Use the PEP 661 sentinel type where appropriate

[agronholm]

NOTE Erasing or replacing the contents of this template will result in your pull
request being summarily closed without consideration!

Changes

This changes our current sentinels to use the PEP 661 sentinel type instead of an anonymous object.

The only place where this could not be done is the typed_attribute() function, as it is typed to return Any.

This PR needs a newer mypy for type checks to pass, as it doesn't yet recognize typing_extensions.sentinel.

Checklist

If this is a user-facing code change, like a bugfix or a new feature, please ensure that
you've fulfilled the following conditions (where applicable):

• You've added tests (in tests/) which would fail without your patch
• You've updated the documentation (in docs/), in case of behavior changes or new
  features
• You've added a new changelog entry (in docs/versionhistory.rst).

If this is a trivial change, like a typo fix or a code reformatting, then you can ignore
these instructions.

Updating the changelog

If there are no entries after the last release, use **UNRELEASED** as the version.
If, say, your patch fixes issue #123, the entry should look like this:

- Fix big bad boo-boo in task groups
  (`#123 <https://github.com/agronholm/anyio/issues/123>`_; PR by @yourgithubaccount)

If there's no issue linked, just link to your pull request instead by updating the
changelog after you've created the PR.

[tapetersen]

I know I saw at least one place a 1 member enum was used instead just for the debuggability reason. I'll see if I can find it again.

[tapetersen]

Found it (and it seems like the only one).

anyio/src/anyio/lowlevel.py

Lines 102 to 103 in b1f6e40

	class _NoValueSet(enum.Enum):
	NO_VALUE_SET = enum.auto()

[tapetersen]

Ok it uses it to have typing support in the code as well and until that is implemented it's probably not worth trying to change it.

I tried replacing it but mypy hasn't implemented it and with the flag to enable it in pyright they don't support the final semantics it seems yet (Sentinel vs., sentinel)

[agronholm]

Ok it uses it to have typing support in the code as well and until that is implemented it's probably not worth trying to change it.

I tried replacing it but mypy hasn't implemented it and with the flag to enable it in pyright they don't support the final semantics it seems yet (Sentinel vs., sentinel)

Tracking issue here.

[agronholm]

I think we could merge this as-is, and follow up with more when the tooling improves.

[EmmanuelNiyonshuti]

Far as i can tell, this is good aside from waiting on mypy to ship the sentinel support. we can get the actual sentinel type in too (instead of object), once it does!

[agronholm]

Far as i can tell, this is good aside from waiting on mypy to ship the sentinel support. we can get the actual sentinel type in too (instead of object), once it does!

I expect us to switch to Zuban soon which already supports sentinels properly. As for the second part of your comment, I'm not sure what you're referring to.

[EmmanuelNiyonshuti]

Far as i can tell, this is good aside from waiting on mypy to ship the sentinel support. we can get the actual sentinel type in too (instead of object), once it does!

I expect us to switch to Zuban soon which already supports sentinels properly. As for the second part of your comment, I'm not sure what you're referring to.

I was trying to mean we'd follow up with the actual sentinel type:

diff --git a/src/anyio/_backends/_trio.py b/src/anyio/_backends/_trio.py
index 2e67786..b063d09 100644
--- a/src/anyio/_backends/_trio.py
+++ b/src/anyio/_backends/_trio.py
@@ -203,7 +203,7 @@ empty_start_value = sentinel("empty_start_value")
 
 
 class _TrioTaskStatus(abc.TaskStatus[T_contra], Generic[T_contra]):
-    early_start_value: T_contra | object = empty_start_value
+    early_start_value: T_contra | empty_start_value = empty_start_value
     real_task_status: trio.TaskStatus[T_contra | None] | None = None
 
     def started(self, value: T_contra | None = None) -> None:

Basically the follow-up to what I think you've already addressed that this PR needs a newer mypy for type checks to pass.

[agronholm]

Thanks for the review!