pin tifffile to a newer version to allow unpinning of ome-zarr.

[ilan-theodoro]

Since version 2023.3.15, tifffile micromanager_metadata["IndexMap"] is stored as a numpy array instead of a dict. So right after i get micromanager_metadata["IndexMap"], I am converting it again to a dict as specified in the tifffile releases before 2023.3.15.

However, I identified in the tests that there is shape mismatch in the test_converter.py if zarr<2.15 or tifffile>=2024.1.30, so I pinned those versions.

iohub/tests/test_converter.py

Line 253 in a4d43a2

assert np.prod([d for d in converter.dim if d > 0]) == np.prod(

But it solves the problem in #207 and #165.

[ziw-liu]

Seems like black 24.1.1 has different behavior compared to 23.1... @ilan-francisco can you change those lines so the CI passes?

[ilan-theodoro]

Seems like black 24.1.1 has different behavior compared to 23.1... @ilan-francisco can you change those lines so the CI passes?

My bad, I tried to fix it before I got your message. I will push it again

[ilan-theodoro]

@ziw-liu actually I discovered that not all files were used locally when I tested. So the problem persists with tifffile=2023.3.15. I am having a bad time trying to solve it. The problem is indeed with the dimensions of some files. For example, reading from mm2.0-20201209_4p_2t_5z_1c_512k_1 gives different results for the versions 2023.2.28 and 2023.3.15. Somehow reading from tifffile=2023.3.15 kind of gives us an extra dimension. Do you know something about it that could give me some light?

[ziw-liu]

@ilan-francisco cgohlke/tifffile#187 gives some context. Unfortunately upstream is not interested in the particular solution I proposed.

[ziw-liu]

Closing in favor of #185.

Thanks @ilan-theodoro!