As far as MaRCoS is affected, the compiler uses a lot of fixed-width signed and unsigned integers implicitly in some fairly involved logic operations. I've added various casts and conversions to fix the bugs that arose with Numpy 2, however others may remain undetected until specific value ranges are hit in the input data.
@josalggui, @catkira please let me know if these changes cause you any issues/problems in your workflows.
Numpy 2 changes the way certain type conversions are implicitly handled, and is stricter about overflows - see https://numpy.org/doc/stable/numpy_2_0_migration_guide.html#changes-to-numpy-data-type-promotion and https://numpy.org/neps/nep-0050-scalar-promotion.html#nep50 for detailed info.
As far as MaRCoS is affected, the compiler uses a lot of fixed-width signed and unsigned integers implicitly in some fairly involved logic operations. I've added various casts and conversions to fix the bugs that arose with Numpy 2, however others may remain undetected until specific value ranges are hit in the input data.
@josalggui, @catkira please let me know if these changes cause you any issues/problems in your workflows.