Michael Woerister
Michael Woerister
That's a good point, @jackh726. I think this should be addressed by the guidance section of the policy since it is one of the main practical issues we currently have....
For reference, during the initial development of incremental compilation we had the [cargo-incremental](https://github.com/nikomatsakis/cargo-incremental) tool that did something similar. The main problem I remember was that we tested large parts of...
The commit message could indicate which command was used to build that commit.
> I wonder if it might be worth it to implement HashStable for FxHashSet directly, and remove the implementation for HashSet. FxHashSet is just an alias for the regular HashSet...
Yeah, those errors are really hard to decipher. @bjorn3 is right: https://github.com/rust-lang/rust/blob/master/compiler/rustc_middle/src/query/mod.rs defines all queries in the compiler. All query keys and results must implement HashStable -- and HashSet (which...
> The current problem is that some places require .iter(), I recognize that as one of the things we'd like to avoid in StableSet. I have a few suggestions, maybe...
Is the always the right choice `create_stable_hashing_context`.
One thing that I'm noticing here is that we probably can reduce the number of instances where StableSet is used. We only need to use StableSet for things that implement...
I resolved a few more compilation errors here: https://github.com/michaelwoerister/rust/commit/6cf3e797364abf1a7fa6e83344307eded0efd259 But yes, I think it would be good to come up with an API design that allows us to safely iterate...