https://docs.python.org/3.6/library/stdtypes.html#hashing-of-numeric-types
Look at semantics example for hash_complex function. The variable 'hash' is clearly misused - it is not defined or refers to builtin function 'hash'. Either way the example is incorrect.

Moreover this example contains following code
if hash_ == -1:
   hash_ == -2
return hash_
that doesn`t make too much sense.

Indeed, nicely spotted! Patch attached (hash_complex_type_1.patch)

P.S.: I don't like the 'hash_' name as a variable for this example; it  looks like grit on Tim's monitor to me, so I went ahead and renamed it to 'hash_value' in that block. Attached as a separate patch for good measure (hash_doc_renamed_1.patch)

Thanks)

Do note that nothing has been changed in the docs yet. I submitted the patch, and now other developers need to review it, decide whether it's correct or need modifications, and then someone needs to commit it. Let the bikeshedding start =)

Adding a _ after a keyword is a very common trick.  The fact that one got lost indicates that it isn't necessarily optimal :).  I don't really have an opinion one way or the other on the variable rename.

The changes look good to me. Thanks for spotting this!

New changeset a6be689eacf8 by Berker Peksag in branch '3.5':
Issue #27601: Improve example in the "Hashing of numeric types" section
https://hg.python.org/cpython/rev/a6be689eacf8

New changeset 06bbcbaeeb18 by Berker Peksag in branch 'default':
Issue #27601: Merge from 3.5
https://hg.python.org/cpython/rev/06bbcbaeeb18

Looks good to me too. 'hash_value' looks more readable to me, thanks!

Originally I emphasized that this bug occurs at every py3k version since 3.2. For some reason version list somehow shrinked to only tow of them - 3.5 and 3.6. I am very sorry, this is probably my fault.

> I am very sorry, this is probably my fault.

It's not your fault :) I removed them.

3.2, 3.3 and 3.4 are in security-fix-only mode so they won't get any documentation updates. You can read https://docs.python.org/devguide/devcycle.html#secbranch for details about security branches.

Thanks for the report!