Repository navigation
asyncio: expose the Future in FutureIter for introspection? #158192
Description
Activity
Exposing it also means that we don't want users to actually mess with it I think. It could be good for debugging but it may be an unstable API that may or may not be reliably protected (like, we don't want UAFs etc. The fact that we expose it to pure Python means adding more checks possibly).
I don't see why we shouldn't expose it. If this thought occurred to me back in the time when i was adding the C version I'd just do this.
Exposing it also means that we don't want users to actually mess with it I think. It could be good for debugging but it may be an unstable API that may or may not be reliably protected (like, we don't want UAFs etc. The fact that we expose it to pure Python means adding more checks possibly).
There's already some existing unsafety in the concurrent handling of
future; #158468 cleans that up, and then my PR for this will depend on that- added a commit that references this issue
on Sep 30, 2026
Feature or enhancement
Proposal:
I am aiming to do some introspection in order to find the Future that an async generator is blocked on.
If it is blocked on a
_PyFuture, chasing down the chain ofcr_awaits andag_awaits eventually gets to a generator forFuture.__await__, from whichselfcould be extracted from the frame.But more realistically, it is blocked on a C future, and the chain leads to a
FutureIterobject. TheFutureIterhas afuturemember, but it isn't exposed.It would be pretty easy to expose it; could we? As
fi_future, maybe? No strong opinion on the name.The most annoying question is what to do about the
_PyFutureversion: should it stay as it is, and introspection would have to look at the frame, or should it match the C version and expose it as an attribute?The implementation of this is super straightforward; I can put up a PR if we think we want to do this.
Has this already been discussed elsewhere?
This is a minor feature, which does not need previous discussion elsewhere
Links to previous discussion of this feature:
No response
Linked PRs