Control flow not optimized

[quotemstr]

BPO	3275
Nosy	@birkenfeld, @benjaminp

^{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 2008-07-03.20:51:29.437>
created_at = <Date 2008-07-03.20:43:13.035>
labels = ['interpreter-core', 'performance']
title = 'Control flow not optimized'
updated_at = <Date 2008-07-04.09:05:53.312>
user = 'https://bugs.python.org/quotemstr'

bugs.python.org fields:

activity = <Date 2008-07-04.09:05:53.312>
actor = 'georg.brandl'
assignee = 'none'
closed = True
closed_date = <Date 2008-07-03.20:51:29.437>
closer = 'benjamin.peterson'
components = ['Interpreter Core']
creation = <Date 2008-07-03.20:43:13.035>
creator = 'quotemstr'
dependencies = []
files = []
hgrepos = []
issue_num = 3275
keywords = []
message_count = 3.0
messages = ['69230', '69231', '69248']
nosy_count = 3.0
nosy_names = ['georg.brandl', 'benjamin.peterson', 'quotemstr']
pr_nums = []
priority = 'normal'
resolution = 'rejected'
stage = None
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue3275'
versions = ['Python 2.5']

[quotemstr]

Consider:

import dis
def foo():
  if 0 and 1: return "hi"

dis.dis(foo)

What I get is

2 0 LOAD_CONST 1 (0)
3 JUMP_IF_FALSE 15 (to 21)
6 POP_TOP
7 LOAD_CONST 2 (1)
10 JUMP_IF_FALSE 8 (to 21)
13 POP_TOP
14 LOAD_CONST 3 ('hi')
17 RETURN_VALUE
18 JUMP_FORWARD 1 (to 22)
>> 21 POP_TOP
>> 22 LOAD_CONST 0 (None)
25 RETURN_VALUE

What I'd expect to see is:

1 0 LOAD_CONST 0 (None)
3 RETURN_VALUE

[benjaminp]

A patch for this was just recently rejected. See bpo-1394.

[birkenfeld]

What real-life use case do you have for a condition that is a boolean
operation on two constant values anyway? Things like

while 1:
   ...

are properly optimized since they serve a useful purpose.