Python is DOOMED! Again!

Mario Figueiredo marfig at gmail.com
Thu Jan 22 23:25:19 CET 2015


In article <6eb91c4b-92ff-44a8-b5a9-6ef04c71f4cb at googlegroups.com>, 
rantingrickjohnson at gmail.com says...
> 
> So if the purpose is "static analysis", what is the
> justification for injecting new syntax into function sigs?

1. Annotations where created exactly for this purpose. So there's some 
preassure to put them to work on what they were always meant to be used 
for.

2. Docstrings are meant as source of code documentation and to fill the 
object's __doc__ attribute, as per PEP 257. As such they aren't good 
consistent candidates for type hinting for purposes of static analysis 
tools.

3. Decorators are meant as function transformation tools, as per PEP 318 
and, as such, are them too not suitable to include type annotations for 
the purpose of static analysis.

4. There is no other formal annotation mechanism, other than these 
potential three.

Arguably, you can make a case against 1. And that's my own issue with 
type hinting as it is being proposed on PEP 484. The real problem is 
annotations as they were defined are a really bad idea. To couple type 
annotations so tightly with code, just results in plain ugly and hard to 
read code.

Personally I would love to see the present annotations just go away and 
a new docstring-like structure to emerge in their place. But that's 
probably not going to happen.

What I agree can't be done though, is to use docstrings or decorators as 
a source for type annotations.



More information about the Python-list mailing list