[Python-ideas] Tail recursion elimination

[Joao S. O. Bueno]

On 19 January 2014 12:27, spir <denis.spir at gmail.com> wrote:
> On 01/19/2014 12:54 PM, Joao S. O. Bueno wrote:
>>
>> On 19 January 2014 08:52, Mark Lawrence <breamoreboy at yahoo.co.uk> wrote:
>>>
>>> On 19/01/2014 07:39, musicdenotation at gmail.com wrote:
>>>>
>>>>
>>>> I propose tail-call optimization to be added into CPython.
>>>
>>>
>>>
>>> Then implement it so everybody else can use it.
>>
>>
>> On a second though,  it actually could be done, at the VM level.
>> I am not a proponent, but after my second though I am from "-1" to "+0".
>>
>> I believe that anytime one have the sequence:
>>
>>               20 CALL_FUNCTION            1
>>               23 RETURN_VALUE
>>
>> in byte code, the current stack frame could be discarded prior
>> to making the function call. Looking from 10000 meters, it feels
>> like it would not impact any other aspect of the language but for
>> enabling automatically tail recursion calls.
>
>
> You also need to adjust frame size, possibly even its structure (dunno,
> depends on implementation details of python's "calling convention" so to
> say), to get a right space (and disposition) for the callee's input
> variables.

Not in this suggestion -  I did not propose re-using the frame,
as seens to be the case around the calls, just because of that:
these frames in Python seen to be tied to the code object
within it. My suggestion is simply to discard the current frame
before building the frame for the call. (Maybe adding some logging
information on this next frame so that the stack trace could be complete)

  js
 -><-

>
> Denis
>
> _______________________________________________
> Python-ideas mailing list
> Python-ideas at python.org
> https://mail.python.org/mailman/listinfo/python-ideas
> Code of Conduct: http://python.org/psf/codeofconduct/