On 03:24 pm, itamar@itamarst.org wrote:

Until now we haven't AFAIK used decorator syntax, but I think it's time we start to. In places where we were already using decorators without the special syntax, that is.

Consider:

def foo(cls, arg): # lalalal foo = classmethod(foo)

One has to type "foo" three times (a potential source of error), and the special semantics of the function are known to the reader only *after* reading the function. On the other hand:

@classmethod def foo(cls, arg): # lalalal

This suffers from none of these problems.

Any disagreement?

Can we start with a limited experimental usage? Some new code can be written this way, and we'll see if there's any negative feedback following the next release. I don't want existing code to get ported to the new syntax just because it's possible now. Jean-Paul

On 03:24 pm, itamar@itamarst.org wrote:

Until now we haven't AFAIK used decorator syntax, but I think it's time we start to. In places where we were already using decorators without the special syntax, that is.

Consider:

def foo(cls, arg): # lalalal foo = classmethod(foo)

One has to type "foo" three times (a potential source of error), and the special semantics of the function are known to the reader only *after* reading the function. On the other hand:

@classmethod def foo(cls, arg): # lalalal

This suffers from none of these problems.

Any disagreement?

Can we start with a limited experimental usage? Some new code can be written this way, and we'll see if there's any negative feedback following the next release. I don't want existing code to get ported to the new syntax just because it's possible now. Jean-Paul