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
For minimal modification (and a tiny performance advantage), this can be done in the NumpyArray constructor. (Also check the NumpyForm and conversions from JSON into NumpyForms.) That way, the unnecessary character can be removed once in one place, without having to fix every instance in NumpyArray where format_ is used.
If NumPy arrays are created as big-endian and then are converted with astype to little-endian, they have this extra character in their formats. Uproot4 is using a work-around so that a new version of Awkward with this fix is not needed right away, but it's important to fix anyway.
This would be an easy fix, though it's entirely in C++, not Python.
The text was updated successfully, but these errors were encountered:
This will happen again when dealing with fixed-width objects (i.e. TLorentzVector), so it's worth fixing and dealing with the Awkward1 ↔ Uproot4 version synchronization.
For minimal modification (and a tiny performance advantage), this can be done in the NumpyArray constructor. (Also check the NumpyForm and conversions from JSON into NumpyForms.) That way, the unnecessary character can be removed once in one place, without having to fix every instance in NumpyArray where
format_
is used.If NumPy arrays are created as big-endian and then are converted with
astype
to little-endian, they have this extra character in their formats. Uproot4 is using a work-around so that a new version of Awkward with this fix is not needed right away, but it's important to fix anyway.This would be an easy fix, though it's entirely in C++, not Python.
The text was updated successfully, but these errors were encountered: