Lucas Åström

Results 7 comments of Lucas Åström

Makes perfect sense, I've adjusted accordingly. I also got rid of the `atomic` feature and just used `spin` instead as it's more clear.

> But do we need to `spin`-based at all if we have either `core::sync::atomic` or `atomic-polyfill`? Does `atomic-polyfill` not support MIPS and PPC? [I don't think so.](https://crates.io/crates/atomic-polyfill) Any architectures not...

> That would mean that platforms not supporting the atomics fallback of our choice do not have access to prepared queries, but that appears to be a rather graceful degradation...

As atomics are used elsewhere in `hecs`, keeping only `spin` was not an option for supporting both MIPS/PPC and non-CAS targets. I've instead kept only `atomic-polyfill` and put prepared queries...

> If atomics are used elsewhere and hence `atomic-polyfill` is a mandatory dependency, we should be able to keep prepared queries without a feature gate as well as the world...

Gonna revisit this with the new changes to `portable-atomics` whenever I have time.

Thanks for taking a look! I don't see this as a high priority feature or anything, I was mostly just wondering if I missed something obvious. You're right, it seems...