Others

We're using rayon now. The biggest future optimization is this one for determining what's rooted: * Track an "invalidation number" for each handle, a u64 that's incremented whenever a lockout...

Well this is because the collector allocates some global datastructures that are never cleaned up. I guess this is technically “working as intended”, but I understand how it’s interfering with...

I doubt you'd see much benefit from using asyncio for shredder. Async shines with IO workloads, shredder is all compute bound I don't want to move away from a global...

So the logs at the top of the issue indicate this is purely a problem with `shredder` not cleaning up its threads at shutdown. If you allocate some actual Gcs,...

Can you test going off the latest content on GitHub? There shouldn’t be any dash map stuff on master, maybe that would clean it up a bit? On Wed, Oct...

I think this should be pretty easy to add--I'll take a look tonight. You may also be able to work around this with the existing API, by getting the &T...

That looks correct. It is actually never possible for the pointer returned by .get().deref() to change (if you look at the implementation it's synthesized directly from a static stored pointer)

`get` is not free, and actually potentially blocking (if collection is happening). So yeah, this definitely needs to be moved internally. See #72

The issue here is the potential performance overhead/collector complexity. I have thought about this a bit, but I'm punting on deciding whether to implement this until the next time I...

@jrmuizel a 4x speedup on the `stress` test I use to benchmark the collector: https://github.com/Others/shredder/blob/master/tests/stress.rs On my machine it goes from ~12 to ~3 seconds. This benchmark kinda blows, and...