upstream snmalloc-rs as part of snmalloc
is this still in progress?
I am still okay with this and I can continue contributing to the rust part at upstream.
If this is the decided approach to go, I will archive the downstream and put a pointer to upstream in its README.
Pros and cons on top of my head:
Pros
- Easier to sync up.
- Unified CI to make sure everything works
Cons
- CXX project will need to pull in extra rust code.
- Loss of downstream history
Some other issues:
Maybe it is better to unify the versions in some future releases.
Additional pro: Bazel users can directly depend on the snmalloc-rs target (assuming the pr is merged). They won't have to add both.
Cons
2. Loss of downstream history
You could look into git subtree merge if you want to preserve the history. There are some tangentially applicable docs on that in https://docs.freebsd.org/en/articles/committers-guide/#vendor-import-git