(original) (raw)
On Wed, Jan 3, 2018 at 12:32 PM, Glenn Linderman <v+python@g.nevcal.com> wrote:
On 1/3/2018 11:16 AM, Guido van Rossum wrote:
Maybe I should clarify again what run() does. Here's how I think of it in pseudo code:
def run(self, func, \*args, \*\*kwds):old = \_get\_current\_context()new = old.copy()\_set\_current\_context(new)try:return func(\*args, \*\*kwds)finally:\_set\_current\_context(old)
I find it interesting that self isn't used in the above pseudo-code. I thought that Context.run() would run the function in the "context" of self, not in the context of a copy of "current context".
Heh, you're right, I forgot about that. It should be more like this:
def run(self, func, \*args, \*\*kwds):
old = \_get\_current\_context()
\_set\_current\_context(self) # <--- changed line
try:
return func(\*args, \*\*kwds)
finally:
\_set\_current\_context(old)
This version, like the PEP, assumes that the Context object is truly immutable (not just in name) and that you should call it like this:
contextvars.copy\_context().run(func, )
The PEP definitely needs to be clearer about this -- right now one has to skip back and forth between the specification and the implementation (and sometimes the introduction) to figure out how things really work. Help is wanted!