Joshua Liebow-Feeser

> The only open question I had was would it help to define the layout of `struct Foo()` to be the same as `()`. This would be very helpful. E.g.,...

I haven't looked at the internals at all, so I could be totally off the mark here, but I'd expect epoch GC to notice that the person iterating still had...

I see; that all makes sense. The use case is for an RCU implementation. We want to mimic crossbeam epoch's approach to storing generation indices - in our case, each...

Ah so based on that issue, and from perusing the code, I suspect that just removing the dependency on `json` may do the trick. In particular: - Mutagen itself never...

@stjepang Thanks so much for this comment! It's great to know that this stuff is being actively developed. @ezrosent will probably have is own comments in addition to mine, but...

Also, a quick follow-up to that: With respect to your question about avoiding allocations for no-std, you don't actually need to avoid allocations, you just need to ensure that the...

Also also: I read the [crossbeam RFC](https://github.com/crossbeam-rs/rfcs/blob/master/text/2017-05-23-epoch-gc.md), and with the exception of [dedicated GC threads](https://github.com/crossbeam-rs/rfcs/blob/master/text/2017-05-23-epoch-gc.md#dedicated-gc-threads), I think that everything I've described is orthogonal to the changes that you guys are...

I realized that I'd made a mistake in my original comment: the global static instance should itself be a handle. I've updated the original comment, and I also wrote up...

No problem! > We do want to support the `Alloc` API, although gating it under the `nightly` feature might be painful. I hope it gets stabilized soon. :) No worries...

That's fantastic! Good to know that proposal is actually reasonable enough to be implemented :) It looks like your global [`pin`](https://github.com/jeehoonkang/crossbeam-epoch/blob/dea30b1bb86f3f0c94e80880d43b622ed588bf2e/src/global.rs#L84) implementation unconditionally calls `with`. Do you have an idea...