numexpr
numexpr copied to clipboard
evaluate global_dict argument ignored
>>> ne.evaluate('a', local_dict={'a': 1})
array(1, dtype=int32)
>>> ne.evaluate('a', global_dict={'a': 1})
KeyError: 'a'
The problem is a regression introduced in https://github.com/pydata/numexpr/commit/21ff376a0853aff6aea63343c6010d6434917319 because global_dict is not passed to re_evaluate (which does not support global_dict either)