On 2020-05-01 4:43 p.m., Chris Angelico wrote:
On Sat, May 2, 2020 at 5:21 AM Soni L. email@example.com wrote:
On 2020-05-01 3:41 p.m., Chris Angelico wrote:
On Sat, May 2, 2020 at 4:38 AM Soni L. firstname.lastname@example.org wrote:
On 2020-05-01 3:10 p.m., Brandt Bucher wrote:
I have pushed a first draft of PEP 618:
Please let me know what you think – I'd love to hear any *new* feedback that hasn't yet been addressed in the PEP!
What about using an optional kwarg for a handler for mismatched lengths? I made a post about it on the other thread and it's not addressed in the PEP. It'd make zip capable of doing zip_shortest, zip_equal (aka zip(strict=True)) and zip_longest, it's not stringly-typed, and it's user-extensible. Something along the lines of zip(foo, bar, baz, and_then=lambda consumed_items, iters: ...).
# iterates in chunks, e.g. a very large file that wouldn't fit all in RAM zip(*[iter(x)]*32, and_then=lambda res, _: (yield res))
I'm honestly not sure how useful this really is in practice. Iterating over a file is already going to be chunked. What do you gain by wrapping it up in an opaque zip call?
# strict zip sentinel = object() def zip_eq(res, iters): if res or any(next(x, sentinel) is not sentinel for x in iters): raise ValueError zip(a, b, c, and_then=zip_eq) # this would ideally be zip.strict e.g. zip(a, b, c, and_then=zip.strict), but w/e.
So.... a messier and noisier spelling of what's already in this proposal...
# normal (shortest) zip but using an explicit function def no_op(*args, **kwargs): pass zip(a, b, c, and_then=no_op)
... and a messier and noisier spelling of what we already have.
I say again, YAGNI. Give an actual use-case for the excessive generality of your proposal - namely, the ability to provide a custom function. And show that it's better with zip than just with a custom generator function.
we can finally push for the no_op function, for starters. but the main benefit is, again, being able to get the iterated values which were silently swallowed by zip when the iteration stopped.
also: I don't like booleans. they're not extensible, unless you consider None. you either get it right the first time, add a new boolean argument later, or use enum.Flag from the beginning. this callback-based API sidesteps all these issues.
and just in case maybe zip(strict=True) should be zip(errors=True) and we can later change it to zip(errors="replace", value=Foo) to get zip_longest >.< (no really bools = bad please don't use them in new APIs)
ChrisA _______________________________________________ Python-ideas mailing list -- email@example.com To unsubscribe send an email to firstname.lastname@example.org https://mail.python.org/mailman3/lists/python-ideas.python.org/ Message archived at https://email@example.com/message/UAQGBS... Code of Conduct: http://python.org/psf/codeofconduct/