This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Title: Should map(function, iterable, ...) replace StopIteration with RuntimeError?
Type: behavior Stage:
Components: Versions:
Status: open Resolution:
Dependencies: Superseder:
Assigned To: Nosy List: rhettinger, veky, xavieryao
Priority: normal Keywords:

Created on 2022-02-03 04:46 by xavieryao, last changed 2022-04-11 14:59 by admin.

Messages (3)
msg412419 - (view) Author: Peiran Yao (xavieryao) Date: 2022-02-03 04:46
Currently, StopIteration raised accidentally inside the `function` being applied is not caught by map(). This will cause the iteration of the map object to terminate silently. (Whereas, when some other exception is raised, a traceback is printed pinpointing the cause of the problem.)

Here's a minimal working example:

def take_first(it: Iterable):
    # if `it` is empty, StopIteration will be raised accidentally
    return next(it) 

iterables = [iter([1]), iter([]), iter([2, 3])] # the second one is empty
for i in map(take_first, iterables):

`take_first` function didn't consider the case where `it` is empty. The programmer would expect an uncaught StopIteration, instead of the loop terminating silently after only one iteration.

Similar to the case of generators (described in PEP 497), this behaviour can conceal obscure bugs, and a solution could be catching StopIteration when applying the function, and replacing it with a RuntimeError.

Beside the built-in map(), imap() and imap_unordered() in the concurrent and multiprocessing modules also have similar behaviour.

PEP 479 -- Change StopIteration handling inside generators
msg412420 - (view) Author: Raymond Hettinger (rhettinger) * (Python committer) Date: 2022-02-03 05:14
-1 for being a breaking change, for addressing a minor issue that rarely arises in real life, and for being a slippery slope.
msg412449 - (view) Author: Vedran Čačić (veky) * Date: 2022-02-03 17:00
Just for the record, I consider PEP 479 one of (very rare) design bugs in Python, and would like it reversed some day. (So anything that helps this outcome, including -1 on this, is welcome.)

It subverts the natural property of exceptions (that they bubble through frames undisturbed until caught) for no benefit at all, and it has made me write almost every chained generator since then in a more complex way, adding boilerplate code that converts inner StopIteration to return. I'm sure many others have done so too. Ceterum censeo PEP479em esse delendam.
Date User Action Args
2022-04-11 14:59:55adminsetgithub: 90779
2022-02-03 17:00:32vekysetnosy: + veky
messages: + msg412449
2022-02-03 05:14:13rhettingersetnosy: + rhettinger
messages: + msg412420
2022-02-03 04:46:26xavieryaocreate