PEG parser doesn't allow lambda in for_if_clause

[slebedev]

BPO	41848
Nosy	@gvanrossum, @terryjreedy, @benjaminp, @serhiy-storchaka, @lysnikolaou, @pablogsal, @isidentical
PRs	bpo-41848: Allow simple lambdas in if clauses of comprehensions #22394

^{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 = None
closed_at = <Date 2020-09-26.23:36:40.147>
created_at = <Date 2020-09-23.23:33:02.167>
labels = ['interpreter-core', 'type-bug', '3.9', '3.10']
title = "PEG parser doesn't allow lambda in for_if_clause"
updated_at = <Date 2020-09-26.23:36:40.145>
user = 'https://bugs.python.org/slebedev'

bugs.python.org fields:

activity = <Date 2020-09-26.23:36:40.145>
actor = 'gvanrossum'
assignee = 'none'
closed = True
closed_date = <Date 2020-09-26.23:36:40.147>
closer = 'gvanrossum'
components = ['Interpreter Core']
creation = <Date 2020-09-23.23:33:02.167>
creator = 'slebedev'
dependencies = []
files = []
hgrepos = []
issue_num = 41848
keywords = ['patch']
message_count = 5.0
messages = ['377435', '377511', '377514', '377545', '377546']
nosy_count = 8.0
nosy_names = ['gvanrossum', 'terry.reedy', 'benjamin.peterson', 'serhiy.storchaka', 'lys.nikolaou', 'pablogsal', 'BTaskaya', 'slebedev']
pr_nums = ['22394']
priority = 'normal'
resolution = 'wont fix'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue41848'
versions = ['Python 3.9', 'Python 3.10']

[slebedev]

Reproducer:

[x for x in [] if lambda: x]

This parses fine in 3.8, but doesn't parse in 3.9 because the grammar expects a disjunction after if in for_if_clause *. While this change has zero practical significance, I think it might be useful to maintain a list of such changes for tooling authors (who might need to support multiple Python versions).

[terryjreedy]

After adding parens,
>>> [x for x in [] if (lambda: x)]
[]

[gvanrossum]

I propose not to fix this, since it is pointless (a lambda is always truthy).

We can call it out in What's New in 3.9 -- as a very small footnote. :-)

[pablogsal]

I concur with Guido: it makes the grammar slightly more complex for no gain. Also, the originally allowed rule does not allow *any* lambda but only the ones that are simple, which is a special case that makes it even more complex. In any case you can use parentheses to break this limitation.

Like Guido, I propose to add a very small comment to What's New in 3.9

[gvanrossum]

Okay, let's close this.