added Authorization header if access_token provided

[kaseyhackspace]

This fixes issue #952 by adding an Authorization: token <access_token> header to Github API requests from Binderhub if access_token is passed by either binderhub_config.py or GithubRepoProvider in secret.yaml.

[betatim]

Thanks for the PR!

I think @sgibson91 has been looking into this as well so getting your input here would be great. I've never even tried to use the features of a BinderHub for "private" deployments so don't have a good feeling if this would break other things or if it should have been solved some other way.

[betatim]

This change seems to have broken one of the tests here (sorry for the very verbose output of our CI).

[betatim]

There is also #940 to look at/compare to/check with.

[kaseyhackspace]

This change seems to have broken one of the tests here (sorry for the very verbose output of our CI).

@betatim I was actually looking into this, and it seems to be a weird point of failure from the CI test. The failure is from a null result of getting gitlab ref. The code changes I made were inside the class GitHubRepoProvider which should not be related to GitlabRepoProvider.

Is there a way to re-trigger the travis-ci tests?

[betatim]

You think this is a flakey test? I restarted it now. I thought it was related to the change because something in the GitLab provider broke.

[kaseyhackspace]

You think this is a flakey test? I restarted it now. I thought it was related to the change because something in the GitLab provider broke.

Looked more into this and the failure was also the same as PR #951 so I looked at the actual test. I think we are getting failures because the repository we are testing against was updated from gitlab-ce to gitlab-foss. I did another branch on my fork with an updated repo name and tried a PR there. Waiting for the test to finish to see if that fixes that issue. Will keep you updated here 😄

[kaseyhackspace]

@betatim so the ci tests have finished for the new branch I made that rectified the test_gitlab_ref case that was failing and it is now passing (https://travis-ci.com/kaseyhackspace/binderhub/jobs/236599088#L3086). This branch should fix both this PR and the error in PR #951. Question now is, how do we proceed:

• Do I PR the branch (https://github.com/kaseyhackspace/binderhub/tree/update-test-repoproviders) to jupyterhub/binderhub's master?
• should I merge and push the branch to this current PR?

[betatim]

Ah, I just opened #956 to fix the failure after looking at a different PR (but not this one) during my morning coffee. Sorry for the double effort. If you could take a quick look at #956 and leave a review we could merge that. After that I'd rebase this PR and #951 onto master.

[kaseyhackspace]

Ah, I just opened #956 to fix the failure after looking at a different PR (but not this one) during my morning coffee. Sorry for the double effort. If you could take a quick look at #956 and leave a review we could merge that. After that I'd rebase this PR and #951 onto master.

It's all good @betatim , it's never a wasted effort 😄 #956 is good to merge.