You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The NEP 51 changes implemented for numpy 2.0, prompting asdf/asdf#1605 indicates that ASDF should more carefully consider how it is serializing and deserializing numpy scalars. This may become a more pressing issue when NEP 41 related changes start becoming more widely used (for things like dtype based units).
It may be worth adding a "scalar" tag/schema similar to the "ndarray" tag/schema which enables encoding the specific dtype information in addition to the scalar's value. The implementation should be very careful to fully represent the full value stored so that this case will perfectly round trip.
The text was updated successfully, but these errors were encountered:
How should ASDF support handling
numpy
scalars?The NEP 51 changes implemented for
numpy
2.0, prompting asdf/asdf#1605 indicates that ASDF should more carefully consider how it is serializing and deserializingnumpy
scalars. This may become a more pressing issue when NEP 41 related changes start becoming more widely used (for things likedtype
based units).It may be worth adding a "scalar" tag/schema similar to the "ndarray" tag/schema which enables encoding the specific
dtype
information in addition to the scalar's value. The implementation should be very careful to fully represent the full value stored so that this case will perfectly round trip.The text was updated successfully, but these errors were encountered: