Yi Lin

Yes. We should. If we have more than 256 variants in the enum, the casting to `u8` will wrap around.

* [ ] For `Timer` (`LongCounter`), `print_total()` and `get_total()` return value in different units (`ms` vs `ns`).

I started CI tests for bindings -- we need to check if the metadata address changes break any binding (they may hard-code the value). Also I also requested a review...

The V8 test failed. But that seems to be a problem from previous PRs. We should fix that in the binding. The API check failed. I am not sure if...

> The failure in Public API Check failed in the "Documenting mmtk" step. Some comments contain cross-references to code which the tool does not recognise. They are not related to...

binding-refs OPENJDK_BINDING_REPO=qinsoon/mmtk-openjdk OPENJDK_BINDING_REF=tph-version

> Is it possible to force regenerating `built.rs` at every `cargo build`? As for the overhead, I guess Rust's incremental build system can avoid rebuilding the entire `mmtk-core` if it...

>This shouldn't be a problem, because we are not tracking git version of the bindings. Actually I just added that change to the OpenJDK binding. In the OpenJDK binding, we...

https://github.com/mmtk/mmtk-core/pull/564 adds a functional implementation for weak reference. Here lists the performance results for the PR: https://github.com/mmtk/mmtk-core/pull/564#issuecomment-1092328984 Remaining work based on https://github.com/mmtk/mmtk-core/pull/564: * [ ] Should we keep Java's weak...

We should look at how servo reports heap usage: https://github.com/servo/servo/blob/master/components/malloc_size_of/lib.rs and this project: https://github.com/paritytech/parity-common/blob/master/parity-util-mem/src/malloc_size.rs