In #1349 we added a script to dump documents from a running instance of CatColab on AWS infra. The data is in some sense incidental to the goals of the change: ensure that semantic-preserving changes are reservable and preserve document validity.
The proposal is to leverage the typescript api itself, or failing that, the lower-level document methods to randomly generate valid notebooks as the data source for these tests.
This will allow us to avoid:
- gating running these tests on AWS access
- committing a blob snapshot of the data source to the repo
- potential desync between the nature of the data and the current changes to the API
There appear to be medium-low effort ways to do this (eg: randomly picking a shape, if it requires references to other cells pick among the generated one or make news, etc) and this should be investigated before we make concrete decisions about wiring these tests into the CI.
In #1349 we added a script to dump documents from a running instance of CatColab on AWS infra. The data is in some sense incidental to the goals of the change: ensure that semantic-preserving changes are reservable and preserve document validity.
The proposal is to leverage the typescript api itself, or failing that, the lower-level document methods to randomly generate valid notebooks as the data source for these tests.
This will allow us to avoid:
There appear to be medium-low effort ways to do this (eg: randomly picking a shape, if it requires references to other cells pick among the generated one or make news, etc) and this should be investigated before we make concrete decisions about wiring these tests into the CI.