Chris Markiewicz

Oh man. If we're depending on Neurodocker, we're going to have a bad time.

Maybe something like: ```Python def cap_dtype(dt, nbytes): return np.dtype(f'{dt.byteorder}{dt.kind}{min(nbytes, dt.itemsize)}') ``` ```python In [1]: import numpy as np In [2]: def cap_dtype(dt, nbytes): ...: return np.dtype(f'{dt.byteorder}{dt.kind}{min(nbytes, dt.itemsize)}') ...: In [3]:...

Thanks! This looks reasonable on first glance. Would you mind creating a test that exercises this edge case?

@gagnonanthony Unfortunately, we can't accept this template as-is. The stated license is GPL, which does not apply to data and imposes unclear restrictions on end-users. Further, the [documentation](https://www.nitrc.org/frs/download.php/17110/Introduction-for-UNC-BCP-4D-Infant-Brain-Volumetric-Atlas-Ver2.pdf) states: >...

The issue remains that we are not yet validating JSON-defined columns.

Here's a quick script that should raise an exception if Zenodo does so internally: ```py # /// script # requires-python = ">=3.13" # dependencies = [ # "pyyaml", # "zenodo-legacy",...

I guess we did not get around to writing documentation, so here are some notes: 1) If you want the old `img.get_data()` behavior, just use `np.asanyarray(img.dataobj)`. This will keep the...

Hi Mattea. Nibabel doesn't currently support NRRD. We'd need a contribution from someone familiar with the format.

So it looks like @hjmjohnson's updates to pynrrd were incorporated, which I assume is why this stalled out. Have you looked into that? Would still welcome a contribution, but it's...

Yeah. Absolutely get your stuff done first. If you find some time to open a PR, cool. If not, we've survived without it for 2+ years.