tuples can be created from iterators

[lbaumbauer]

Bug report

Bug description:

tuples can be created from iterators and this code should not need the creation of a list in between the iterator and the tuple.

    if typed:
        key += tuple([type(v) for v in args])
        if kwds:
            key += tuple([type(v) for v in kwds.values()])

as seen in ./Lib/functools.py

CPython versions tested on:

3.16

Operating systems tested on:

No response

[aisk]

I think this is intentional, see #143825. Feel free to reopen this if I misunderstood anything.

[Michael202119]

no se programar....,quisiera una guia

[lbaumbauer]

I think this is intentional, see #143825. Feel free to reopen this if I misunderstood anything.

I said the code should not need it . So my expectation is that the language is changed so that it is not necessary to add the List

[aisk]

Hi, since this issue was filed as a bug report, I read it as "the code in functools.py is wrong and the list should be removed". That's why I closed it, because the list comprehension there is intentional (see #143825), so it's not a bug.

Currently, using a list comprehension is still much faster on the main branch, so we still need it:

./python -VV && \
./python -m timeit -s "args=(1,'a',3.0,b'x')" "tuple([type(v) for v in args])" && \
./python -m timeit -s "args=(1,'a',3.0,b'x')" "tuple(type(v) for v in args)"

Python 3.16.0a0 (heads/master:5918085bb6f, Aug  4 2026, 00:14:14) [GCC 15.3.0]
500000 loops, best of 5: 570 nsec per loop
100000 loops, best of 5: 2.01 usec per loop

If you want to improve this kind of performance, in my opinion, you should give at least some ideas about how to optimize it, instead of just asking for it, because many Python statements can be made faster. For example, #131738 optimized tuple/any/all calls with a generator expression argument. And there is already a related issue tracking this: #138325.

If I misunderstood anything, please feel free to reopen this issue. (I'm not sure if the issue author can reopen it. If you think it should be reopened but don't have the permission, please leave a comment and I will reopen it.)

[lbaumbauer]

@aisk It does not need to be reopened. I did not know the other ticket. Thank you for providing the link. I also added my analysis that showed the existing analysis was incomplete.