Robin Krahl

For reference, here is my current implementation: - https://github.com/Nitrokey/trussed/pull/9 - https://github.com/Nitrokey/trussed/pull/10 I’ll update this PR once we’ve finished testing the integration into the firmware. While this adds some additional type...

> this could be improved with more code generation I’d rather reduce the amount of code generation in Trussed because I feel that the code generation macros often are not...

I would separate three different cases: - Firstly, we have types like `Client` that are the main Trussed API and are used in all components related to Trussed. I agree...

Closing in favor of https://github.com/trussed-dev/trussed/pull/68

I cannot reproduce this issue with a Nitrokey 3 CN running firmware version 1.2.0 on cloud.nitrokey.com. For me, all steps (registration, passwordless login, 2FA) work.

Good to see `from_array` being added! Unfortunately, the implementation is not `const`. It would still be useful to be able to create non-empty constants. My use case for `fill` is...

I don’t think `extend` + `repeat` would work as [`extend`](https://docs.rs/heapless/latest/heapless/struct.Vec.html#method.extend) “[p]anics if the vec cannot hold all elements of the iterator.” Here is a real world example: https://github.com/trussed-dev/ctap-types/blob/084db87a800e1204bcfdc3e25ca46bc4a81650ba/src/ctap2.rs#L148-L175 `serialize` receives...

> Presumably you need to do that audit anyway and (hopefully) already do it for all the other panics (`unwrap()`, slice indices and ranges, or all the other **already panicing**...

> Also our developer uses usb-ip heavily and does not see this issue. In our normal setup, we only use it for the firmware though. I tried repeating the tests...

I suspect that this is caused by a bug in the USB descriptor parsing in the `ctap`/`ctap_hmac` crate. See https://github.com/trussed-dev/usbd-ctaphid/issues/1#issuecomment-1430148638 for more information. I’ve prepared a patch for the `ctap`...