The documentation erroneously changes "primary" for "away" in the power operator syntax:

https://docs.python.org/3.6/reference/expressions.html#the-power-operator
https://docs.python.org/3.5/reference/expressions.html#the-power-operator

The docs are correct. See the definition of the "await" expression: https://docs.python.org/3.6/reference/expressions.html#await-expression

Yurys answer and resolution misses the point. Nobody complained about the await operator. Fact is, that in the _power_ operator (chapter 6.5) as base specification now "await" is given. This is likely to be an inadvertent copy of the previous chapter, makes no sense here and should probably be replaced by "u_expr" or similar.

The original bug report is (apart from spelling) correct. This is a bug and it should be closed.

Please reopen.

This patch reverts the erroneous change in:

changeset:   96185:548d5704fcb3
user:        Yury Selivanov <yselivanov@sprymix.com>
date:        Thu May 21 11:50:30 2015 -0400
summary:     Issue 24180: Documentation for PEP 492 changes.

That should have read "... should _not_ be closed."

The docs looks correct to me. It reflects the fact that the "await" keyword has higher priority than the power operator. "await a ** b" is identical to "(await a) ** b", not "await (a ** b)".

Fwiw, I'm -1 on changing this. I think that the current names are correct.

But to avoid confusion I propose following patch. This may be more correct, since it is strange to name the expression without "await" "await expression".

I agree that what is currently in the docs is technically correct.  The section on the power operator:

   power ::=  await ["**" u_expr]

should be read as:

   power ::=  (["await"] primary) ["**" u_expr]

just as the definition of 'u_expr' below should be read as:

   u_expr ::=  (await ["**" u_expr]) | "-" u_expr | "+" u_expr | "~" u_expr

and 'await' should be read as:

   await ::= ["await"] (atom | attributeref | subscription | slicing | call)

Each definition builds upon previous ones.

However, it is confusing.  Could a name other than "await" be used, since "await" is easily confused with the keyword await?

May be "w_expr"? ("a_expr" is already used for addition level).

Let's call it "await_expr"

+1 for await_expr

The confusion is in the definition of await expression with optional "await". Without "await" the expression doesn't suspend the execution of coroutine. And await expression can only be used inside a coroutine function, but the power operator can be used outside.

Ack to the new patch. It is semantically confusing that the await expression also served as the power base without any await.

Sounds like it is a bit more than just confusion: given that power can be used outside a coroutine but await can't, Serhiy's formulation would seem to me to be more semantically correct, even if syntactically it is the same as the current.

I think it would regardless be better to replace 'await' with 'await_expr', so +1 on that from me as well.

I'm with David here, in that a change would improve ease of comprehension:
I still find it strange, that the grammar symbol has the same name as the
keyword (or in the proposed version keyword+'expr'), which is never done on
more widely used levels. I would have expected something like
'coroutine_suspension', which reflects what it is and not which keyword is
required to state it.

2016-05-06 15:04 GMT+02:00 R. David Murray <report@bugs.python.org>:

>
> R. David Murray added the comment:
>
> Sounds like it is a bit more than just confusion: given that power can be
> used outside a coroutine but await can't, Serhiy's formulation would seem
> to me to be more semantically correct, even if syntactically it is the same
> as the current.
>
> I think it would regardless be better to replace 'await' with
> 'await_expr', so +1 on that from me as well.
>
> ----------
> nosy: +r.david.murray
>
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue26156>
> _______________________________________
>

I think "await_expr" is good. It is not new that the keyword is a part of the grammar symbol name (yield_expression, comp_for, and_expr, or_test, lambda_expr).

New changeset e876ff4e9e95 by Serhiy Storchaka in branch '3.5':
Issue #26156: Make expressions grammar description more semantically correct.
https://hg.python.org/cpython/rev/e876ff4e9e95

New changeset 61e3160310ae by Serhiy Storchaka in branch 'default':
Issue #26156: Make expressions grammar description more semantically correct.
https://hg.python.org/cpython/rev/61e3160310ae