Thomas Eizinger

Bumped the priority because we got a customer request for it.

I think I am running into the same bug. For me, the event is an invite received from Google Calendar. Here is the event exported from "Etar" where it shows...

> I think it's important to distinguish the possible types of determinism: > > * Determinism over multiple runs of the same binary is easy to provide, and something I'm...

> Currently it is not a breaking change for rcgen for example to change the default set of certificate fields. I don't think we should promise this to never change....

Yeah that totally makes sense! This makes me think that a "best-effort deterministic certificate" is a better API than exposing randomness. For one, it is closer to the actual usecase...

> > WebTransport offers the ability to connect to servers with self-signed certificates if you know the certificate hash in advance which is tightly coupled to deterministically generating certificates. >...

> This still sounds like you want to do a more elaborate rcgen API on top of the basic _ring_ API. I (still) think that's a bad idea -- the...

That all makes sense but it is not going to work because `ring`'s `SecureRandom` trait is sealed, meaning you can't implement it for your own struct. See https://docs.rs/ring/0.17.2/src/ring/rand.rs.html#23. So I...

> AFAICT since ring's `EcdsaKeyPair` constructors now all require a `&dyn SecureRandom` and given that that trait is sealed, it basically means that _ring_ itself doesn't support the feature you're...

> > AFAICT since ring's `EcdsaKeyPair` constructors now all require a `&dyn SecureRandom` and given that that trait is sealed, it basically means that _ring_ itself doesn't support the feature...