On Thu, 27 May 2021 at 15:49, Chris Angelico <[email protected]> wrote:
>
> On Fri, May 28, 2021 at 12:25 AM Paul Moore <[email protected]> wrote:
> >
> > On Thu, 27 May 2021 at 15:04, Chris Angelico <[email protected]> wrote:
> > def static(**statics):
> > def deco(func):
> > for name, value in statics.items():
> > setattr(func, name, value)
> > func.__globals__["__me__"] = func
> > return func
> > return deco
> >
> > @static(a=1)
> > def f():
> > print(__me__.a)
> >
>
> Can't use globals like that, since there's only one globals() dict per
> module. It'd require some compiler magic to make __me__ work the way
> you want. But on the plus side, this doesn't require a run-time
> trampoline - all the work is done on the original function object.
Rats, you're right. Hacking globals felt like a code smell. I
considered trying to get really abusive by injecting some sort of
local but that's not going to work because the code's already compiled
by the time the decorator runs:
>>> def f():
... print(__me__.a)
...
>>> dis.dis(f)
2 0 LOAD_GLOBAL 0 (print)
2 LOAD_GLOBAL 1 (__me__)
4 LOAD_ATTR 2 (a)
6 CALL_FUNCTION 1
8 POP_TOP
10 LOAD_CONST 0 (None)
12 RETURN_VALUE
So yes, without compiler support you can only go so far (but you could
always use the _getframe approach instead).
> So, yeah, add it to the pile.
Yep. It's an interesting exercise, is all, and honestly, I don't think
I'd use static much anyway, so something "good enough" that works now
is probably more than enough for me personally.
I do think that having a compiler supported way of referring
efficiently to the current function (without relying on looking it up
by name) would be an interesting alternative proposal, if we *are*
looking at actual language changes - it allows for something
functionally equivalent to statics, without the performance advantage
but in compensation it has additional uses (recursion, and more
general access to function attributes). I'm not going to push for it
though, as I say, I don't have enough use for it to want to invest the
time in it.
Paul
_______________________________________________
Python-ideas mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3/lists/python-ideas.python.org/
Message archived at
https://mail.python.org/archives/list/[email protected]/message/TFAU3JX3RH2RAQ3E2KMVD43HZ34454DH/
Code of Conduct: http://python.org/psf/codeofconduct/
