by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 07:18
    adamhsparks opened #898
  • May 24 15:43
    choldgraf commented #1445
  • May 24 15:19
    consideRatio commented #1445
  • May 23 17:14

    choldgraf on master

    Add an issue template for reque… Rephrase Add numbers and 5 more (compare)

  • May 23 17:14
    choldgraf closed #1447
  • May 23 17:14
    choldgraf closed #1446
  • May 23 17:14
    choldgraf commented #1447
  • May 23 16:24
    GeorgianaElena synchronize #1447
  • May 23 15:48

    betatim on master

    binderhub: https://github.com/j… Merge pull request #1448 from h… (compare)

  • May 23 15:48
    betatim closed #1448
  • May 23 15:32
    henchbot opened #1448
  • May 23 15:20

    betatim on master

    add bs4 and pytest to requireme… replace assertRaisesRegexp with… remove pytest from requirements… and 4 more (compare)

  • May 23 15:20
    betatim closed #1104
  • May 23 15:19
    betatim commented #1104
  • May 23 13:55
    betatim commented #1102
  • May 23 13:48
    betatim synchronize #1102
  • May 23 10:55
    kpaschen synchronize #1104
  • May 22 22:06
    manics commented #994
Nathan Weeks
@nathanweeks
https://grafana.mybinder.org is working, though it seems the web server is configured to disallow displaying in an iframe from another domain (per the following error in the chrome developer console: status.html:1 Refused to display 'https://grafana.mybinder.org/d-solo/fLoQvRHmk/status?panelId=6&orgId=1&tab=general&theme=light' in a frame because it set 'X-Frame-Options' to 'deny'.)
Axel Dürkop
@xldrkp
2020-05-13_15-52.png
Hi everybody, I try to get widgets working in a JupyterBook, but the cell with the widget throws the following error when I click the "Run" button:
Chris Holdgraf
@choldgraf
Ah yeah - widgets are a bit tricky right now, because the way jupyter handles front-end javascript libraries needed for its packages it a bit complicated
it looks there like the ipywidgets libraries haven't been loaded
Nic30
@Nic30
Hello, is there any way how to prebuild mybinder docker image? I would like to automatically build a image after each commit or tag from travis-ci.
(I would like to use mybinder for interactive tutorial, however the image takes too long to build and it does not look for new users of my lib.)
Martin
@hackfin
@Nic30 how about the docker hub services/auto-builds?
Sarah Gibson
@sgibson91
@Nic30 you can use repo2docker (https://repo2docker.readthedocs.io) to build the container in your CI. The issue is that the main docker registry mybinder.org uses is private so you won't be able to push your image there, therefore, mybinder.org won't be able to find it and will just rebuild it anyway
Nic30
@Nic30
I was thinking that the binder requires it's own credentials and name of the image, I guess I can build an another image, I currently do not know how to use it directly but will find out. I was just curious if there is some rest api or something where I can just trigger the build using wget or something.
@sgibson91 the rebuild of the image is exactly what I am trying to avoid.
Sarah Gibson
@sgibson91
You could pin your binder link to a GitHub tag rather than master branch so that it won't rebuild unless you move the tag?
Nic30
@Nic30
I was thinking about it, but it seems to me that I will forget to update links quite often and that is why I would like to avoid it.
Martin
@hackfin
If you do frequent updates, you could also use some bootstrap second stage approach that pulls the latest NBs from inside the Binder, to avoid rebuilds. But it would be a nice feature to just be able to run any prefabricated container from Dockerhub instead of having to inherit 'FROM' it and rebuild. But you can always try squashing that last layer.
Sarah Gibson
@sgibson91
I've seen examples where the Binder environment is hosted in one repo, the notebooks are in another, and then nbgitpuller is used to pull the notebooks into the Binder environment. That way, rebuilds are only triggered when you update the environment, not the notebooks
(It may even have been an environment branch of the same repo, not separate repos actually)
Nic30
@Nic30

I would like to have content of repo same as content of repo in docker image.

The best way how to achive this (which I know about) is to publish docker image from CI on every sucessful test pass and then extend this image in binder image (because I do not know how to specify docker image without dockerfile).

Tim Head
@betatim
we try to arrange the layers in the docker images on mybinder.org so that content changes should not need a full rebuild if you are rebuilding a lot
it is all a bit of a trade off though as we can't keep all previous layers for all repos in the cache on all nodes of the cluster
you can trigger a build by visiting a repo's "launch URL" and the build will continue in the background so you don't have to watch it build
Tim Head
@betatim
you can automate that call on merge-to-master if you like to. We'd ask that you use the power to trigger rebuilds by API sparingly though :) If too many people trigger too many rebuilds too many times we will have to implement some kind of rate limiting/build budget per repo to make sure the build queue doesn't get monopolised by those who automate builds
:D
overall what works well for people is to use repo2docker locally for super fast iteration cycles while developing things (it is most likely much faster than triggering a build on mybinder.org as it has the full docker layer cache)
and then use a combination of manually triggering a new build when they have "a bunch of changes" or using a tag that they move when there is a "new release" of materials for their blog/class/workshop
and for super popular repos that are constantly being launched there is no need to automate stuff because your users do all the work for you :D
Nic30
@Nic30
@betatim I do not plan to break binder with too many build requests, I am also planning to "build slow patrt of image elsewhere" as I see that binder is bussy sometimes.
Tim Head
@betatim
cool :D
Erik Sundell
@consideRatio
@betatim sorry for the self-merge
BinderHub dev question - what is webpack.config.js etc doing? Serving static content? Is the binderhub server also doing that? I'm confused in general what the python package binderhub does and what the node package does
Tim Head
@betatim
i dont think there is a node package
we use webpack to build the js and css bundle used in the websites binderhub serves
Michael Patterson
@iampatterson

Just

@iampatterson the config-file? There is a lot of custom stuff on it, any specific part you think may be the culprit here?

Sorry for not catching this earlier, I am jammed up a bit. Did you get your problem fixed?

Has anybody have nbinteract working on a private Git repo via Binder? I have a client that I would love to use this with but I can't make the repo public because of the data I will be interacting with. Any help would be appreciated. The code works if I make the git public.
Michael Patterson
@iampatterson
Sorry I found the answer. I will let you know how it goes using google
Michael Patterson
@iampatterson
So I ended up doing this. Kinda the opposite side of binder, unless there is a way of using private gits this is the next best thing I could find.
Daniel Shapero
@danshapero
Hi all -- I'm trying to use Binder for some demonstrations in a class, but I depend on a pretty big library (PETSc)
I already have a docker image with all my dependencies built and I was able to launch a binder instance following the instructions on using custom docker images -- it was amazing how easy all this was
the one thing I'm wondering about -- do I need to manually pull in all the demonstration notebooks into that image? they don't appear in the binder instance for my repo, but it looks like whatever notebooks are in your repo do get pulled in automatically if you can get away without using a custom docker image
Sarah Gibson
@sgibson91
Hi @danshapero I would put your notebooks into the repo that has the Binder custom Dockerfile, I think they should be pulled in then
@iampatterson there's this documentation for accessing private repos from your own BinderHub deployment https://binderhub.readthedocs.io/en/latest/zero-to-binderhub/setup-binderhub.html#accessing-private-repositories it's one level up from z2jh-k8s
Jan Janssen
@jan-janssen
Can I make a suggestion for https://github.com/binder-examples ? I would like to add https://github.com/jan-janssen/code-server-example which demonstrates how to use code server (similar to visual studio code) on mybinder.
Arnim Bleier
@arnim
@jan-janssen I got a "500 : Internal Server Error The error was: could not start nbvscode in time"
Sarah Gibson
@sgibson91
Another one of @betatim's awesome repos that should be copied across to binder-examples!
Arnim Bleier
@arnim
;)
Tim Head
@betatim
yes and no, vscode-binder is simultaneously the "source" of the package you need to use if you want to do this and an example of how to use it. so maybe we should split it up and add the "how to use it" part off it as a binder-examples repo
Sarah Gibson
@sgibson91
:+1:
Jan Janssen
@jan-janssen
@betatim That would be great. Having vscode inside jupyter lab is really helpful for us as scientists because it allows to shift the software development to the login node of the HPC.