Jeff Burdges

I doubt some sponge parameters trait could be particularly meaningful. In reality, one should build Merkle trees, etc. around specific sponge configurations, like absorbing exactly 9 field elements. It could...

I'll move this from elsewhere.. In theory, we could temporarily cap no-shows for glutton cores, but instead count no-shows, provided we foresee some benefits to pushing faster and collecting data...

Is it possible to map these to native x86-64/aarch64 instructions directly? AES diserves discussion, because its faster on the hardware we actually use, but I've mostly pushed people towards chacha...

I've suggested this approach for symetric cryptography like hashing before: We've some _lightweight pure host function_ (LPFH) which avoids a context switch, but still uses native code, and enforces jump...

I suppose host function overhead breaks down into primarily context switches and memory copies, with other overhead being very small, yes? And memory copies are being discussed in https://github.com/paritytech/polkadot-sdk/issues/6761 and...

I'd think this works for upgrades too, no? Is there a reason to warn against that?

In part, we're asking: What does the community want in the way of audit notes? We know there is wider interest in audit notes for say unsafe blocks, but that's...

We're okay with "sponges" that do not support converting from sponge back into absorb to give runtime errors? I'd think runtime errors would be the only option for something like...

I'd think scoped traits wind up unsound.. crate x ``` pub trait SomeTrait { .. } put unsafe trait SomeTraitConforms : SomeTrait { } pub fn bar(..) { } pub...

Alright fair enough, but `SomeTraitConforms` need not be a subtrait of `SomeTrait`, so then `pub fn baz(..)` creates problems. Also, you could disable seemingly unrelated impls unexpectedly, right? You cannot...