Wrong bytecode generated for 'in' operation

[phmc]

BPO	18208
Nosy	@brettcannon, @birkenfeld, @amauryfa, @ncoghlan, @benjaminp, @phmc

^{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 2013-06-14.14:59:31.484>
created_at = <Date 2013-06-14.08:42:34.558>
labels = ['interpreter-core', 'invalid']
title = "Wrong bytecode generated for 'in' operation"
updated_at = <Date 2013-06-14.14:59:31.483>
user = 'https://github.com/phmc'

bugs.python.org fields:

activity = <Date 2013-06-14.14:59:31.483>
actor = 'pconnell'
assignee = 'none'
closed = True
closed_date = <Date 2013-06-14.14:59:31.484>
closer = 'pconnell'
components = ['Interpreter Core']
creation = <Date 2013-06-14.08:42:34.558>
creator = 'pconnell'
dependencies = []
files = []
hgrepos = []
issue_num = 18208
keywords = []
message_count = 3.0
messages = ['191108', '191112', '191134']
nosy_count = 7.0
nosy_names = ['brett.cannon', 'georg.brandl', 'amaury.forgeotdarc', 'ncoghlan', 'benjamin.peterson', 'pconnell', 'isoschiz']
pr_nums = []
priority = 'normal'
resolution = 'not a bug'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue18208'
versions = []

[phmc]

The following two expressions should have the same value:

Python 3.4.0a0 (default:fae92309c3be, Jun 14 2013, 09:29:54) 
[GCC 4.8.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> 1 in [2] == False
False
>>> (1 in [2]) == False
True

It looks like this is a compiler issue - there shouldn't be a jump if the 'in' expression is false:

>>> dis.dis("1 in [2] == False")
  1           0 LOAD_CONST               0 (1)
              3 LOAD_CONST               1 (2)
              6 BUILD_LIST               1
              9 DUP_TOP             
             10 ROT_THREE           
             11 COMPARE_OP               6 (in)
             14 JUMP_IF_FALSE_OR_POP    24
             17 LOAD_CONST               2 (False)
             20 COMPARE_OP               2 (==)
             23 RETURN_VALUE        
        >>   24 ROT_TWO             
             25 POP_TOP             
             26 RETURN_VALUE        
>>>

[amauryfa]

This is one case of chained comparisons:
http://docs.python.org/3/reference/expressions.html#not-in

"x <= y <= z" is equivalent to "(x <= y) and (y <= z)"
"x in y == z" is equivalent to "(x in y) and (y == z)"

There is a jump if the 'in' expression is false, because 'and' should short-circuit the second comparison.

[phmc]

Thanks Amaury. That's quite surprising, but I wouldn't advocate changing such long standing behaviour.

I'm closing the issue.