Automatically deploy pull requests #472
Conversation
|
This pull request has conflicts ☹ |
This patch automatically deploys pull requests with static mock data to https://test.admin-interface.opencast.org similar to how the test deployment of the Opencast Editor works. Unfortunately, we cannot set a path prefix for mock data which means that all deployments share the lattest set of mock data. But since they shouldn't change very often, that's probably okay for this to get started.
|
Use Local test with mock data Proxy data from develop.opencast.org It may take a few seconds for the interface to spin up. |
There was a problem hiding this comment.
The workflow reads reasonably, mostly because it reads very similar to the respective editor workflow, so we should be good.
Seeing as how this would be quite difficult to test without access to the required github secrets, how about we go ahead and merge this? We can still hotfix it if anything breaks.
Also this is not necessarily a change request, but why do we not do a opencast developer/committer membership check like in the editor?
Because this cases the code from a pull request to be checked out and executedin an environment which has access to secrets like the private deployment key. A malicious actor can steal that. A malicious actor can aalso use this to publish any code, binary, … under an opencast.org domain. Again, not great. GitHub has a limited built-in protection in that this will only run automatically for users which have an accepted contribution merged into the default branch. But that also means a contributtion from 10 years ago with no way for us to remove that user even if we know that their account has been compromised. |
+1 from me. I justt verified tthat #471 did not break this. We already discussed this PR on Tuesday. If you waant to merge tthis, please go ahead. |
This patch automatically deploys pull requests with static mock data to https://test.admin-interface.opencast.org similar to how the test deployment of the Opencast Editor works.
Unfortunately, we cannot set a path prefix for mock data which means that all deployments share the lattest set of mock data. But since they shouldn't change very often, that's probably okay for this to get started.