Py/PyAny: remove PartialEq impl and add is()

[birkenfeld]

and fix the fallout, mostly in tests.

See #2175 (comment) for discussion.

[birkenfeld]

The guide entry presumes #2175 as merged.

[adamreichold]

Should this inherit the #[inline] attribute from the PartialEq method? (Similarly for the method on PyAny.)

[davidhewitt]

I think probably makes sense for this to be #[inline].

[birkenfeld]

will do

[davidhewitt]

LGTM, thanks for updating migration notes too. It's a shame we can't deprecate trait implementations, however I would hope that any breaking changes users have to undergo are also removing possible bugs from their code 😄

[davidhewitt]

I think probably makes sense for this to be #[inline].

[davidhewitt]

Clippy fix needs rebasing on #2184

[birkenfeld]

I actually found a semi-bug; you can see it in the tests (test_sequence and _pyproto), where vector of objects was compared directly to vec![1.into(),...].

It works since small integers are singletons in CPython, so I kept the logic the same for now, but that's not a guarantee and maybe should be changed to use eq.

[birkenfeld]

BTW, a simple cargo test always results in

   Doc-tests pyo3
error[E0554]: `#![feature]` may not be used on the stable release channel
 --> /home/gbr/devel/pyo3/src/lib.rs:2:21
  |
2 | #![cfg_attr(docsrs, feature(doc_cfg, doc_auto_cfg))]
  |                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

error: aborting due to previous error

for me now.

I can run with nightly, but then the UI tests usually fail, which is not pretty.

[mejrs]

BTW, a simple cargo test always results in

Yeah I was getting annoyed by this too. It should be fixed with #2176

You can delete the rustdocflags in .cargo/config to get around it for now.

[davidhewitt]

Looks great, I'm just going to hold off on merging this for a moment until #2175 is ready.