Op 08-01-18 om 17:25 schreef Cody Piersall: >> Let's put it this way. Suppose that __eq__ existed and __ne__ didn't, >> just like with __contains__. Go ahead: sell the notion of __ne__. >> Pitch it, show why we absolutely need to allow this. Make sure you >> mention the potential confusion when subclassing. Be sure to show why >> it's okay for "not in" to force to boolean but "==" should allow any >> return value. > > __ne__ and __eq__ are important for building mask arrays in NumPy, > which allow complex indexing operations. A lot of NumPy's design was > inspired by MATLAB, so being able to index the same way as in MATLAB > is a pretty killer feature.
They are only important if you find it necessary to build these mask arrays wih an operator. > Indexing an array using mask arrays like this is idiomatic: > > some_arr = np.array([-1, 0, 1, 2, 3, 4, 5, 2, -1, 3, -1, 6, 7, 3]) > valid = some_arr[some_arr != -1] I guess it was rather useful because list comprehension wasn't included in the language at that moment, but I don't find it that much more useful than: valid = somearr[[number != -1 for number in somearr]] -- Antoon Pardon. -- https://mail.python.org/mailman/listinfo/python-list
