[docs] Document bytecode changes in 3.6 (issue 28810)

[levkivskyi at gmail.com]

On 2017/02/22 10:04:36, storchaka wrote:
> http://bugs.python.org/review/28810/diff/19372/Doc/library/dis.rst
> File Doc/library/dis.rst (right):
> 
>
http://bugs.python.org/review/28810/diff/19372/Doc/library/dis.rst#newcode980
> Doc/library/dis.rst:980: pushes the return value.
> On 2017/02/22 09:56:13, levkivskyi wrote:
> > I think it is not completely clear what happens with
> > 
> > f(*x, *y, **z, **w)
> > 
> > As I understand, there will be always at most one element on the
stack built
> > with BUILD_MAP_UNPACK_WITH_CALL, and only one element built by
> > BUILD_TUPLE_UNPACK_WITH_CALL. The latter two are not documented yet.
I think
> it
> > would make sense to also document them, and mention them here.
> 
> You understand it correctly. There is separate issue26213 for
documenting
> bytecode changes in 3.5.

The situation is a bit more complex, according to
_bootstrap_external.py, BUILD_MAP_UNPACK_WITH_CALL was added in 3.5, but
BUILD_TUPLE_UNPACK_WITH_CALL was added in 3.6, also semantics of some
bytecodes has changed (e.g. BUILD_MAP), I will therefore make two
separate PRs (one then should be backported to 3.6, another to 3.5 and
3.6). I will reference issue #26213 in the PRs

http://bugs.python.org/review/28810/