override generated __eq__ to work with numpy fields

[chobeat]

I would like to know how to override generated methods in attrs. From the documentation is not clear.

My problem is that if I put some numpy arrays as values, the regular "==" operator fails because np.array_equal should be used instead.

The error is the following:

    def __eq__(self, other):
        if other.__class__ is not self.__class__:
            return NotImplemented
        return  (
            self.raw,
            self.processed,
        ) == (
            other.raw,
>           other.processed,
        )
E       ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

[hynek]

I’m afraid you’ll have to disable attrs’ eq and write it yourself. This is a super edgy edge case so I don’t think it makes a lot of sense to add special hooks for it. Are you aware of the reason why numpy arrays don't allow for comparison using ==?

[chobeat]

Yes, there's a very explicit error message if you try do it and it makes sense: for the way they make use of python syntax, a == would be ambigous if used to compare 2 arrays, so they offer numpy.array_equal.

I’m afraid you’ll have to disable attrs’ eq and write it yourself.

Can you explain me how? Just writing a eq method in my class doesn't work because I believe attrs eq is generated afterward.

[hynek]

Your approach might work in the future (see #324) until then you'll have to pass @attr.s(cmp=False).

[chobeat]

Oh, perfect, thank you very much