On Nov 1, 2007, at 2:16 PM, Michael Watkins wrote:

> * David Binger wrote on [2007-11-01 11:28:08 -0400]:
>> def f(x:int, y:list) -> float:
>>
>> I'm proposing using a similar annotation syntax for annotating the
>> function name, using a colon and either "xml" or "str" to say that
>> the function is a template and that it should be compiled using
>> the "xml" transformation, or the "str" transformation.
>
> Without any thought at all to whether this would be implementable in
> Python < 3.x, using Python 3 annotation syntax as a guide. (I say
> this not knowing if the new annotation syntax is an all or nothing -
> i.e. both parameters and return values - approach or not.)

No problem there.  The proposed qpy syntax is a SyntaxError in all
versions of python (except with the qpy compiler).

>
> Python 3:
>> def f(x:int, y:list) -> float:
>
> Would this make some sense, and is it implementable:
>
>     def f(annotated, parameters, or not) -> xml:
>     def f(annotated, parameters, or not) -> string:
>
> Having just written that, it is perhaps not such a good idea as it
> may perpetuate the notion that only the return value is "xml" or
> "string".

This would make it impossible to use the annotation to say that
an ordinary, non-template function returns an xml or string.
Also, since it would not be a SyntaxError in py3k, you'd run the
risk of having these compile as non-templates.

>
> As for your proposal:
>> def f:xml():
>> def f:str():
>
> I like this better than ``@[xml]`` if for no other reason than it is
> easier to type. I also think it looks cleaner than the
> decorator-like syntax.

I think so.