not operator expression raising a syntax error

[candide]

BPO	24612
Nosy	@scoder, @benjaminp, @aroberge, @ezio-melotti, @stevendaprano, @vadmium, @serhiy-storchaka, @pablogsal, @iritkatriel
PRs	bpo-24612: Improve syntax error for 'not' after an operator #28170

^{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 = None
created_at = <Date 2015-07-11.15:23:53.358>
labels = ['interpreter-core', 'type-bug', '3.11']
title = 'not operator expression raising a syntax error'
updated_at = <Date 2021-09-05.01:21:27.256>
user = 'https://bugs.python.org/candide'

bugs.python.org fields:

activity = <Date 2021-09-05.01:21:27.256>
actor = 'aroberge'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Interpreter Core']
creation = <Date 2015-07-11.15:23:53.358>
creator = 'candide'
dependencies = []
files = []
hgrepos = []
issue_num = 24612
keywords = ['patch']
message_count = 10.0
messages = ['246606', '246607', '246608', '246611', '246612', '246613', '246619', '246620', '401055', '401063']
nosy_count = 11.0
nosy_names = ['scoder', 'benjamin.peterson', 'aroberge', 'ezio.melotti', 'mrabarnett', 'steven.daprano', 'martin.panter', 'serhiy.storchaka', 'candide', 'pablogsal', 'iritkatriel']
pr_nums = ['28170']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue24612'
versions = ['Python 3.11']

[candide]

Expressions such as

a + not b
a * not b
+ not b

• not b

raise a SyntaxError, for instance :

>>> 0 + not 0
  File "<stdin>", line 1
    0 + not 0
          ^
SyntaxError: invalid syntax
>>> - not 0
  File "<stdin>", line 1
    - not 0
        ^
SyntaxError: invalid syntax
>>>

if the not expression is wrapped in parenthesis, expected evaluation occurs:

>>> - not 0
  File "<stdin>", line 1
    - not 0
        ^
SyntaxError: invalid syntax
>>> 0 + (not 0)
1
>>> - (not 0)
-1
>>>

The problem has been first submitted in comp.lang.python :

https://groups.google.com/forum/?hl=fr#!topic/comp.lang.python/iZiBs3tcuak

suggesting a bug report.

[serhiy-storchaka]

"not not 0" is compiled successful, while "+ not 0", "- not 0", and "~ not 0" are rejected.

[stevendaprano]

On Sat, Jul 11, 2015 at 03:23:53PM +0000, candide wrote:
> 
> New submission from candide:
> 
> Expressions such as
> 
> a + not b
> a * not b
> + not b
> - not b
> 
> raise a SyntaxError, for instance :
> 
> 
> >>> 0 + not 0
>   File "<stdin>", line 1
>     0 + not 0
>           ^
> SyntaxError: invalid syntax

That has been invalid syntax since Python 1.5, if not older. I don't
think that it needs to be changed.

[steve@ando ~]$ python1.5
Python 1.5.2 (#1, Aug 27 2012, 09:09:18)  [GCC 4.1.2 20080704 (Red Hat 
4.1.2-52)] on linux2
Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam
>>> 0 + not 0
  File "<stdin>", line 1
    0 + not 0
          ^
SyntaxError: invalid syntax

[scoder]

It looks like perfectly valid syntax to me and I cannot see why it should be semantically rejected. I disagree that it shouldn't be changed. I think it's a (minor) bug that should eventually get fixed.

[mrabarnett]

"not" has a lower priority than unary "-"; this:

not a < b

is parsed as:

not (a < b)

How would you parse:

0 + not 0 + 0

?

Would it be parsed as:

0 + not (0 + 0)

?

Similar remarks could apply to "yield":

0 + yield 0

which is also a syntax error.

[scoder]

Hmm, right. I see your point and also the analogy with "yield". I could live with (maybe) giving it a better error message suggesting to use parentheses for clarity and otherwise keeping it a SyntaxError.

[vadmium]

Funny, I ran into this one or two days ago, when refactoring some code that used the bitwise exclusive-or operator, since there is no boolean exclusive or:

-if (x == a) ^ (y != b): ...
+aa = x == a
+bb = y == b
+if aa ^ not bb: ...

It is fairly clear what I wanted to do above, but with “is not” you would have to avoid ambiguity:

>>> "spam" is not "ham"  # Using “is not” operator
True
>>> "spam" is (not "ham")  # Two distinct operators
False

I think it would be too complicated to make unary “not” bind more tightly than “and” in some cases but not in others. How would you handle these cases?

a + not b and c
a ** not b ** c
a is not not b

The way I see it, there is no equivalent problem with the other unary operators: arithmetic (+, -), bitwise (~), and “await”. Await has higher priority than all other binary operators, so no problem there. The other three are equal with the highest priority binary operator, exponentiation (**):

>>> 2 ** -1 ** 2  # Evaluated right to left
0.5
>>> 2 ** (-1) ** 2  # Negation first
2
>>> (2 ** -1) ** 2  # Left-hand exponentiation first
0.25

BTW, in the operator precedence table <https://docs.python.org/dev/reference/expressions.html#operator-precedence\>, I think exponentiation should be in the same priority group as the arithmetic and bitwise unary operations. At the moment it says exponentiation has higher priority, but has a footnote saying this is reversed on the right-hand side of an exponentiation. This is unclear when applied to my example above.

[vadmium]

BTW “yield” is not a fair comparison because its syntax is even stranger (due to originally being a “yield” statement):

def g():
    x = yield + 1  # Yield the value +1
    y = (yield) + 1  # Add 1 after yielding
    return (yield)  # Mandatory brackets

[iritkatriel]

Reproduced on 3.11:

>>> 0 + not 0
  File "<stdin>", line 1
    0 + not 0
        ^^^
SyntaxError: invalid syntax
>>> - not 0
  File "<stdin>", line 1
    - not 0
      ^^^
SyntaxError: invalid syntax

[pablogsal]

I think the best outcome here is to refine the syntax error. Making it behave a bit better is going to be quite a pain because of how unary "-" and "not" work on the priority level in the grammar.

I also don't think that facilitating the concatenation of operators without parentheses is a good idea (for readability reasons).

I will prepare a PR to improve the syntax error

[encukou]

Fixed in #28170.
Thanks to everyone involved!