Günther Foidl

Thinking about this more... Current crypto-methods have a pattern that "needs" an allocation, like https://github.com/dotnet/Kerberos.NET/blob/a24a9fa221679b30f8126a3a5972110ecf2f93bc/Kerberos.NET/Crypto/Pal/IHashAlgorithm.cs#L7 (For the returning result an allocation must happen.) Is it possible to change to a...

> I'm not terribly concerned about breaking changes throughout these particular interfaces Fine. So I'll try something...(which should ideally result in a PR :wink:)

> .NET Standard 2.0 is the highest this can go as it MUST stay backwards compatible with the full framework Without duplication of all the native interop from .NET Core...

With a target for .NET ST 2.1 it's possible to remove quite a lot of allocations, but this [results in a mess](https://github.com/dotnet/Kerberos.NET/compare/bca6b1c2a340916872cc802cb01dda6f6fbd7538...gfoidl:565dcb1a9842c3d8273e5f2a6ffed3ffc12ea986) (scroll half-way down to see what I mean)...

Ah, I gotcha. Now I've a plan for Windows only.

> Please don't close, merge instead 😉

> It's not stale, just no reaction from the maintainers, so please don't close.

Yeah, you wrote that above. My comment is just for the stupid bot, that want's to close issues even when there was no interaction from a maintainer (I'm not one)....

From https://playcode.io/988904  The seconds input has no `invalid` set, also the red border isn't shown. But the `feedbackInvalid` is shown, which is wrong. This PR corrects that behavior. It...

Is related to https://github.com/dotnet/designs/pull/173? Or is this one here more about AOT?