[docs] asyncio.docs: Document subprocess_exec and subprocess_shell (issue 20694)

guido at python.org guido at python.org
Fri Feb 21 00:51:53 CET 2014


http://bugs.python.org/review/20694/diff/11109/Doc/library/asyncio-eventloop.rst
File Doc/library/asyncio-eventloop.rst (right):

http://bugs.python.org/review/20694/diff/11109/Doc/library/asyncio-eventloop.rst#newcode479
Doc/library/asyncio-eventloop.rst:479: list of strings,
:func:`subprocess_exec` takes multiple string arguments.
On 2014/02/21 00:47:30, Yury Selivanov wrote:
> On 2014/02/21 00:32:26, gvanrossum wrote:
> > Maybe add "similar to os.execl()"?
> 
> I think it's too low-level and not really widely used/known to add
reference to
> it, but up to you.

OK, NM.

http://bugs.python.org/review/20694/diff/11109/Doc/library/asyncio-eventloop.rst#newcode503
Doc/library/asyncio-eventloop.rst:503: * *bufsize*: The buffer size to
be used when creating a pipe; this is
On 2014/02/21 00:47:30, Yury Selivanov wrote:
> On 2014/02/21 00:32:26, gvanrossum wrote:
> > I don't actually see a use case for setting this nonzero on any
platform, as
> > buffered I/O doesn't mesh well with async I/O. I'd rather not call
it out in
> the
> > argument list, but instead add a note to the next bullet saying
"(however,
> > *bufsize* should be omitted or zero)". The same is true for shell
and
> > universal_newlines -- they are only present in the signature so we
can
> > explicitly check that they aren't set to different values than what
we need.
> 
> OK
> 
> > Maybe it makes more sense to give an explicit list of the parameters
that *do*
> > make sense to pass to Popen?
> 
> Maybe you can draft a list?

I take it back. They are all extremely obscure (and seem to vary by
version). Let's say "except for bufsize, universal_newlines and shell,
which should not be specified at all". That we actually allow you to
pass redundant values for those will be our little secret. :-)

http://bugs.python.org/review/20694/


More information about the docs mailing list