BUG: <TypeError: Indexing arrays must have integer dtypes when reading H5 files with h5py >= 3.0' prefix>

[alex-attinger]

Describe the issue:

Running suite2p on an H5 input file raises the following error:

File "suite2p/io/h5.py", line 88, in h5py_to_binary
    im = f[key][irange, ...]
TypeError: Indexing arrays must have integer dtypes

In h5py_to_binary, the index array irange is computed using float division:

irange = np.arange(
    ik / ncp, min(ik / ncp + nbatch / ncp, nframes_all / ncp),
    1)

np.arange returns a float64 array. h5py >= 3.0 no longer silently casts float indices to integers and raises a TypeError instead. This was a deliberate breaking change introduced in h5py 3.0.
Suggested fix
Use floor division (//) and explicit dtype=int to keep the index array in integer space:

Reproduce the code example:

python -m suite2p --settings --verbose
creating log file 
2026-03-16 15:54:43,993 [INFO] 
platform:       	win32 
python version: 	3.11.15

Example input dataset used to reproduce the issue:

No response

Error message:

Version information:

v1.0.0.1

Context for the issue:

No response