Cory Benfield

I think having an OKHTTP based solution for Android is a good idea. I don't think we'll do exactly that here, but I have some other ideas.

The answer is basically by analogy to the nonces in `Crypto`, however a piece of code is missing. The CBC IV should conform to at least `Sequence` though, if not...

We may need to play clever games here, because any extra check here will slow both paths down. Not a lot (it should branch predict well), but it won't be...

The downside of that is it makes the logger even larger, and it also produces slightly-less-than-optimal outcome. What we would have ideal is something close to a two instruction operation,...

So FWIW, retaining the existing semantics won't hurt the performance of the new optimal path, and will only very slightly hurt the performance of the existing path (specifically, getting and...

So you can do it like this: ```diff diff --git a/Sources/Logging/Logging.swift b/Sources/Logging/Logging.swift index 544a739..3059bba 100644 --- a/Sources/Logging/Logging.swift +++ b/Sources/Logging/Logging.swift @@ -69,6 +69,11 @@ public struct Logger { @usableFromInline internal var...

This change also avoids making the structure any larger than it has to be (`LogLevel` and `LogLevel?` are the same size and have the same alignment requirements), and so minimise...

I think that works fine, though I'd still suggest offering the convenience of an actual API spelling for that case.

> I don't think that necessarily swift-service-lifecycle. No disagreement from me, but we also don't need to let perfect be the enemy of good here. > A related question is...

Presumably at this stage that's only possible with a history rewrite though, right? The test-cases haven't really changed so we aren't storing multiple copies.