Title: yield from expression can be any iterable
Type: behavior Stage: patch review
Components: Documentation Versions: Python 3.6, Python 3.5
Status: open Resolution:
Dependencies: Superseder:
Assigned To: docs@python Nosy List: docs@python, terry.reedy
Priority: normal Keywords:

Created on 2016-07-28 18:30 by terry.reedy, last changed 2016-07-28 18:30 by terry.reedy.

Messages (1)
msg271577 - (view) Author: Terry J. Reedy (terry.reedy) * (Python committer) Date: 2016-07-28 18:30 says
  "When yield from <expr> is used, it treats the supplied expression
  as a subiterator. All values produced by that subiterator ...".
To me "treats..expression as a subiterator" means that the expression must *be* an iterator, such as returned by iter or calling a generator function.  Hence I was surprised upon reading "yield from <non-iterator iterable>" in stdlib code.

I confirmed that this usage is correct by trying

>>> def g():
	yield from (1,2)

>>> i = g()
>>> next(i), next(i)
(1, 2)

and then reading the PEP380 Formal Semantics, which begins with "_i = iter(EXPR)".  Hence I suggest the following replacement for the quote above:
  "When yield from <expr> is used, the expression must be an iterable.
  A subiterator is obtained with iter(<expr>).  All values produced
  by that subiterator ...".

Note that 'subiterator' is spelled in the following sentences 'underlying iterable' (which I am not sure I like) and 'sub-iterator' (and 'sub-generator').  I think we should  be consistent for at least the two short 'yield from' paragraphs.
Date User Action Args
2016-07-28 18:30:50terry.reedycreate