Wednesday, 2014-06-11

*** tpb has joined #timvideos00:00
mithromorning people01:44
mithroCARAM_: did you see ?01:51
tpbTitle: Create a docker package for flumotion · Issue #56 · timvideos/streaming-system · GitHub (at
mithroshenki: ping?01:53
ayush3504mithro: good morning02:08
mithrohello ayush350402:08
mithrolooking at your report now02:08
ayush3504yes  i see a wombat there ;)02:09
CARAM_mithro: I absolutely have not02:35
CARAM_I have been trying to get collector / encoder configured correctly the past couple of days, but this looks really good...02:35
mithroYou should subscribe to the bug.02:43
ayush3504mithro: doing diagram, could you review again02:51
mithroayush3504: diagram URL?03:04
mithroCARAM_: do you need any help with that?03:04
mithroCARAM_: did you see the tool which generates the configuration for them?03:04
CARAM_yeah and I got CarlFK's notes on it as well03:05
CARAM_so I'm good, but thanks03:05
mithroCARAM_: please do ask questions when you get stuck03:06
mithroaps-sids: ping?03:06
mithroayush3504: I don't see any changes in the diagram?03:07
ayush3504will be pasting just a part of it in the document03:08
ayush3504fixing shapes of connectors and other things03:08
mithroayush3504: which part? draw a box around it03:10
ayush3504mithro: two diagrams, one explaining how the connectors align so that everything fits inside a box. other one showing that a daughter board can be made a adapter with female db903:13
mithroayush3504: sounds good03:13
mithroayush3504: I was also wondering if we could consider having an optionally populated PIC on the motherboard03:13
ayush3504mithro: hmm would it be for?03:15
mithroayush3504: two options03:16
mithroayush3504: a) using the board without the HDMI2USB03:16
mithroayush3504: b) emulating the Numato production board03:16
ayush3504mithro: can u tell a bit more about a)03:17
mithroayush3504: There are plenty of budget PICs which support USB03:17
ayush3504mithro: ah got it03:18
ayush3504mithro: the ones i found cheap for this purpose was PIC18F14K5003:18
mithroayush3504: Yeah there is a bunch of choices03:18
mithroif b) is a primary goal, then we should us the same PIC they used03:19
ayush3504mithro: you mean the PIC24 in digilent board?03:20
mithroayush3504: no. The Numato board has a PIC18F series chip for the PMIC which also does extra peripherals like the serial ports03:20
ayush3504mithro: okay03:20
mithrobut lets discuss that once you have the other stuff finished03:21
mithroayush3504: so?04:06
ayush3504mithro: give me 10-15 minutes to finish documenting the board layout changes04:09
*** CarlFK has quit IRC04:20
*** Niharika has joined #timvideos04:32
ayush3504mithro: review the doc once more04:36
mithroayush3504: looking now04:41
mithroayush3504: diagram looks good04:42
ayush3504mithro: great04:43
*** xobs has joined #timvideos04:44
mithroayush3504: we also have the board going to the wrong edge I think04:44
mithrothe edge it is going to has limited connectors, where as if it went to the other edge it would share with the ethernet, 3.5mm audio, hdmi, power04:45
ayush3504can you point which edge should it be, currently it's the upper edge in the diagram04:46
*** CarlFK has joined #timvideos04:46
*** ChanServ sets mode: +v CarlFK04:46
mithrowhich edge has the thernet, 3.5mm audio, hdmi, power?04:46
ayush3504bottom one.04:49
ayush3504i think i should fix the arrow to point it to upper edge.04:50
ayush3504mithro: see again, the 3rd diagram04:51
mithroayush3504: right, the bottom one04:54
ayush3504mithro: yes, working on middle one right now04:54
mithroshenki: nope!05:07
shenkimithro: ok. im in canberra, so perhaps i'll rsync the ISE dir over to one of our servers while im here05:08
mithroshenki: I had a couple of questions for you05:08
shenkimithro: ok. im about to play counterstrike, but shoot05:09
mithroshenki: a) do we actually understand what the bottle neck for the fps is yet?05:09
shenkinope. that's why ajit is doing the simulation. he is confident that will help us understand the bottle neck05:09
mithroshenki: we get almost twice as much FPS when the quality is 50% right?05:10
shenkii am less confident, but i think it's a task that is worth doing05:10
mithroshenki: that didn't lead anywhere?05:10
mithroshenki: as well, did he get any further with having the uart output the FPS internally?05:10
shenkimithro: i have a strong suspicion that the issue is throughput to/from the DDR05:10
mithroshenki: lastly, do you know much about usb-serial - It looks like with a CDC-ACM device we need 2 endpoints per serial port which kind of sucks. The usb-serial driver appears to support devices which have many serial ports per endpoint?05:11
shenkimithro: i don't know heaps about it. i could do some research05:12
shenkiwhy do you need many serial ports?05:12
*** emusan has joined #timvideos05:12
shenkiwouldn't it be smarter to define a protocol over which you multiplex many operations?05:12
shenkii presume this is for the PTZ stuff. is there an existing protocol/software you're planning on being compatable with, or will we need to develop host side softare05:13
shenkibbl, PS3 time05:13
ayush3504mithro: should i post it now? take a look if you want05:14
mithroshenki: we'd like the multiplexing to be transparent to the software so it can just treat it like a serial port05:14
mithroshenki: your playing PS3 in the middle of the day!? You don't work for Google :P05:15
mithroayush3504: the last diagram needs to be updated right?05:15
emusanmithro: do you have a GSoC student working on CI?05:15
mithroayush3504: the edge issue05:15
mithroemusan: no05:15
ayush3504mithro: i did make the arrow longer05:16
ayush3504mithro: so that it points to the upper edge05:16
emusanah, I could take a stab at it if you'd like, I'm not too familiar with EC2 though05:16
mithroemusan: we where looking at if we could speed up student development / compile time then on their local machines05:16
mithroshenki: was the last person looking at it05:16
mithrobut he has just gone to play games05:17
mithroshenki: writing less drivers is also good :P05:17
mithroshenki: I also wanted to know if there is any reason the Cypress FX2 couldn't emulate a USB hub?05:17
emusanI have a setup right now that will check the last commit before a push and if the word BUILD is in it then it will run a makefile and email the log and any bitfile generated05:17
mithroayush3504: you want to flip the board so the connectors along the bottom edge05:18
mithroemusan: github can do triggers in many different ways05:18
emusanyeah, I use a post-recieve hook05:18
ayush3504mithro: oh okay05:19
emusanand then I just check the last commit and grep out a BUILD, it's not fancy or anything, but it works for me05:19
emusanis there anything more you'd be looking for from it?05:19
mithroemusan: I think we should just always do it05:19
emusanah, okay, that's even easier then05:19
mithroemusan: all types of stats would be useful :P05:19
mithroemusan: but start easy first? :P05:19
mithroemusan: I wonder if we could use travis-ci as an intermediary so we get the green/red badges and such?05:20
emusanI'm not too familiar with that, perhaps05:21
emusanmithro: I don't think travis-ci would work, you need access to the build machine to install the Xilinx tools05:27
mithroemusan: the travis-ci wouldn't actually do the build. It would just call the EC2 system to start if off and then get a response05:27
emusanahh, okay, that could work05:28
emusanthe only other thing I'm not sure about would be licensing...05:28
emusanI don't know if you really care too much about the details of that, just not sure how much a webpack allows one to do in terms of a build-server type thing05:28
emusanI'll look into it some though, I've been considering setting this sort of thing up as a service on my own anyway05:30
ayush3504mithro: see again?05:30
mithroayush3504: looks good!05:31
ayush3504mithro: cool :)05:31
*** Niharika has quit IRC05:46
*** slomo has joined #timvideos06:07
*** emusan has quit IRC06:10
*** slomo_ has joined #timvideos06:15
*** slomo has quit IRC06:15
*** slomo_ has quit IRC06:21
*** slomo_ has joined #timvideos06:22
*** slomo__ has joined #timvideos06:27
*** slomo_ has quit IRC06:29
*** slomo__ has quit IRC06:37
*** slomo has joined #timvideos06:57
*** slomo has joined #timvideos06:57
shenkimithro: big beefy servers don't seem to make it any faster06:57
mithroshenki: yes, I think we discovered that so far?06:58
shenkiwell, i just ran it on a 24-core Xeon sever06:58
shenkijust to confirm06:58
shenkiit saved 40 seconds over EC206:58
shenki5% improvement06:59
shenkinot really worth it imo07:00
mithroshenki: ahh07:01
mithroshenki: kind of sucky :/07:01
mithroshenki: I wonder where they are bound?07:01
shenkiyeah. i wonder what it's doing07:01
shenkithey must have a bunch of solvers that run for a fixed number of seconds07:02
mithroshenki: It would be interested to see how the output compares between a fast machine and slow machine07:04
mithroshenki: IE are we getting some type of better optimisation?07:04
shenkimithro: i wondered that too07:05
shenkimithro: i will write a blog post about my findings. hopefully we can get some answers from people in the know07:05
mithroshenki: that would be great07:05
mithroemusan was more interested in doing the CI stuff07:06
*** slomo has quit IRC07:06
shenkiwho's that ?07:09
*** slomo has joined #timvideos07:12
mithrosee scrollback07:18
*** slomo has quit IRC07:22
shenkithey have gone07:23
shenkiyeah, what he said sounds good07:24
shenkiuse travis to trigger an EC2 instance to start up07:24
shenkido the build, copy out hte contents and log to s3, and update the badge on github07:25
shenkimithro: fireing up a ec2 instance is really quick; much faster than the few minutes we were guessing the other day07:25
mithrotravis generates the badges07:26
shenkithose steps i was describing i was assuming would be run by travis07:31
*** slomo has joined #timvideos07:39
*** slomo has quit IRC07:47
*** slomo has joined #timvideos07:59
*** Niharika has joined #timvideos08:02
mithroaps-sids: ping?08:02
aps-sidsmithro: pong08:02
mithroaps-sids: we have a discont on eater default of <-- do you know where that message comes from?08:03
aps-sidsmithro: it's  there in feedcomponent0.10.py08:04
mithrocan you link me to your current code with that?08:05
tpbTitle: flumotion-orig/flumotion/component/ at porting-to-gst1.0 · aps-sids/flumotion-orig · GitHub (at
aps-sidsmithro: I haven't commit changes I made for porting httpstreamer since its not working. Should I commit those too?08:07
mithroaps-sids: yeah, you should do it on a separate branch08:08
tpbTitle: httpstreamer ported (stream not working) · f4db274 · aps-sids/flumotion-orig · GitHub (at
*** slomo has quit IRC08:12
mithroaps-sids: so that message is coming from gstreamer then it appears?08:15
mithroyou linked me to the wrong line number?08:16
mithro ???08:16
tpbTitle: flumotion-orig/flumotion/component/ at porting-to-gst1.0 · aps-sids/flumotion-orig · GitHub (at
*** slomo has joined #timvideos08:17
aps-sidsmithro: oh yes, oops08:18
aps-sidsmithro: actually, this is the correct one,
tpbTitle: flumotion-orig/flumotion/component/ at porting-to-gst1.0 · aps-sids/flumotion-orig · GitHub (at
mithroaps-sids: my current guess is that something in that structure has changed08:27
mithro we have a discont on eater default of 3399999774 units between 226 and 3400000000  (flumotion/component/
mithroI think that should be08:28
mithrowe have a discont on eater default of 226 units between 3399999774 and 340000000008:29
aps-sidsyeah, I think so08:32
mithroaps-sids: lets try and solve that problem first?08:33
*** rohitksingh has joined #timvideos08:35
mithroaps-sids: I don't think that is a major issue though08:38
mithroaps-sids: you probably want to filter out packager from the log output too08:41
mithroaps-sids: I think this might be your issue...08:44
tpbTitle: Pastebin: SfGN1vA0 | IRCCloud (at
*** sarwarc has joined #timvideos08:51
*** slomo has quit IRC08:59
mithroaps-sids: I'm following your "Run a Test Stream" page09:09
mithroMy output from run-tests is as follows09:09
tpbTitle: Pastebin: WqLS5rQH | IRCCloud (at
aps-sidsmithro:  that is due to a change I did after that post. Actually, the values test is passing need to be changed from an array to a GstStructure. I'm aware of this.09:11
aps-sidsI was testing some configs just now and I got a happy httpstreamer component. I'm just confirming everything and testing the stream.09:12
mithroI also get the following issue09:16
tpbTitle: Pastebin: tqIMoIs1 | IRCCloud (at
aps-sidsmithro: I'm seeing this for the first time. Also, I got the every component's mood as "happy". It seems there was a problem in configuration I was using.09:23
mithroaps-sids: yay!09:25
aps-sidsmithro: It seems there are discontinuities due to encoder component. Feeding videotest producer directly to httpstreamer causes no discontinuities. Opening stream with gst-launch says "cannot decode stream"09:29
mithroaps-sids: what is your gst-launch command?09:30
*** slomo has joined #timvideos09:30
aps-sidsmithro: gst-launch-1.0 -v playbin uri=http://ubuntu.local:8800/ogg-video/09:30
mithroaps-sids: I doubt there is enough information without encoding for it to figure out the format09:31
*** sarwarc_ has joined #timvideos09:31
*** sarwarc has quit IRC09:33
aps-sidsmithro: I just checked everything once again. Discontinuities were due to config problems (linking httpstreamer with porter). That problem is now solved (even for encoder).09:35
aps-sidsmithro: now, if I run without encoder, I get the expected error " Could not determine type of stream."09:35
aps-sidsmithro:  with encoder, it gives "Could not decode stream."09:36
mithroyou mean *with* the encoder?09:36
mithrocan you give me your full config?09:37
aps-sidsmithro: probably I have missed changing playbin2 and decodebin2 elements to playbin and decodebin respectively. just a sec09:37
tpbTitle: [XML] - (at
*** slomo has quit IRC09:50
*** slomo has joined #timvideos10:03
mithroaps-sids: you are going to need a muxer in there before it can be decoded10:06
aps-sidsmithro: I thought so, I'm porting muxer right now10:08
mithroaps-sids: great!10:08
aps-sidsand after this, we won't need to manually use configration, flumotion-admin can be used directly.10:09
mithroaps-sids: I always manually use configuration, you get much better control that way10:09
*** slomo has quit IRC10:17
*** slomo has joined #timvideos10:32
*** slomo has quit IRC10:39
*** slomo has joined #timvideos10:52
*** Niharika has quit IRC11:06
*** sarwarc_ has quit IRC11:13
*** loadstar91 has joined #timvideos11:22
*** loadstar81 has quit IRC11:23
shenkimithro: i got billed for the AWS time I used11:23
mithroshenki: hrm11:23
shenkimithro: so something didn't work with the consolidated billing?11:23
mithroshenki: it seems to be linked as far as I can see?11:25
shenkimithro: err, i just got an email that you cancelled it11:25
mithroshenki: that was the old one11:27
mithroshenki: I'm seeing that I "paid for" 1.80 of EC2 usage in the Sydney region from you11:27
mithroshenki: so you should try and figure out why you where charged11:32
shenkimithro: ok, i'll do some digging11:33
mithrodid you use anything *before* the consolidated billing request?11:33
shenkii don't recall. i don't think so11:34
*** sarwarc has joined #timvideos11:40
*** slomo has quit IRC11:58
*** slomo has joined #timvideos12:03
*** ayush3504 has quit IRC12:03
*** slomo has quit IRC12:28
*** slomo has joined #timvideos12:46
*** slomo has quit IRC12:46
*** slomo has joined #timvideos12:46
*** thaytan has quit IRC12:49
*** thaytan_ has joined #timvideos12:49
*** sarwarc has quit IRC12:51
*** CARAM_ has quit IRC12:51
*** CARAM_ has joined #timvideos12:53
*** sarwarc has joined #timvideos12:55
*** slomo_ has joined #timvideos13:03
*** slomo has quit IRC13:05
*** slomo_ has quit IRC13:12
*** slomo_ has joined #timvideos13:12
*** rohitksingh1 has joined #timvideos13:18
*** rohitksingh has quit IRC13:20
*** slomo__ has joined #timvideos13:20
*** Niharika has joined #timvideos13:23
*** slomo_ has quit IRC13:23
*** mparuszewski has joined #timvideos13:39
*** CarlFK has quit IRC13:41
*** slomo_ has joined #timvideos13:47
*** slomo__ has quit IRC13:50
*** CarlFK has joined #timvideos14:04
*** ChanServ sets mode: +v CarlFK14:04
*** slomo_ has quit IRC14:14
*** slomo has joined #timvideos14:26
*** slomo has quit IRC14:38
*** slomo has joined #timvideos14:51
*** sarwarc has quit IRC15:15
*** CarlFK has quit IRC15:17
*** CarlFK has joined #timvideos15:23
*** ChanServ sets mode: +v CarlFK15:23
*** skay_ has joined #timvideos15:26
*** skay_ has left #timvideos15:26
*** slomo has quit IRC16:38
*** sarwarc has joined #timvideos16:45
*** Niharika has quit IRC16:52
*** sarwarc has quit IRC17:22
*** sarwarc has joined #timvideos17:23
*** sarwarc has quit IRC17:27
*** slomo_ has joined #timvideos18:18
*** slomo_ has joined #timvideos18:18
*** mparuszewski has quit IRC18:51
*** rohitksingh has joined #timvideos19:18
*** rohitksingh1 has quit IRC19:21
*** mparuszewski has joined #timvideos20:01
*** shenki has quit IRC20:39
*** tariq786_lab has quit IRC20:39
*** shenki has joined #timvideos20:40
*** ChanServ sets mode: +v shenki20:40
*** tariq786_lab has joined #timvideos20:40
*** rohitksingh has quit IRC20:44
*** slomo_ has quit IRC21:39
mparuszewskimithro: can we update Ruby on
mparuszewskiI have some differences between my generated website and website generated on server and I can't find causes.22:07
mithromparuszewski: no22:47
mithromparuszewski: the ruby you should be using is the one in trusty22:47
mithroCARAM_: Apparently AppEngine has docker support now?22:54
tpbTitle: Google Cloud Platform Blog: An update on container support on Google Cloud Platform (at

Generated by 2.12.1 by Marius Gedminas - find it at!