Currently, test discovery is either ran from scratch or a cached instance is used. There's checks in place to try to clear the cache. The cached test discovery is also not leveraged for --collect-only
There could be a cached test discovery. When new test functions are added / renamed / removed, the cache could be updated.
The primary beneficiary of this would be VS Code users.
Note that at Carta I created a "nose style" test discovery which was very fast and eliminated the need for this kind of thing. Likewise, PyCharm doesn't use pytest for test discovery so its already fast. This would be a somewhat complex undertaking and so going with nose style test discovery may just be preferable as a first iteration solution. This nose style test discovery does not need to be a part of this project.
Currently, test discovery is either ran from scratch or a cached instance is used. There's checks in place to try to clear the cache. The cached test discovery is also not leveraged for
--collect-only
There could be a cached test discovery. When new test functions are added / renamed / removed, the cache could be updated.
The primary beneficiary of this would be VS Code users.
Note that at Carta I created a "nose style" test discovery which was very fast and eliminated the need for this kind of thing. Likewise, PyCharm doesn't use pytest for test discovery so its already fast. This would be a somewhat complex undertaking and so going with nose style test discovery may just be preferable as a first iteration solution. This nose style test discovery does not need to be a part of this project.