Thursday, 2017-01-12

*** tpb has joined #timvideos00:00
*** hyadez has joined #timvideos00:52
techman83jea: no perl! blasphemy!01:04
* techman83 got hired as a python developer01:04
techman83jea: I'll be about at the conference if you want to pick my brain, I've actually wanted to re-develop eventstreamr in Python.01:07
jeatechman83: haha. perl is not a problem, just that I don't know enough of it to do useful things with it01:08
techman83jea: though for data gathering + metrics, it might worth using something like prometheus and just write some custom exporters. No need to write a whole stack then.01:08
jeaI have a mostly rewritten version of eventstreamr in python, but it hasn't been used properly yet. most of the bits that I didn't convert can go away now too due to not using dvswitch01:09
techman83jea: That's the main problem, not enough perl devs.01:09
techman83jea: nice01:10
jeaand yes, prometheus is the plan for the monitoring. i got it setup the other day, now i just need to generate the data!01:10
* techman83 dances01:10
jeabut first i need to work out why these numbers are complely off. seems like it is not really using a long...01:11
xfxftechman83: turned to the dark side? ;)01:18
xfxfcongrats on new job btw01:18
techman83xfxf: haha, yeah. A job offer popped up. Requirement was that I learn Python.01:19
techman83Thanks. It's much more interesting and far less soul destroying than my previous job!01:19
tumbleweedCarlFK: https://docs.python.org/3/library/re.html#re.sub01:20
tpbTitle: 6.2. re — Regular expression operations Python 3.6.0 documentation (at docs.python.org)01:20
techman83I figured if I wanted a dev job, Python was where it's at. So I'd been working in it for a little while as it was.01:20
mithro_florent_: So, libuip is definitely not being rebuilt correctly at the moment02:16
*** nrossi has joined #timvideos02:18
mithro_florent_: Found a bug in the Makefile for libuip02:19
mithroxfxf: Where is this spreadsheet?02:20
xfxfhttps://docs.google.com/spreadsheets/d/1W3p1J-Lb6l-sWmh6n9-YJ_U5o7a0pDmWAR6QIKfCzI0/edit#gid=118460116502:20
tpbTitle: Google Sheets - create and edit spreadsheets online, for free. (at docs.google.com)02:20
*** CarlFK has quit IRC02:21
mithro_florent_: https://gist.github.com/mithro/a5ed4d86e53a81259b835ed559fcf2b1 my check_csr.py script02:21
tpbTitle: Hacky script to check the csrs dont violate constraints in litex · GitHub (at gist.github.com)02:21
*** CarlFK has joined #timvideos02:21
*** ChanServ sets mode: +v CarlFK02:21
mithroxfxf: What do the numbers in the cell means?02:22
xfxfexplained up the top02:22
mithroxfxf: I would have never seen that myself :-P02:23
xfxfit was the easiest way to help identify patterns02:23
xfxfa 2 so far always means it works but only after a power cycle02:24
xfxfdo i have to check every single cable.  this shit is tedious02:24
xfxfsurely redmere + white/std is enough?02:24
xfxfmithro: i think you just moved values in the column i'm currently testing in...02:27
mithroxfxf: Oh sorry - I rearranged it to be input first, then cable type02:27
xfxfalso the order you've stuck things in is not how i'm testing them02:27
xfxfi'm doing different cables on both inputs, not power cycling + test, then power cycle + test each time02:28
xfxfer rather, above is confusing02:28
xfxfi'm doing: take a cable, plug it in to input0, see what happens, power cycle device if it didn't work, see what happens, then repeat on input 1.  then onto a new cable02:29
mithroxfxf: I need to be able to rearrange the table so I can understand the data - is there a way to do that?02:29
xfxfnot sure, i just want this to flow in how i'm doing things so i don't make data entry errors02:30
xfxfpretty sure input0 is boned on this board anyway02:30
mithroxfxf: I added a test-order column which lets me quickly switch back and forth between the two sorting orders02:31
xfxflooking at this i'm pretty sure i can reduce the testing to 1 redmere cable, 1 white cable, on all devices, on both input0/input1, with the power cycle/non power cycle test02:31
xfxfnp02:31
xfxfpls confirm above02:31
xfxfi assume i'll be able to do a subset of devices too which are acting different (i.e. mDP dongle vs HDMI port on my mac, versus the xa20)02:32
mithroxfxf: Looking at it at the moment02:32
xfxffeel free to re-sort, i'll go outside for a sec02:32
mithroI was sure there was a way to make the sort order specific to the user with the "views" stuff02:34
mithrobut I don't have the mental capacity to figure that out again02:34
mithroxfxf: Looking at the spreadsheet, there is one weirdness02:35
xfxfwhich is?02:36
xfxfoutside of the xa20's always working02:36
mithroxfxf: the "Lenovo W530 - mDP" was different on the Blue/White Redmere cable?02:36
xfxfit was? let me look02:37
mithroxfxf: I'm wondering if that is just an entry error or something02:37
xfxfoh02:38
xfxfright02:38
mithroxfxf: You written "same behaviour as blue 3ft redmere below"02:38
xfxfso i suspect it's the same problem02:38
xfxfexcept the cable might have worked once02:38
xfxfboth are problematic02:38
xfxfoh wait no02:39
xfxfboth are meant to be 302:39
xfxffixed02:39
mithroxfxf: Can you see if you can reproduce that at all? I don't think we have ever seen a difference between redmere cables?02:39
*** paradisaeidae_ has joined #timvideos02:39
xfxfsure02:39
xfxfthis is again, the most tedious shit ever :P02:39
mithroxfxf: Yeah sorry, we really need some minions :(02:40
xfxfis there anybody else we could grab?02:41
xfxfi really should be working on this monitoring stuff.. it is thu arvo02:41
xfxfsomebody def should test all boards though02:41
mithroxfxf: Can we extract the following extra data for each cell on this board02:42
mithroxfxf: If you are seeing the "dvisampler0: IDELAY busy timeout"02:43
mithroxfxf: If we are seeing the "pixelclk" value02:43
xfxfno02:43
xfxfwtf02:43
mithroxfxf: If we are seeing the WER02:43
xfxfthe redmere cable works on input0 sometimes?02:43
xfxfit seems random02:43
mithroI think they are all in your notes already02:43
xfxfi'm seeing errors02:43
xfxfno idea busy timeout02:43
xfxfwait, nm, i am02:43
mithroxfxf: You okay with me added extra columns under each device and putting that there?02:44
xfxfum, i guess, how much more confusing is this going to be?02:44
xfxfokay, so occasionally when i unplug it / plug it back in, it works02:45
xfxfboth redmere cables02:45
xfxfi don't even know how to record this data anymore, this is too confusing to display two dimensionally02:45
xfxfmithro i'm struggling dealing with representing sane test data on what appears to be a faulty board that doesn't always do the same thing02:49
xfxfi've marked it as '3' and added notes02:49
mithroxfxf: Looking at the data so far, It looks like there is an issue with the IDELAY stuff on this board in Input 0 of this board02:50
xfxfokay, well, this data is for you, so you tell me if what i'm providing is enough02:50
xfxfi'm going to try and rope somebody else in to test all ~17 or whatever boards but we need repeatable test instructions first02:51
xfxfi don't think me spending the rest of today doing this is a good idea considering everything else i need to be on top of02:51
xfxfmy next steps should be trying a new board, continuing along the existing board with new devices, something else?02:52
mithroxfxf: So if IDELAY timeout occurs, then a high WER rate and the input not working is expected02:54
xfxfok, answer to Q above?02:55
mithroHold on, writing some stuff02:56
xfxfi really want to continue the monitoring s/w (given the extra complexity there now) so want to progress on this rapidly02:56
xfxfokay, rad, ta02:56
mithro_florent_: https://github.com/enjoy-digital/litex/pull/1503:25
tpbTitle: Add strcasecmp function to lib by joeladdison · Pull Request #15 · enjoy-digital/litex · GitHub (at github.com)03:25
jeamithro: how do i update the submodule now?03:37
mithrojea: There is a "merge-upstream.sh" script in ./third_party/03:37
jearight03:38
mithrojea: But upstream needs to merge the change before that script works03:38
jeait has been merged now03:38
mithrojea: It will merge all the upstreams too :-P03:39
jeamithro: I'm getting a thing about 'dirty'. do I need to clean something first?03:39
mithro_florent_: http://hopper.mithis.com/~tim/opsis_video_out_debug_lm32.tar.bz203:43
*** paradisaeidae_ has quit IRC03:49
*** paradisaeidae_ has joined #timvideos03:51
mithroxfxf: https://www.raspberrypi.org/documentation/configuration/config-txt.md04:05
tpbTitle: config.txt - Raspberry Pi Documentation (at www.raspberrypi.org)04:05
*** paradisaeidae_ has quit IRC04:29
_florent_mithro: I'm not able to get the file04:47
tumbleweedCarlFK: https://github.com/xfxf/lca2017-av/commit/e8bcaf7fb282d4158c9893542169624498d8549804:49
tpbTitle: Add r3mix · xfxf/[email protected] · GitHub (at github.com)04:49
tumbleweedCarlFK: https://anonscm.debian.org/cgit/debconf-video/ansible.git/04:53
tpbTitle: debconf-video/ansible - Ansible playbooks for the DebConf Video Team's machines (at anonscm.debian.org)04:53
*** CarlFK has quit IRC06:23
*** CarlFK has joined #timvideos06:26
*** ChanServ sets mode: +v CarlFK06:26
mithrojea: Is that change almost ready to merge?07:26
mithrojea: I'd really like to check the blackmagic stuff ASAP07:27
mithrotumbleweed: There is some confusing stuff in https://linux.conf.au/wiki/conference/chat/07:35
tpbTitle: linux.conf.au 2017 | Chat Platforms (at linux.conf.au)07:35
CarlFKmithro:   Rooms:  Plenary Hall,  Tasman Hall A,  Tasman Hall B/C,  Boardwalk Gallery,  Wellington Room 1,  Wellington Room 2,07:35
tumbleweedmithro: thanks. and yes, that's confusing :)07:37
mithroNormally they are something like "#lca2017-plenary", "#lca2017-halla", etc...07:38
mithroUsing https://plus.google.com/+MikeWiseguy/posts/6Cyq9EwWpPY07:40
mithro /msg alis list *lca*07:40
mithroDoesn't seem to list any channels07:40
tumbleweedmithro: thanks07:47
mithrotumbleweed: The IRC channel names are08:40
mithro7:18 PM F<faulteh> #lca2017_plenary, _tasman_a, _tasman_bc, _boardwalk, _wellington_1, _wellington_208:40
*** hyades has joined #timvideos08:45
tumbleweedmithro: thanks!09:55
tumbleweedCarlFK: ^^09:55
mithro_florent_: The problem seems related to my TOFE module somehow09:57
mithro_florent_: I just pushed all the latest code I have to the nextgen branch10:09
mithro_florent_: I was playing with the modules in this https://github.com/mithro/HDMI2USB-litex-firmware/blob/nextgen/targets/opsis/base.py#L25010:09
tpbTitle: HDMI2USB-litex-firmware/base.py at nextgen · mithro/HDMI2USB-litex-firmware · GitHub (at github.com)10:09
_florent_mithro: ok I look at that10:10
mithro_florent_: I'm currently trying build with an empty TOFE module10:10
mithro_florent_: I also uploaded a bunch of the builds at http://hopper.mithis.com/~tim/10:10
tpbTitle: Index of /~tim (at hopper.mithis.com)10:10
mithro_florent_: Do you think it is worth setting you up a Linux laptop which has the same environment that myself, xfxf and jea are using?10:13
_florent_mithro: not for now10:14
mithro_florent_: I'm probably going to go to bed soon, I'm not feeling great10:22
_florent_mithro: ok, I'm looking at your build10:22
_florent_mithro: is you design with empty TOFE module working?10:23
mithroSeems so10:23
mithroI just rebuilt with the I2C module re-enabled and it worked again10:23
mithro_florent_: I should just build the same gateware like 10 times and see if they all fail / all pass10:25
xfxf_florent_: poke myself/Carl/tumbleweed if you decide you do want one - we have lots of spare laptops10:26
_florent_mithro: not sure if it can be related but there a warning on your TOFE module:10:29
_florent_https://www.irccloud.com/pastebin/CJ3vLNPL/10:29
tpbTitle: Snippet | IRCCloud (at www.irccloud.com)10:29
mithroFor when micolous turns up, first thing with streaming would be to import the schedule database into the streaming-system website10:32
mithro_florent_: The one that is currently working on the TV has that warning10:32
xfxfmithro: you're way more on top of what's required for streaming than I am, can you instruct micolous on what you'd like him to do over the next couple of days when he arrives?10:35
xfxffrom my POV i want to make sure anything he used to do the YT streaming last year is committed up to a repo, and there's a handoff to jea before the conf starts10:35
mithroHe probably knows better then I do what the current state of it is10:36
xfxfi'd imagine there'll be some integration with the ansible stuff tumbleweed/CarlFK are working on, so it's integrated correctly into the systemd foo they're doing + the right bits are on the machine10:36
mithroNFI on that part10:37
xfxfme either, poke tumbleweed, chat with micolous tomorrow (he arrives anytime soon) about what of his YT/gstreamer streaming stuff needs to end up on the machines10:39
xfxfdebconf use icecast normally, correct?10:39
tumbleweedyep10:41
mithroxfxf: I seem to recall that nothing was run on the systems locally - everything was run on a system in the NOC and it just connected to voctomix's streaming port remotely10:41
mithrobut I could be wrong10:41
jeathat does sound familiar from what i remember of last yera10:44
jea*year10:44
jeamithro: i've updated the PR for custom video mode now10:45
mithrojea: Yeah I just saw10:45
jeamithro: cool10:45
mithroReady to merge?10:45
jeayes, I think so10:46
jeamithro: are issues done against your repo or enjoy-digital?10:48
mithrojea: Just put it against my repo for now10:48
*** cr1901_modern has quit IRC10:48
jeamithro: not sure if it is enabled for issues? i can't see a button10:49
jeamithro: ok, its there now. opening issue10:50
mithrojea: I'm going to merge now10:54
jeamithro: thanks for that10:54
mithro_florent_: I don't think the actual changing of the TOFE module is causing the failure/passing.... it just seems to be random10:57
mithro_florent_: It also looks like some non-determinism has crept back into the verilog generation code in some way?10:58
*** cr1901_modern has joined #timvideos10:58
xfxfa wild micolous has appeared11:05
mithroHrm, maybe not...11:07
mithro_florent_: I'm trying to reproduce - but it looks like it might be the lsio_sws....11:07
*** hyades has quit IRC11:11
mithro_florent_: Hrm.. I just disabled the lsio_sws and it start working again11:25
mithro_florent_: I just re-enabled the lsio_sws and memtest started failing again....11:34
CarlFKtumbleweed: http://veyepar.nextdayvideo.com/main/veyepar.cfg  room=plenary_hall11:36
mithro_florent_: I've just disabled it again and rebuilding, if it succeeds, it is a pretty strong smoking gun11:38
mithro(it'll have happened 4 times now)11:38
mithro /join #lca-matrix if you want to talk to the people who set up the IRC rooms11:39
mithroI told faulteh that you'd join #lca-matrix to ask any questions you had about the IRC rooms11:42
mithro_florent_: Okay, it works....11:44
mithro_florent_: There is something in these lines which are causing the memtest failure -> https://github.com/mithro/HDMI2USB-litex-firmware/blob/nextgen/targets/opsis/base.py#L239-L24611:44
tpbTitle: HDMI2USB-litex-firmware/base.py at nextgen · mithro/HDMI2USB-litex-firmware · GitHub (at github.com)11:44
micolousHmm... I can get onto our wifi network from upstairs11:48
micolousHowever my room is right next to the common area.11:48
tumbleweed\o/11:53
xfxfmicolous: please don't log into the router and replace 'ls' with 'sl'11:54
micolousThe program 'sl' is currently not installed. You can install it by typing:11:57
micoloussudo apt install sl11:57
mithro_florent_: After I merged jae's changes - it has just failing with the lsio_sws disabled :(12:00
mithro_florent_: I now have ~15 build output directories12:02
mithrohttps://www.irccloud.com/pastebin/ICN61aol/12:03
tpbTitle: Snippet | IRCCloud (at www.irccloud.com)12:03
micolousI was thinking another option for networking stuff is to run a network cable up the balcony and put an AP on the end of that :P12:11
micolousEven though it's connected, the signal is very sweak and the connection is very laggy12:12
micolousThere's 10s+ latency spikes12:12
xfxfdo what we said, connect to the ressie network, change the gateway ip, works well12:16
xfxf10.217.157.9612:16
micolousok12:16
micolousThat's better, packet loss went away12:22
mithrojea: ping?12:38
mithrojea: How do I convert this to a modeline?12:38
mithrohttps://www.irccloud.com/pastebin/L3K2uWjn/12:38
tpbTitle: Snippet | IRCCloud (at www.irccloud.com)12:38
jea74.250 1280 1390 1430 1650 720  725  730  750 +hsync +vsync12:39
jeaactually no12:39
jea74.25 1280 1390 1430 1650 720  725  730  750 +hsync +vsync12:39
jeatry that12:39
jeamithro: I don't think it works with 3 decimal places on the dotclock at the moment12:40
jeaso the second should work12:40
mithroRUNTIME>video_mode custom 74.25 1280 1390 1430 1650 720  725  730  750 +hsync +vsync12:40
mithroParsing custom mode...12:40
mithroCustom video mode set.12:40
jeayay12:40
mithrojea: What should happen after I do that?12:41
mithrojea: Do I have to change to it or something?12:41
mithrojea: Is there a way to get it to print out what it thinks the current mode line is?12:42
jeamithro: that should be it. status will sort of show you, and the edid will also have it12:42
jeamithro: nothing in there at the moment to get the current mode line12:43
mithrooutput0: [email protected] from pattern12:43
mithrojea: Do you have a working output on your opsis with your firmware?12:53
jeamithro: I haven't tried many modes to see if they work properly or not. i know it parses and sets edid and stuff correctly, but haven't fully checked the input12:58
mithrojea: But at the moment, your have firmware on your computer which has your code stuff and is currently outputting things?12:59
mithrojea: What about his mode?13:06
mithrohttps://www.irccloud.com/pastebin/Dkh5obdb/13:06
tpbTitle: Snippet | IRCCloud (at www.irccloud.com)13:06
jeamithro: 74.25 1280 1720 1760 1980 720  725  730  750 +hsync +vsync13:07
mithro_florent_: http://hopper.mithis.com/~tim/all-the-tests.tar.bz213:15
micolousxfxf: https://pbs.twimg.com/media/C115cv8XUAI2D5T.jpg13:28
jeawow13:55
*** CarlFK has quit IRC14:18
*** rohitksingh has joined #timvideos14:30
*** rohitksingh has quit IRC14:31
*** CarlFK has joined #timvideos14:42
*** ChanServ sets mode: +v CarlFK14:42
*** kflux has joined #timvideos15:59
kfluxhi guys16:05
kfluxi have the numato opsis with all the extension boards16:06
kfluxi am not a developer though16:06
kfluxi supported the project hoping for a videomixer firmware for it16:06
kfluxi read that there was a SoC project to make such a firmware16:06
kfluxi was wondering what the status is on that16:07
kfluxwhat is the current funtionality of VideomixerSoC and how to i put it onto my opsis16:08
kfluxthe "documentation" is very confusing for non devs16:48
*** kflux has quit IRC18:29
*** kflux has joined #timvideos20:09
*** nrossi has quit IRC20:20
*** its_only_me has joined #timvideos20:46
its_only_mehello20:47
its_only_mequick question about crowdsupply campaign: it is a great device and i wonder what is throughput of that expansion slot. I would like to do hdmi switch 2x8 for example (routing 2 signals between 8 outputs), do you think it is possible ?20:49
tumbleweedCarlFK: o/22:12
micolousYou're awake? I'll come down then22:13
tumbleweedyep22:13
*** hyades has joined #timvideos22:13
micolousI've just been trying to get the YT stuff running again, without success. I think there are rate limits on this network because "uploading to youtube" has the same traffic profile as "uploading to p2p"22:14
micolousits_only_me: there's one person who can answer that with certainty, and I think he's asleep22:15
*** CarlFK has quit IRC22:58
tumbleweedxfxf: will you try to find out the story with the network?23:12
*** its_only_me has quit IRC23:16
xfxftumbleweed: emailed him23:19
mithrokflux: Sorry about that, the board is really only suitable for developers/power users at the moment, we are slowly working towards making it easier for non-developers but it still needs a lot more effort23:55

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!