Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Charles Cross
    @spiderkeys
    there is one difference, being that the diagnostics one places the webcomponent in a div
    Dominik Fretz
    @codewithpassion
    that's where I started
    one difference in appearance is as well that the diagnostics one show like this:
    blob
    see the white border around the navbar
    Charles Cross
    @spiderkeys
    Yea, thats weird.
    Dominik Fretz
    @codewithpassion
    ok, I'll finish the UI first
    then look at why the weird placement
    don't worry
    Dominik Fretz
    @codewithpassion
    so, I found out what the reason for the white border is
    <style include="foundation5-styles"></style>
    <style include="normalized-styles"></style>
    if I add those to the configurator the same thing happens
    Charles Cross
    @spiderkeys
    Where do those come from? are those hidden somewhere inside a core polymer element?
    Dominik Fretz
    @codewithpassion

    <dom-module name="orov-settings-manager">

    <template>

    <style include="foundation5-styles"></style>
    <style include="normalized-styles"></style>

    very top
    Dominik Fretz
    @codewithpassion
    although in the Diagnostics component it's not there and it has the same behaviour
    a bit wiered
    badevguru
    @BrianAdams
    Ah. The normalized-styles was an experiment, I think that is where the extra border is coming from.
    Dominik Fretz
    @codewithpassion
    @BrianAdams was talking to @spiderkeys about the layouting. how/why was foundation5 choosen?
    anyway, we should choose either foundation or bootstrap
    badevguru
    @BrianAdams
    So the point is that the infrastructure does not require any framework. Each applet/theme/control is completely isolated form one another. So I don’t really care what the author uses, as long as they can get it to look like they want. Now… for those that we at OpenROV are supporting… why yes, we can choose a general layout standard so that we minimize what we have to support and learn. I picked foundation for the JSON edit contorol because it looked the best of all of the frameworks when applied to his control which is all that is in the settings applet. Feel free to advoate for your favorite.
    Dominik Fretz
    @codewithpassion
    ok
    where did you get the foundation webcomponent from?
    as foundation updated to V 6
    of their framework
    Bran Sorem
    @bransorem
    Foundation is pretty nice. Check out Bulma for another option (flexbox based): http://bulma.io/
    Charles Cross
    @spiderkeys
    might be interesting to try out: https://code.facebook.com/posts/1840075619545360
    badevguru
    @BrianAdams
    Yarn looks interesting. They claim to bundle in the bower dependencies in to their workflow as well, through I am still looking for actual explanantion in the docs. I cannot help but think “shots fired” with NPM. As for us, they claim it is copying from the global cache to local repo instead of linking. Will need to see if that behavior could be changed.
    Charles Cross
    @spiderkeys
    npm blog seemed pretty supportive of it: http://blog.npmjs.org/post/151660845210/hello-yarn
    badevguru
    @BrianAdams
    Yea, that is the public face. But this is ultimately a response to NPM being stagnet and leaving lots of bugs in the core features of their client. I suspect NPM will shift a large amount of their effort in fixing the NPM solutions as a result. Just my prediction.
    Charles Cross
    @spiderkeys
    global cache sounds like a good thing for image building and desktop development though. wouldn't really help when deployed on actual hardware, but if the build server has cached global dependencies, seems like it could massively speeed up cockpit installation times
    badevguru
    @BrianAdams
    Could be.
    Dominik Fretz
    @codewithpassion
    Safari 9 does not understand const and let
    Bran Sorem
    @bransorem
    Yarn is great, but not production ready. I've heard of a case where it blew up bower deps (removing and breaking some). I'd wait for a little more stability - probably won't be long
    Charles Cross
    @spiderkeys
    @bransorem Yea, that is what I experienced as well when I tried to experiment with moving cockpit's install to yarn. It was super fast, but there were a couple of packages that weren't happy with it, particularly projects that used node-gyp, like node-serialport
    Dominik Fretz
    @codewithpassion
    Dominik Fretz
    @codewithpassion
    Hey!
    @BrianAdams do we yet use a centralised logging service for the web services? Somthing like logz.io, loggly.com or somehting simillar? preferably something that already has a winston transport?
    badevguru
    @BrianAdams
    No. Have not done anything with the services yet. Had been working on a local version so that we did not get killed by the service fees of the hosted services. Have a prototype so far that I am sure has a winston transport. Backed be elasticsearch+kabana
    I suspect there are several docker images (that is what I was using) that have been updated since then that allow a simple docker run to get it up and going.
    Dominik Fretz
    @codewithpassion
    is that deployed anywhere on our infrastructure?
    so that OpenExplorer could log there?
    badevguru
    @BrianAdams
    Probably. If you can give me till sometime tomorrow, I can make sure its all up and running correctly. Been a while.
    Dominik Fretz
    @codewithpassion
    :thumbsup:
    Dominik Fretz
    @codewithpassion
    Regarding centralised logging, did you see elk: https://hub.docker.com/r/sebp/elk/
    It's one image to combine elastisearch, logstash and kibana
    Dominik Fretz
    @codewithpassion
    5 or 10$/month should do espeacially if we're using it primaraly for log
    badevguru
    @BrianAdams
    Yea, that is basically what I stood up, the docker instance with a pattern for upgrading. My preference would be for the docker version only because I don’t know what the pattern is for keeping the one-click-apps version up to date. Otherwiser we can go either way.
    Charles Cross
    @spiderkeys
    Storing this here for now until I codify it in a repo somewhere:
    # Stream 10s of video to a file from mxuvc
    sudo mxuvc stream --time 10 --vout1=input.h264 --stats
    
    # Stream indefinitely to a file from mxuvc
    sudo mxuvc stream --vout1=input.h264 --stats
    
    # Convert raw geo stream h264 to fragmented mp4 (1 frame per box!)
    ffmpeg -r "30" -i test.h264 -acodec copy -vcodec copy -f mp4 -frag_size 1 -movflags frag_keyframe+empty_moov output2.mp4
    
    # Trim the preceeding P frames from a raw geo h264 stream so that the SPS/PPS frames are the first!
    ffmpeg -re -i input.h264 -acodec copy -vcodec copy -f h264 output.h264
    
    # Tool for analyzing h264 frames
    https://github.com/shi-yan/H264Naked
    
    # Tool for analyzing mp4 segments
    http://download.tsi.telecom-paristech.fr/gpac/mp4box.js/filereader.html