Make sure that the v2 deploy function works well from a single spec file
So rather than deploying what's expanded on the filesystem, we can deploy from a single spec file.
If the repo already has a project state yaml file, it should use this to find the endpoint and stuff. I guess precedence is:
- accept endpoint and project id from args
- load endpoint and project id from alias or UUID (works if this isn't the first deploy)
- load endpoint and project id from openfn.yaml (maybe? feels a bit loose?)
- accept endpoint and project id from env
v1 deploy should not work with the v2 format. We should probably add a nice error
Make sure that the v2 deploy function works well from a single spec file
So rather than deploying what's expanded on the filesystem, we can deploy from a single spec file.
If the repo already has a project state yaml file, it should use this to find the endpoint and stuff. I guess precedence is:
v1 deploy should not work with the v2 format. We should probably add a nice error