[Python-Dev] PEP 567 v2 (original) (raw)

Guido van Rossum guido at python.org
Wed Jan 3 00:34:04 EST 2018

On Tue, Jan 2, 2018 at 5:30 PM, Victor Stinner <victor.stinner at gmail.com> wrote:

Hum, it seems like the specification (API) part of the PEP is polluted by its implementation. The PEP just require a few minor changes to better describe the behaviour/API instead of insisting on the read only internal thing which is specific to the proposed implementation which is just one arbitrary implemention (designed for best performances).

Yeah, we need some more words. I still hope someone proposes some, but if threats fail I will have to try to find time myself.

IMHO the PEP shouldn't state that a context is read only. From my point of view, it's mutable and it's the mapping holding variable values. There is a current context which holds the current values. Context.run() switchs temporarely the current context with another context. The fact that there is no concrete context instance by default doesn't really matter in term of API.

I think the issue here is a bit different than Yury's response suggests -- it's more like how a variable containing an immutable value (e.g. a string) can be modified, e.g.

x = 'a' x += 'b'

In our case the variable is the current thread state (in particular the slot therein that holds the context -- this slot can be modified by the C API). The value is the Context object. It is a collections.Mapping (or typing.Mapping) which does not have mutating methods. (The mutable type is called MutableMapping.)

The reason for doing it this way is that Yury doesn't want Context to implement delitem, since it would complicate the specification of chained lookups by a future PEP, and chained lookups look to be the best option to extend the Context machinery for generators. We're not doing that in Python 3.7 (PEP 550 v2 and later did this but it was too complex) but we might want to do in 3.8 or 3.9 once we are comfortable with PEP 567. (Or not -- it's not clear to me that generators bite decimal users the way tasks do. Coroutines always run on behalf of a task so they're not a problem.)

--Guido

Victor

Le 3 janv. 2018 00:34, "Victor Stinner" <victor.stinner at gmail.com> a écrit :

> I would really like to invite more people to review this PEP! I expect I'll be accepting it in the next two weeks, but it needs to go through more rigorous review.

I read again the PEP and I am still very confused by Context.run(). The PEP states multiple times that a context is immutable: * "read-only mapping" * inherit from Mapping, not from MutableMapping But run() does modify the context (or please correct me if I completely misunderstood the PEP! I had to read it 3 times to check if run() mutates or not the context). It would help if the ctx.run() example in the PEP would not only test var.get() but also test ctx.get(var). Or maybe show that the variable value is kept in a second function call, but the variable is "restored" between run() calls. The PEP tries hard to hide "context data", which is the only read only thing in the whole PEP, whereas it's a key concept to understand the implementation. I understood that: * ContextData is immutable * ContextVar.set() creates a new ContextData and sets it in the current Python thread state * When the called function completes, Context.run() sets its context data to the new context data from the Python thread state: so run() does modify the "immutable" context

The distinction between the internal/hiden immutable context data and public/visible "mutable" (from my point of view) context is unclear to me in the PEP. The concept of "current context" is not defined in the PEP. In practice, there is no "current context", there is only a "current context data" in the current Python thread. There is no need for a concrete context instance to store variable variables values. It's also hard to understand that in the PEP. Why Context could not inherit from MutableMapping? (Allow ctx.set(var, value) and ctx [var] = value.) Is it just to keep the API small: changes should only be made using var.set()? Or maybe Context.run() should really be immutable and return the result of the called function and a new context? But I dislike such theorical API, since it would be complex to return the new context if the called function raises an exception. Victor

-- --Guido van Rossum (python.org/~guido) -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.python.org/pipermail/python-dev/attachments/20180102/817448ff/attachment-0001.html>

More information about the Python-Dev mailing list