Jan-Erik Rediger

Fixing the treeherder issue in https://github.com/mozilla/probe-scraper/pull/827

This includes some good changes, that should be included nonetheless. But we should also rethink the "cluster description file" approach (whether we end up with yaml, json, an ini-like format...

What's the output of `swiftc --version` on your machine? I think it might be outdated and the error here makes me believe that your version of Swift doesn't know `async`...

I currently cannot reproduce that in Docker: ``` $ docker build -t rfkelly/uniffi-ci -f docker/Dockerfile-build . $ docker run \ -ti --rm \ -v $HOME/.cargo/registry:/usr/local/cargo/registry \ -v $PWD:/mounted_workdir \ -w...

We're using it in CI. :) Are you using `janerik/uniffi-ci-test:latest`? .... and only just now I realize that we do reference the old name (`rfkelly/uniffi-ci`) in the docs (and in...

Binding generators can be developed externally. As we do not have the need for that usecase right now, it would be up to you to do put in the work...

I don't doubt that it is a valid use case. But we also have only so many hours a day to do stuff and supporting our own use cases (Kotlin/Python/Swift)...

Note: "does not work" is usually not very helpful. If you can it's always helpful to include the full error message you get.

The test I add in #2154 also fails on 0.26. Doc string parsing first appeared in 0.26 here: https://github.com/mozilla/uniffi-rs/blob/cd38ccea8236df7d93aff336c325a3a8e524af5d/uniffi_macros/src/util.rs#L295-L301 So yeah, that never worked since we added that.

I'm not against introducing parallelization where it makes sense, but it does add complexity. I don't think the code generation itself is much of a bottleneck right now. So before...