Preston Tunnell Wilson

The mystery deepens---the crate text for `apache-avro` lists: > # Important > > DO NOT USE THIS VERSION OF THE CRATE! IT IS USED FOR TESTING THE PUBLISHING PROCESS ONLY!...

> An **alternative approach** would be to get this information automatically from sections of the `Cargo.toml` > that contain `patch`. @LukeMathWalker what do you think of manually specifying directories to...

@LukeMathWalker what do you think of the test I added? Do you think it's enough of a minimal example?

Here are what I see as the pro's and con's of each of the different approaches: | Approach | Pro's | Con's | | --- | --- | --- |...

> I am not sure I understand what you mean here. On top of the test, could you provide an actual minimal reproducible example as a GitHub repo? It would...

> (you mentioned compilation errors, but you didn't provide details about them). Hi @LukeMathWalker, sorry for the delay. It's been hard to find time to try to minimize the problem,...

> I still don't fully understand the con you mentioned though: > > > Forces users to list all directories in a dependency as a patch, even if they don't...

> Or are we working under the assumption that the patched version of `Pear/lib` brings the whole workspace/large parts of it due to path dependencies? This is correct. At least...

> This is correct. At least for the example, `Pear/lib` depends on `codegen` locally: https://github.com/SergioBenitez/Pear/blob/master/lib/Cargo.toml#L11 I've been looking into recreating the issue in another repo, and it looks like we...

In trying to recreate the issue, I've come across a separate issue using `cargo chef` with dependencies vendored with `cargo vendor`. I've made a minimal repo at https://github.com/prestontw/cargo-chef-vendored-repo (edit: [issue](https://github.com/LukeMathWalker/cargo-chef/issues/74))...