22
mithroif you do ```patch it'll even color it nicely02:23
mithroCarlFK: I can't see any files in libgstreamer-plugins-bad1.0-dev that would be required?02:23
CarlFKpretty sure that's where the voaacenc stuff is.  but I may have fooled myself02:24
mithroCarlFK: it's in gstreamer1.0-plugins-bad package02:26
mithro$ dpkg --listfiles gstreamer1.0-plugins-bad | grep voaac02:26
CarlFKbut don't we need the -dev to link it into gst-server?02:30
CarlFKhuh. I'll swear it erorred until I installed -dev.02:31
CarlFKbut I am not testing that in a clean environment, so who knows02:31
mithroCarlFK: do you need anything from me regarding LCA now?02:36
mithroCarlFK: just replied to your last question I think02:36
CarlFKthat's good enough for me.  thanks.  I think it is up to Ryan to figure out my air02:38
CarlFKI really really hope I don't have to bring tripods - that seems silly02:38
CarlFKoh oh.. my flight my break him02:39
CarlFKlets see what I can find...02:40
CarlFKzoiks.  $220002:41
mithroyeah, I'm not seeing much :(02:41
mithroalso have to remember USD->AUD conversion02:43
CarlFKwhats the rate?02:44
mithro1 USD == 1.19 AUD02:44
CarlFKoh hell.  wrong direction02:45
mithrocheapest I can see is USD $2,313.80 :/02:45
CarlFKI can help by moving the in/out city to SFO02:46
CarlFKand cover the flight out to visit my mom for her birthday at the end of Jan02:46
mithroyeah, I was just looking into that02:47
*** hyades has quit IRC04:05
*** hyades has joined #timvideos05:31
mithrohyades-: poke?06:06
hyadesmithro: ping06:06
mithrohyades: so we never finish discussing the travis stuff06:07
mithrohyades: can you walk me through the current set up?06:08
mithrohyades: and then I'll explain my idea for how we should change it06:09
hyadesso there are 4 things06:09
hyadesit will first setup using the in all of the environments06:09
mithrohyades: yeah, I understand the scripts bit06:10
mithrohyades: I'm more after what each bit is suppose to cover06:10
hyadesthe first one will do a make unittests06:10
hyadessecond one will do integration06:10
mithro  - TEST=python-api TYPE=unittest06:11
mithro  - TEST=python-api TYPE=integration06:11
mithro  - TEST=module TYPE=python06:11
mithro  - TEST=module TYPE=c06:11
hyadesthird one is supposed to run all tests we got06:11
mithroby first one you mean "TEST=python-api TYPE=unittest" and by "make unittests" you mean "It should run the unit tests for the gst-switch Python API" ?06:12
mithrohyades: and what do the "gst-switch python API unit tests" consist of?06:12
mithrohyades: I assume they run the API through a bunch of stuff with a fake gst-switch server responding or something?06:14
hyadesthey unit test the components of the API only06:14
hyadesthey mock all the "real" stuff06:15
mithrohyades: great06:15
mithrohyades: so what does "TEST=python-api TYPE=integration" cover?06:15
hyadesthese cover the end to end tests06:15
mithroare they the same set of tests as above with real gst-switch server?06:15
mithroend to end tests meaning....06:16
hyadesthey arn't the same06:16
hyadesthey will run the server, do things like add inputs, then change the PIP mode, record the video and then check if the output is correct06:17
mithrohyades: these tests are mainly targeting and making sure the Python API works?06:17
mithrohyades: so, the aim is to test the Python API, not gst-switch itself right?06:17
hyadesum yea06:18
mithrohyades: okay, so what about the remaining two?06:19
hyadesthough gst-switch is also tested  while we do this06:19
hyadesthe TEST=module TYPE=python is something which should ideally run all the tests we got, including pure C tests, but it doesn't right now06:20
mithro"all the tests" mean?06:21
hyadesright now it runs all unittests, integration, lint06:21
hyadesthis is something which doesn't have a lot of significance06:22
mithrowhy is it TEST=module TYPE=python ?06:23
hyadesguess coz they were supposed to test everything related to the python-api06:24
mithrohyades: why do we run the python-api tests separately and then together?06:24
hyadesyea these are a bit redundant. If we have a bunch of tests for the api, we can run them all together here, instead of creating the env again and again06:25
mithrohyades: so my suggestion is we have the following06:27
mithrohyades: first we split the run into "Python API" and "gst-switch" bits06:27
mithrohyades: then for the Python API we have style, unittest and integration06:28
mithrohyades: and for gst-switch we have style, unittest, integration06:28
mithrocurrently gst-switch has "empty" unittest suite06:28
mithrothe style ones just check for style compliance06:28
mithroand don't run any tests or anything06:29
hyadesI guess we have a small thing for testing gst-switch in C, which opens up the UI for I guess around half an hour06:31
hyadesand there are style checks for the C code06:31
mithrohyades: there is a bunch of tests under gst-switch/tests/06:32
hyadesyes, was mentioning those06:36
*** rohitksingh has joined #timvideos08:03
*** slomo has joined #timvideos08:06
*** rohitksingh has quit IRC08:29
*** rohitksingh has joined #timvideos08:31
*** tariq786 has quit IRC09:17
*** tariq786 has joined #timvideos09:19
*** rohitksingh1 has joined #timvideos10:29
*** rohitksingh has quit IRC10:32
*** rohitksingh1 has quit IRC11:01
*** rohitksingh has joined #timvideos11:03
thaytanmithro, pong11:24
mithrothaytan: two questions11:33
mithro1) when are you in Auckland?11:34
mithro2) can you check that gst-switch pull request seems sane and I didn't misunderstand what you were suggesting?11:35
thaytanArriving Wed 7th in the afternoon11:37
thaytanwhich pull request?11:38
thaytan I guess?11:38
tpbTitle: Use sync=false on imagesinks as feeds are live. by mithro · Pull Request #40 · timvideos/gst-switch · GitHub (at
thaytanmithro, if it seems OK, use it11:39
thaytanthe alternative is more complicated - getting proper synch info delays passed through everywhere.11:40
thaytanwithout sync you'll maybe get a bit more display jitter11:40
mithrothaytan: what re sync info delays?12:03
thaytaninfo + delays, I mean12:04
thaytanhow does the original dvswitch sync incoming streams?12:04
mithrothaytan: it does if I understand correctly, they just come in and it switches frames12:11
thaytanwell, that's closest to sync=false :)12:12
thaytanIf I was designing the system, I think I'd use RTP to send packets around12:12
thaytanand rtpjitterbuffer to smooth out network jitter and handle packet losses12:13
thaytanplus network clocks for synchronising distributed capture sources12:13
mithrothaytan: how does the network clocks stuff work?12:16
*** tvCommitBot has joined #timvideos12:17
tvCommitBot[gst-switch] mithro pushed 2 new commits to master:
tvCommitBotgst-switch/master e2d91fd Tim 'mithro' Ansell: Use sync=false on imagesinks as feeds are live....12:17
tvCommitBotgst-switch/master 7c049da Tim Ansell: Merge pull request #40 from mithro/sync-false...12:17
*** tvCommitBot has left #timvideos12:17
mithrothaytan: IE could you replace the tcp+gdppay stuff with rtp?12:18
thaytanthe network clock works like a local ntp12:19
thaytanusing udp ping/pongs to synthesise a remote clock locally12:19
thaytanso you run one network clock provider, and client clocks on each receiver/capture machine12:19
mithrothaytan: that means running more than one gst-launch command on a capture machine, right?12:20
thaytanmithro, you can't do network clocks in gst-launch12:21
thaytanyou'd use a python script instead12:21
mithrothaytan: okay12:24
mithrothaytan: got an example Python script?12:24
thaytan and are the original ones from 0.1012:25
thaytanI should port them and update12:25
mithrothaytan: I'm heading to bed, gnight!12:50
thaytannight :)12:50
*** Niharika has joined #timvideos13:20
*** rohitksingh has quit IRC13:36
*** rohitksingh has joined #timvideos13:41
*** slomo_ has joined #timvideos14:48
*** slomo_ has joined #timvideos14:48
*** slomo has quit IRC14:52
*** slomo_ has quit IRC16:19
*** slomo_ has joined #timvideos16:19
*** rohitksingh1 has joined #timvideos17:05
*** rohitksingh has quit IRC17:06
*** shenki has quit IRC18:03
*** slomo_ has quit IRC18:03
*** shenki has joined #timvideos18:04
*** ChanServ sets mode: +v shenki18:04
*** slomo has joined #timvideos18:06
*** Niharika has quit IRC19:12
*** hyades has quit IRC21:05
*** shenki_ has joined #timvideos21:25
*** ChanServ sets mode: +v shenki_21:25
*** thiagoss` has joined #timvideos21:28
*** micolous_ has joined #timvideos21:33
*** slomo has quit IRC21:36
*** shenki has quit IRC21:36
*** micolous has quit IRC21:36
*** thiagoss has quit IRC21:36
*** rohitksingh1 has quit IRC21:45
*** thiagoss` is now known as thiagoss22:04
*** shenki_ is now known as shenki22:26

Generated by 2.13.1 by Marius Gedminas - find it at!