RetroSearch Browse

Thu Aug 5 21:54:17 CEST 2004 · https://mail.python.org/pipermail/python-dev/2004-August/046921.html

On 5-aug-04, at 21:45, Neil Schemenauer wrote:

> On Thu, Aug 05, 2004 at 09:28:13PM +0200, Ronald Oussoren wrote:
>> On 5-aug-04, at 21:17, Gustavo Niemeyer wrote:
>>>   def funcname(...):
>>>      ...
>>>   funcname.signature = "v@:@i"
>>
>> That should be workable for this specific example.
>
> Even nicer if '_' is bound to the last function defined.
>
>     class SomeClass(objc):
>         def funcname(...):
>            ...
>         _.signature = "v@:@i"
>
> The objc metaclass could take the 'signature' function attribute and
> transform the function.
>
>> It wouldn't work for the objc.accessor example. The objc.accessor
>> function/decorator deduces the right kind of signature from the
>> name and arguments of the function.
>
> Can't the metaclass do that?

Not every method is a property accessor. The metaclass does calculate a 
default signature (all arguments are objects and the result is an 
object) when it doesn't have better information (such as the signature 
of a method with the same name in a superclass). Sadly enough that 
isn't good enough for property setters. Some parts of Cocoa require 
that a setter returns 'void'.

I'm trying to push the automatic detection of the right signature as 
far as I can get to avoid having to use explicit annotations in user 
code, but sometimes you just cannot avoid adding annotations.

Ronald
--
X|support bv            http://www.xsupport.nl/
T:  +31 610271479       F:  +31 204416173

Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Showing content from https://mail.python.org/pipermail/python-dev/2004-August/046921.html below:

[Python-Dev] Call for defense of @decorators