Ask quick dev questions about mybinder.org, BinderHub, and repo2docker. Use discourse.jupyter.org for user questions, support, and discussion.
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'.)
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:
it looks there like the ipywidgets libraries haven't been loaded
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.)
(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.)
@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
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.
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.
(It may even have been an environment branch of the same repo, not separate repos actually)
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).
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
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
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
we use webpack to build the js and css bundle used in the websites binderhub serves
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.
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
@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
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.
there is also https://github.com/betatim/vscode-binder/