bioio-devs
test writing with omezarrwriter and reading with bioio-ome-zarr, likewise with tiff
Feature Description
Idea: Add Integration testing to ensure that our built-in writers maintain compatibility with the corresponding readers.
So this task has evolved into just being 'Make a test in bioio that uses BioImage to import a reader (main branch), then open a file corresponding to that reader as part of the scheduled testing routine. This test should also be required before a bioio branch can be merged into main. This test should take an optional argument to use a branch of a reader repo instead of the main branch. This test should also make use of the cached testfiles for the readers.'
Use Case
The writers in bioio are somewhat lacking in unit tests. Because we don't bundle the tiff reader and zarr reader inside of bioio, but we DO bundle the writers inside it, there is no assurance of compatibility.
Solution
Should we add tests to the reader packages or in bioio or have some new integration test repo? Should we consider bundling the zarr and tiff readers into bioio? (at a cost of increasing the unit test burden on bioio - but how do we test the writers anyway?)
I think the tests would be testing the writing since we have other tests in the readers to guarantee the writing aspect so it should belong in bioio
I think the tests would be testing the writing since we have other tests in the readers to guarantee the writing aspect so it should belong in bioio
I think you meant to say "other tests in the readers to guarantee the READING aspect" ; in any case I will proceed with looking at bioio to test the writing.
I'm thinking we want a new integration test repo; it would be a good place to house the tests that confirm the latest versions of bioio work with all the latest reader versions, as well as the test to confirm the written files are readable.
I have a script that reads a test file, writes it out, and reads it again for: ome_zarr ome_zarr 2 ome_tiff time_series (gif, mp4) 2d (png) It mostly uses the test data stored on aws for files, and this is dependent on solving that to be an automated workflow process. I found some random internet files for the gif, mp4 and png examples.
Dan (and others) have decided the writers should be moved into the readers, so this automation is blocked by that.
This task is replaced with three tasks specific to the repos that have writers: bioio-ome-zarr, bioio-imageio, bioio-ome-tiff