Depending on subwindow size, PyVale can return subset_x and subset_y values in the csv file that are not an integer (ie 7.5 or 11.5)
Newer versions of numpy (v2.1.3 is fine, v2.5.2 breaks) are more strict about casting this directly to an integer, giving the following error:
ValueError: could not convert string '7.5' to int32 at row 0, column 1.
Currently we store the coordinate in memory so changing this to a float will occupy more memory. Is there a reason we keep this coordinates? We could just cast to a float instead (which is used for subwindow size calc) then not save the coordinates?
Depending on subwindow size, PyVale can return
subset_xandsubset_yvalues in the csv file that are not an integer (ie 7.5 or 11.5)Newer versions of numpy (v2.1.3 is fine, v2.5.2 breaks) are more strict about casting this directly to an integer, giving the following error:
ValueError: could not convert string '7.5' to int32 at row 0, column 1.Currently we store the coordinate in memory so changing this to a float will occupy more memory. Is there a reason we keep this coordinates? We could just cast to a float instead (which is used for subwindow size calc) then not save the coordinates?