numpy 1.24.X (current latest) depreciates np.int()

[tsippel12345]

Near line ~190 of recon.py, the following lines:

sino_chunk_start = np.int(sino_start + nSino_per_chunkiChunk)
sino_chunk_end = np.int(sino_start + nSino_per_chunk(iChunk+1))

.... np.int() is depreciated in vers 1.24.x of np and replaced with np.int32() and np.int64()

[tsippel12345]

Issue is currently resolved by installing version 1.23.5 of numpy rather than the latest 1.24.X version
conda install numpy=1.23.5
but perhaps declaration of varliables should be rewritten to conform to future numpy releases. Unclear to me whether np.int32(), np.int64(), or int() should be used, or perhaps choice makes minimal difference in precision / computational time.

[decarlof]

nSino_per_chunk = pow(2, int(params.binning)) is always an int. line ~190 is to convert a native int (flexible-sized (at least in python 3.x)) to a numpy int. I can set this to int32. It will not change precision/performance as it is just to set the number of sinograms to be processed per chunk.

[tsippel12345]

Great! Thanks Francesco! Completely new to python. First exposure is to analyze a set of data you folks (Alan K) graciously helped collect. Don't feel comfortable changing anything in the program!