The collection of test fixtures like cosmological_volume (run test for all example snapshots and all access methods), cosmological_volume_only_single (run test for all access methods on an up-to-date, single-file snapshot example) and cosmological_volume_only_single_local (run test for local filename access method and same single file as previous) have grown pêle-mêle over time and it's not clear that it's useful to run all tests (some of which are quite expensive) on as many cases as they are currently set to run on. It would be a good idea to:
- Revise the available fixtures to systematically offer logical combinations. The naming could also be better so that there is some consistency between e.g. SOAP fixtures and snapshot fixtures (and also SOAP-or-snapshot fixtures).
- Systematically review which tests run on which test cases. This needs some care since I've been surprised by some of the bugs that have been caught by running the tests on cases that I had assumed would cover the same possible pitfalls but actually had subtle differences. But I think we're probably running e.g. the remote file access case on more tests than is necessary.
The collection of test fixtures like
cosmological_volume(run test for all example snapshots and all access methods),cosmological_volume_only_single(run test for all access methods on an up-to-date, single-file snapshot example) andcosmological_volume_only_single_local(run test for local filename access method and same single file as previous) have grown pêle-mêle over time and it's not clear that it's useful to run all tests (some of which are quite expensive) on as many cases as they are currently set to run on. It would be a good idea to: