* Add feature controlling the global reference pool to enable avoiding its overhead.
* Document reference-pool feature in the performance guide.
* Invert semantics of feature to disable reference pool so the new behaviour becomes opt-in
* Remove delayed reference count increments as we cannot prevent reference count errors as long as these are available
* Adjust tests to be compatible with disable-reference-pool feature
* Adjust tests to be compatible with py-clone feature
* Adjust the GIL benchmark to the updated reference pool semantics.
* Further extend and clarify the documentation of the py-clone and disable-reference-pool features
* Replace disable-reference-pool feature by pyo3_disable_reference_pool conditional compilation flag
Such a flag is harder to use and thereby also harder to abuse. This seems
appropriate as this is purely a performance-oriented change which show only be
enabled by leaf crates and brings with it additional highly implicit sources of
process aborts.
* Add pyo3_leak_on_drop_without_reference_pool to turn aborts into leaks when the global reference pool is disabled and the GIL is not held
* deprecate `PyCell::new` in favor of `Py::new` or `Bound::new`
* update deprecation warning
Co-authored-by: David Hewitt <mail@davidhewitt.dev>
---------
Co-authored-by: David Hewitt <mail@davidhewitt.dev>
Adds a "threadsafe" variant of `PyCell::try_borrow` which will fail instead of
panicking if called on the wrong thread and use it in `call_traverse` to turn GC
traversals of unsendable pyclasses into no-ops if on the wrong thread.
This can imply leaking the underlying resource if the originator thread has
already exited so that the GC will never run there again, but it does avoid hard
aborts as we cannot raise an exception from within `call_traverse`.
Since we're just testing a bug during traversal, we don't actually have
to reap the object, we just have to make a reference cycle so that it's
traverse method is called.
Adam Reichold