Using field transformer breaks integration with hypothesis

[jakub-borusewicz]

Adding field transformer according to documentation (i.e. as a function of type Callable[[type, list[attr.Attribute[Any]]], list[attr.Attribute[Any]]])
breaks attrs integration with hypothesis, because attributes transformed with such function are not casted to generated AttrsClass, but remains just an list of attributes (here)

This causes error in hypothesis here, because fields is an list, not an AttrsClass instance, and therefore getattr(fields, name) raises an AttributeError.

Easy workaround for that is just to create an AttrsClass again inside field transformer, like so:

def transformer(cls: type, fields: list[attr.Attribute[Any]]) -> list[attr.Attribute[Any]]:
    attr_names = [f.name for f in fields]
    AttrsClass = _make_attr_tuple_class(cls.__name__, attr_names)
    
    ... do something with fields ...

    return AttrsClass(fields)

...but it's not very handy nor pretty, and messes with typing. So, I would very thankful if it could be resolved within attrs (it would be quite easy, I guess).

BTW. I love your package, awesome job! Cheers!

[hynek]

@sscherfke? 😇

[sscherfke]

I will take a look, but I can’t tell when. I’m currently quite busy with work. :)