Matthew Neeley

Are you using anything from `cirq-google`? If not, you can install `cirq-core` rather than `cirq` and avoid the dependency on `google-api-core` altogether. It's generally a good idea to install the...

> I wasn't aware that `cirq` is a meta package, and `cirq-core` is what one should install most of the time. This is probably something we should document better. Most...

On the core question of whether `cirq.measure` should not accept sets, I don't think we should do that. There's no way in general to detect whether a given iterable comes...

Yes, the expensive computation is an issue. I worry that trying to track properties such as all qubits in a circuit as it's being mutated is going to be very...

Typically we don't install `cirq` but rather `cirq-core`, `cirq-google`, etc to bring in a minimal set of dependencies. I suspect it's one of the other `cirq-*` packages that is actually...

Why not add the `cached_method` decorator on `_json_dict_` directly? Are you worried that the return value is not immutable?

I've always found the `get_qcs_object_for_notebook` function to be awkward, starting from the name, and I wonder if we can get away from it with a little bit of refactoring. For...

The incantation I usually use is `isinstance(op.untagged, cirq.CircuitOperation)`. But we could certainly add a helper function for this.

I agree this can be a bit tedious. Often typing will help with this, since if you do `isinstance(op.untagged, cirq.CircuitOperation)` then the type checker will know that `op.untagged` has a...

Note: I filed https://github.com/grpc/grpc/pull/36096 to fix these errors upstream. We'll see what comes of that, but in any case it will be a while before we can rely on that...