add sys.exception()

[iritkatriel]

BPO	46328
Nosy	@warsaw, @ericsnowcurrently, @sweeneyde, @iritkatriel, @AlexWaygood
PRs	bpo-46328: added sys.exception() #30514

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/iritkatriel'
closed_at = <Date 2022-01-13.12:36:20.282>
created_at = <Date 2022-01-10.11:50:17.000>
labels = ['type-feature', 'library', '3.11']
title = 'add sys.exception()'
updated_at = <Date 2022-01-13.12:36:20.282>
user = 'https://github.com/iritkatriel'

bugs.python.org fields:

activity = <Date 2022-01-13.12:36:20.282>
actor = 'iritkatriel'
assignee = 'iritkatriel'
closed = True
closed_date = <Date 2022-01-13.12:36:20.282>
closer = 'iritkatriel'
components = ['Library (Lib)']
creation = <Date 2022-01-10.11:50:17.000>
creator = 'iritkatriel'
dependencies = []
files = []
hgrepos = []
issue_num = 46328
keywords = ['patch']
message_count = 9.0
messages = ['410204', '410216', '410222', '410224', '410225', '410255', '410258', '410260', '410489']
nosy_count = 5.0
nosy_names = ['barry', 'eric.snow', 'Dennis Sweeney', 'iritkatriel', 'AlexWaygood']
pr_nums = ['30514']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue46328'
versions = ['Python 3.11']

[iritkatriel]

Following to changes in bpo-45711, the interpreter's internal representation of the active exception is just the exception instance (the exc_type and exc_traceback fields were removed).

For backwards compatibility, sys.exc_info() constructs the (typ, val, tb) tuple from the instance and this will continue to be the case for some time because this tuple has leaked into quite a few APIs.

However, now that the redundancy in the exc_info tuple is guaranteed by the way it's constructed, we can confidently add a sys.exception() method that returns just the exception instance (as suggested in PEP-3134's section on future improvements).

This small change will make a difference to learners because the (typ, val, tb) tuple looks quite cryptic to those who don't know about tracebacks, and the redundancy in it is confusing for those who do.

[iritkatriel]

This is part of a larger plan to reduce the footprint of the exc_info triplet on the language. See https://gist.github.com/iritkatriel/3927147548b10a7929cb0b680e3adc52
History

[sweeneyde]

Would there be any value in spelling this as sys.active_exception() or sys.current_exception() or sys.get_exception() or sys.exception_in_flight() or similar?

My only worry is confusion between sys.exception() versus builtins.Exception.

[ericsnowcurrently]

So sys.exception() will be equivalent to sys.exc_info()[1] (or rather, sys.exc_info() will be (type(sys.exception()), sys.exception(), sys.exception().__traceback__))?

That seems good to me.

[ericsnowcurrently]

FWIW, here's an alternative we should not pursue: return just the exception from sys.exc_info(). For compatibility, we would implement __iter__ and __getitem__ on BaseException, to duplicate the behavior of the current tuple. There are a number of good reasons why this is not what we should do.

[iritkatriel]

I thought of something like sys.active_exception() but it seems like a lot to type. sys.exception() was suggested in PEP-3134.

Does this change need a pep?

[warsaw]

sys.exception() seems like a decent enough trade-off. I've always disliked the abbreviations in "exc_info". It doesn't feel big enough for a PEP to me.

[iritkatriel]

Cool. I just removed the do-not-merge label from the PR and I guess it's ready to be reviewed.

[iritkatriel]

New changeset c590b58 by Irit Katriel in branch 'main':
bpo-46328: Add sys.exception() (GH-30514)
c590b58