Well great, I didn't try regenerating the protos as noted in the description/commit message, but it seems if you do regenerate them they no longer compile per the generated-protos-compile CI job

The core problem seems to be that crate::google::protobuf::Duration (i.e. prost_types::Duration) is no longer a Copy type, which means no struct that contains it can be a Copy type any more.

I'm not sure why this change was made. It's a simple struct with two public fields which are themselves Copy types (i64 and i32)

Sidebar: in cosmos-sdk-proto and ibc-proto we've discussed ditching prost-types and generating the google.protobuf protos directly from the protobuf schema using the same conventions as everything else, which among other things allows integrations with pbjson for building JSON serializers /cc @romac

See discussion here: cosmos/cosmos-rust#459 (review)

The core problem seems to be that crate::google::protobuf::Duration (i.e. prost_types::Duration) is no longer a Copy type, which means no struct that contains it can be a Copy type any more.

I'm not sure why this change was made. It's a simple struct with two public fields which are themselves Copy types (i64 and i32)

prost_types::Duration still implements Copy in prost-types 0.13.1: https://docs.rs/prost-types/latest/prost_types/struct.Duration.html#impl-Copy-for-Duration

Sidebar: in cosmos-sdk-proto and ibc-proto we've discussed ditching prost-types and generating the google.protobuf protos directly from the protobuf schema using the same conventions as everything else, which among other things allows integrations with pbjson for building JSON serializers /cc @romac

See discussion here: cosmos/cosmos-rust#459 (review)

That's already what we are already doing in tendermint-rs, see the generated code here: https://github.com/informalsystems/tendermint-rs/blob/main/proto/src/protobuf.rs

These types never implemented Copy in the first place, but we can add it manually. I can take care of that.

We should probably generate code for all google.protobuf types (ie. including Any) instead of copy-pasting a subset of it, and then use those directly from ibc-proto-rs instead of the latter having its own copy of them.

Otherwise one cannot pass a tendermint_proto Duration where an ibc-proto Duration is expected and vice-versa, same for the other well-known types.

What do you think?

@romac sounds good to me! Seems like the first step towards unifying protos between tendermint-proto, cosmos-sdk-proto, and ibc-proto

I'd just ask to retain Any::{from_msg, to_msg}: https://docs.rs/prost-types/latest/prost_types/struct.Any.html#implementations

I'd just ask to retain Any::{from_msg, to_msg}: https://docs.rs/prost-types/latest/prost_types/struct.Any.html#implementations

Yes good idea!

Let's do all this in a separate PR.

@romac I didn't put tendermint-proto on the front initially because the changes impact multiple crates