Monday, 2013-09-02

*** tpb has joined #timvideos00:00 [freenode-info] channel trolls and no channel staff around to help? please check with freenode support:
*** mithro- has joined #timvideos02:04
*** mithro- has quit IRC03:01
*** hyades has joined #timvideos03:08
*** hyades has quit IRC03:41
mithroSewar, ping?04:29
Sewarmithro, pong04:30
mithroSewar, you linked me to the edid-grabber-in-c blog post04:30
mithrobut that seems to be from the 19th August?04:30
Sewaroops, same blog title/url04:31
tpbTitle: Mohammed Alhadab (at
Sewarmithro, iiie, CarlFK, parx:
tpb<$B> (at
mithroAh ha04:33
mithroSewar, I'd recommend using mingw rather then Visual C++04:33
mithroit's a bit harder to get set up04:33
mithrobut then you can cross compile under linux04:33
mithroand you never have to touch windows except for testing04:34
Sewarisn't mingw for use IN windows? how can I compile from linux then?04:37
mithrocygwin is for using Unix in Windows04:37
mithromingw is a gcc compiler for targeting windows04:37
*** hyades has joined #timvideos04:38
Sewarso mingw can be installed in linux to compile for windows?04:39
Sewarthat part is hidden in
tpbTitle: Welcome to | MinGW (at
Sewar"Using MinGW for Cross-Hosted Development"04:40
Sewarmithro: what do you recommend for compiling script (to handle multiple platforms, etc)? make? automake? configure? I never used any before so I have no idea where to look04:58
mithrofor something as compilicated as your code, just make04:59
mithroSewar, only problem I can see automake helping with would be AMD64 verse i686 stuff05:04
SewarI see05:07
*** iiie has joined #timvideos05:26
*** mithro- has joined #timvideos05:38
*** mithro- has quit IRC05:45
*** mithro- has joined #timvideos06:06
hyadesmithro: mithro- how do I send both python and c coverage reports to coveralls?06:20
mithro-hyades, I wouldn't recommend doing that06:21
mithro-hyades, have two build targets06:21
mithro-one for the C and one for the python06:21
hyadesoh fine06:21
mithro-hyades, will explain more when I'm not in a meeting06:21
hyadesi got it.06:22
*** mithro- has quit IRC07:07
mithrohyades, okay I have some time to explain further if you want07:14
mithroyou've read up about the build-matrix stuff that travis provides?07:28
hyadesi was going through it, but not very clear about it07:30
mithrohyades, so the idea is that you have multiple different environments you want to test against07:31
hyadesso the test runs multiple times?07:31
mithroIE for web-animations project I work on at work07:32
hyadesmy whole build is divided into, first installing and then building07:32
mithrowe want to test under the major browsers07:32
hyadessorry installing and then testing07:32
mithroeven though we want to run the same tests, we want to run them against Firefox, Chrome, etc07:33
mithrowe also want to run the lint and style checks separately07:34
mithroso this is what our travis.yml looks like07:34
tpb<> (at
hyadesso, the install and script are run for each of your browsers?07:35
mithroeach one of the lines under matrix: is a different environment we want to run under07:35
mithrowe then have a set up script (because it got to complicated to just keep in .travis.yml) which downloads the correct browser07:36
tpb<> (at
mithroit looks at the environment variables set via the matrix07:36
mithrolastly we have a run script07:36
tpb<> (at
mithrowhich against looks at the environment and runs different things07:37
hyadesyes :)07:37
mithrohyades, so you can see in the run script, if MODE=check we just do the linting, else we run the tests07:37
mithroso you'll probably want something which goes07:37
mithro  TESTING=api  MODE=lint07:38
mithro  TESTING=api  MODE=test07:38
mithro  TESTING=project  MODE=lint07:38
mithro  TESTING=project  MODE=test07:38
mithro(I'm unsure if "project" is a good name - but something like that)07:38
mithrohyades, that make any sense?07:39
hyadesyes, got it07:39
mithrohyades, there might not be a lint test for gst-switch core yet07:39
mithromaybe TESTING should be "MODULE" or something?07:39
hyadesthere is a lint test i guess07:40
hyadesmithro:  any hint on why i get the TypeError here07:41
hyadesin the end07:41
mithrohyades, so where are we at?07:53
hyadesthis TypeError doesnt come when I do the same in my quantal VM07:54
mithrohyades, probably an older version of gi.py07:54
mithrohyades, what is our current state, what goals do we have for this week?07:54
mithrohyades, and since we have 2 weeks left, what are we going to consider "success"?07:55
hyadesthe gcov part, I had edited the c file to generate coverage files when it receives SIGUSR107:55
hyadesI have a coverage of c files.07:59
hyadesthough its not very great07:59
tpbTitle: Ubuntu Pastebin (at
mithrohyades, I think the travis environment is Ubuntu Precise08:05
hyadesthe Gio docs i had found were for 12.10,08:06
tpb<> (at
hyades - html of the coverage08:07
tpb<> (at
mithrohyades, can you upload it somewhere I can actually view it?08:08
tpbTitle: LCOV - (at
hyadesthere maybe a better report, I will send you by evening08:11
hyadesyes, this report has some problem. Will send you again.08:15
mithro<mithro> hyades, what is our current state, what goals do we have for this week?08:23
mithro<mithro> hyades, and since we have 2 weeks left, what are we going to consider "success"?08:23
hyadescurrent state is that you can try out the build using the travis steps atleast on a quantal machine08:25
hyadesit will work08:25
hyadesthis week, should be to improve the coverage of integration test in both the api and c code08:27
mithrobut we don't have coverage ending up on coveralls yet?08:28
mithroso first I think we want to get the split travis build and coveralls upload stuff first?08:28
hyadesi will do that by today08:29
hyadesso coveralls will show two coverage reports?08:29
hyadeswhat else should be done?09:00
hyadesand also "success" would mean what ?09:02
mithrowell, that is what we need to figure out09:02
mithroour original goal was to09:03
mithro a) create a Python API for gst-switch09:03
mithro b) Use the Python API to make a reliable test-suite for gst-switch09:03
hyadesit will test the major things in gst-switch controllable over dbus09:04
mithroSo for a)09:06
mithroI'd like to see the following09:06
mithro(these might all be finished already)09:06
mithroOn every change, travis runs09:07
mithro  * gst-switch Python API unit test coverage09:07
mithro  * gst-switch Python API integration test coverage09:07
mithro * The unit test coverage should be 100%09:07
mithroAnything I missed there?09:09
mithroWe should also move towards having a always "green master"09:10
hyadeslooks fine. (readthedocs maybe also)09:10
mithrohyades, yes - we should be publishing documentation automatically too09:10
mithrohyades, let me put this in a document somewhere :)09:10
tpb<> (at
hyadesthe lint checker runs when you edit a c file. It checks if the code style is fine.09:24
mithrohyades, I'm not sure what you mean?09:25
hyadesthere are a couple of scripts in scripts/apps09:26
tpb<> (at
tpb<> (at
mithrohyades, how does a person run them?09:27
hyadeswhen you commit code, it runs these tests09:28
hyadesit doesnt commit unless you have resolved the issues09:28
mithrohyades, okay09:28
mithroI'm pretty sure you can override that09:29
mithrohyades, how does that look?09:31
hyadesCode style okay!09:31
hyadeson success :)09:31
tpb<$> (at
mithrohyades, can you think of anything which I've missed?09:34
mithrodoes the second section look okay?09:35
hyadesCreate a "how to write tests for gst-switch" document09:36
hyadesusing the API?09:36
hyadeslooks good09:37
hyadesregarding the master branch09:37
hyadessome dbus variables need to be changed for master branch. Same code will work, parameters like the interface, object path should be provided. Defaults wont work.09:37
mithroso, we can redefine what the master branch is09:37
mithroI think it probably makes sense to make master the branch you are working off of09:38
mithrohyades, so I assume we still haven't started the09:40
mithroTests should exist which exercise the major parts of gst-switch09:40
hyadeswell, i need clarification here09:40
hyadesa code which tests PIP modes tests gst-switch functionality?09:41
hyadessame for other methods09:41
mithrohyades, code which tests all the functionality09:45
hyadesmithro: "Remove the currently failing, unreliable tests." which tests are these?09:55
mithrohyades, the ones in the tests directory09:55
hyadesthe one which runs by doing make demo?09:56
hyadeslast time I tried them, they were working I guess09:56
mithroI don't think they work reliably09:58
mithrohyades, some performance tests would be good too10:01
hyadesI have a parameter in most of my integration tests called FACTOR. increasing its value will lead to loading the system.10:03
hyadessystem and gst-switch10:03
mithrowhat does that mean?10:03
tpb<> (at
mithrohyades, "FACTOR is badly named"10:04
mithrohyades, it should be something like "number of connections"10:04
mithroThere are multiple ways to increase the load10:05
mithro a) increase the number of input streams10:05
mithro b) increase the resolution of the input streams10:05
mithro c) cause lots of PIP transitions10:05
hyadesa) and b) can be handled by changing NUM and FACTOR10:07
mithrohyades, you shouldn't have to change anything10:07
mithrohyades, and those are really bad names10:07
mithrohyades, and not really documented?10:07
hyadesyes, will change names10:07
hyadesbut,what should I confirm by doing these tests?10:08
mithrohyades, what do you mean?10:08
hyadescannot do frame comparisons for these tests. It takes lots of time10:08
mithro"torture tests" probably shouldn't run on travis10:09
hyadeseven on my laptop, each frame comparison takes around 1.5 secs10:09
hyadescannot do it for something like rapidly changing PIP modes10:10
mithrohyades, the comparision should happen *after* the test is finished10:10
hyadesokay, fine!10:11
mithroIf you change 100 times within 1 second, that is till only a couple of minutes to check the result10:11
mithrohyades, torture tests are about making sure the program still works under heavy load10:11
hyadesregarding multiple times a second. The gst-switch-srv crashes if the time between two PIP mode change requests is less than 0.6 secs (0.5 it crashes)10:12
hyadescrash = seag fault10:12
mithrohyades, well that is a bug!10:12
mithrohyades, and exactly the type of bug we want to find10:13
mithrohyades, so - your working as expected :)10:13
mithrohyades, oh10:14
mithroone bigger thing I only just remebered10:14
mithrohyades, I'd really like to see if we could migrate from the script/ thing to using gnome's jhbuild10:15
mithrohyades, as I think that would make it more reliable/reproducable10:15
hyadesi dont use stage script anymore10:15
mithroand less stuff for us to maintain10:15
hyadesjust a autogen is enough10:15
mithrohyades, ?10:15
mithrohyades, how do you get the dependencies?10:16
mithrohyades, or is this on travis?10:16
hyadesyes it is10:16
hyadesin the .travis.yml file10:16
hyadesone patch is required10:16
tpb<> (at
hyadesthis path should be the one specified in the prefix argument to autogen.sh10:17
mithrohyades, both the original and your patch are wrong10:18
hyadesyes, i know10:18
hyadesi need the prefix there10:18
mithroI don't know automake/configure to give you the right value there10:18
mithrobut I'm sure there is a help channel here on IRC that could help you out10:19
hyadesso is jh-build still needed?10:25
mithrohyades, yes10:28
mithrohyades, I'm not sure how you are building gst-switch without jh-build?10:28
hyadesI am first installing all dependencies, then installing all gstreamer + plugins10:29
mithroinstall all gstreamer + plugins, you mean?10:29
mithrodownloading from git and compiling them?10:30
mithrohyades, so you manually reproduced what the stage script use to do?10:30
hyadesthe stage script doesn't provide the PKG_CONFIG_PATH10:31
hyadesthe stage script needs to be edited.10:32
mithrohyades, your welcome to edit it10:34
mithrohyades, but I think we should dump the stage script for jh-build - I think that would more reliably build the enivornment we need and is better maintained10:36
mithroI'm going to head home so I can do the washing I didn't do on the weekend. I'll be back in about 30 minutes.10:37
hyadesI will look into it :)10:37
*** mithro- has joined #timvideos11:14
mithro-hyades, back now11:31
mithro-but my internet seems stuffed11:31
mithro-hyades, so are you happy you know what to do?11:32
*** mithro- has quit IRC11:47
*** mithro- has joined #timvideos11:47
*** mithro- has quit IRC11:53
*** hyades has quit IRC12:01
*** mithro- has joined #timvideos12:10
*** mithro- has quit IRC12:51
*** hyades has joined #timvideos12:55
*** mithro- has joined #timvideos13:08
*** mithro- has quit IRC13:39
*** hyades has quit IRC13:40
*** hyades has joined #timvideos14:04
*** hyades has quit IRC14:24
*** hyades has joined #timvideos16:23
*** hyades has quit IRC16:44
*** hyades has joined #timvideos16:55
hyadesmithro: gcov reports for c files after doing all the tests -
tpbTitle: LCOV - out.infotools (at
hyadesaround 70% coverage (Rest of the c files are not executed)18:06
hyadesirc logs no longer work?19:04
hyadesmithro: the docs will be hosted here
tpbTitle: Welcome to gstswitch’s documentation! gstswitch documentation (at
mithrohyades, IRC logs should still work?23:48
mithrohyades, they where still being logged just the website wasn't updating23:51
mithrofixed now23:51
mithrohyades, so which files are not tested?23:52

Generated by 2.12.1 by Marius Gedminas - find it at!