[Python-Dev] Not Deprecating Arbitrary Function Annotations

Steve Wedig stevewedig at gmail.com
Mon Oct 5 14:55:37 EDT 2015


Congratulations on the release of 3.5 and Pep 484. I've used Python
professionally for 10 years and I believe type hints will make it easier to
work with large codebases evolving over time. My only concern about Pep 484
is the discussion of whether or not to deprecate arbitrary function
annotations.
https://www.python.org/dev/peps/pep-0484/

I would like to request that arbitrary function annotations are not
deprecated for three reasons:
1. Backwards Compatibility
2. Type Experimentation
3. Embedded Languages

1. Backwards Compatibility
After reading Pep 3107 my team has made significant use of non-standard
annotations. It would be a serious burden to be forced to port our
annotations back to decorators. This would also make our codebase
considerably less readable because function annotations are much more
readable than input/output annotations relegated to decorators.
https://www.python.org/dev/peps/pep-3107/

2. Type Experimentation
Arbitrary function annotations allow developers to experiment with
potential type system improvements in real projects. Ideas can be validated
before officially adding them to the language. This seems like an advantage
that should be preserved. After all, Pep 484 says it was strongly inspired
by MyPy, an existing project.
http://mypy-lang.org/

3. Embedded Languages
Python's flexibility makes it an amazing language to embed other languages
in. In this regard, Python 3's addition of arbitrary function annotations
and class decorators complements Python 2's dynamic typing, function
decorators, reflection, metaclasses, properties, magic methods, generators,
and keyword arguments. Arbitrary function annotations are a crucial part of
this toolkit, and this feature is not available in most other languages.
For anyone interested in the utility and mechanics of embedded languages,
I'd recommend Martin Fowler's book: Domain Specific Languages.
http://www.amazon.com/Domain-Specific-Languages-Addison-Wesley-Signature-Series/dp/0321712943

So I agree with the course of action mentioned in Pep 484 that avoids
runtime deprecation of arbitrary function annotation: "Another possible
outcome would be that type hints will eventually become the default meaning
for annotations, but that there will always remain an option to disable
them." I would only add that there should be a way to disable type checking
for an entire directory (recursively). This would be useful for codebases
that have not been ported to standard annotations yet, and for codebases
that will not be ported for the reasons listed above.

Thanks for your consideration.

Best,
Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-dev/attachments/20151005/f1a5e2bb/attachment.html>


More information about the Python-Dev mailing list