rustdoc does not emit Message impl for generated types

[fschutt]

I am not sure if I am too dumb, but I couldn't figure out how to actually read a file using the generated code an example would be really helpful.

Using prost, I've generated the rust bindings for the OpenStreetMap .pbf files: https://crates.io/crates/osm-proto-rs

And here is a small PBF file, for testing: https://github.com/sharazam/small-test-pbf-file

How do I read the file, for example the header block? Do I use bytes or just File::open or is there a hidden "Reader" somewhere? Thanks in advance.

[danburkert]

Hi @Sharazam I left a comment on reddit here: https://www.reddit.com/r/rust/comments/6jyhmx/prost_a_protocol_buffers_implementation_for_the/djikvj9/, but I'll expand a bit on your options for decoding from a file. The simplest way is, as I mentioned, to read the entire file into a Vec<u8> using something like Read::read_to_end. If the file is really big and you want to do it more lazily, you could also read-as-you-go, since the OSM format appears to be length-prefixed. Read the length-prefix, then read that much into the buffer. Another option is to use the memmap crate to map the file to a slice, which can then be wrapped in a Cursor to get a Buf impl.

[fschutt]

@danburkert Yes, I know how to read a file. What I don't get is how I should serialize the file into the structs. That's what protobuf is for, isn't it? I've published version 0.2 of the osm-pbf-rs library with the fileformat.proto file added: https://docs.rs/osm-proto-rs/0.2.0/osm_proto_rs/

But there is no "decode" method like you mentioned, nothing. How should I serialize the PBF file into structs, I thought that was what protobuf was for?

[danburkert]

Interesting, I think there may be a bug in rustdoc. The encode method comes from an impl of the Message trait. Every Rust struct generated by prost has a derived impl of Message. I do see the issue where the rustdocs aren't listing that, though, and I'll look into that.

[danburkert]

This was indeed due to a rustdoc bug. See the commit message in 0bdd878 for the gory details. Thanks again for the report!