Add detailed version information information represented by the Git revision to the final binary for better bug tracking.
Can be retrieved by GIT_VERSION string in the STATUS command, in the non-release builds.

Connected:

• Nitrokey/nitrokey-websmartcard-tests#2
• Nitrokey/nitrokey-websmartcard-tests#3
• Nitrokey/nitrokey-websmartcard-tests#1

• Why is everything in a lib directory and not directly in src?
• Please run cargo clippy
• use standard Rust idioms. For example the ERROR_ID) enum should be called Error, with PascalCase variants, and no need for the ERR prefix in each variant.
• The default PIN should probably be 123456 and not 1234 for consistency with other apps
• Why re-export heapless and heapless-bytes?
• You import serde_cbor but don't use it
• Not a fan of adding the peek method to ctaphid_dispatch just for that, but I'm not sure how we could do differently. It should also IMHO be documented why exactly it is there. The dispatch crates already support multiple apps, so the commit names are not clear. Thinking about it, could this be made as a form of "middleware" or a wrapper around the FIDO app? That way ctaphid_dispatch stays "clean"? -> We will talk about it in the trussed steering call
• __MAX_SIZE: You can use #[doc(hidden)] instead of prefixing with __
• Please use pub use transport::Webcrypt. Type alias don't show up properly in doc
• Is the dependency on git-version really necessary? Can't we just use env!("CARGO_PKG_VERSION")?
• You can get rid of the min helper by using value1.min(value2)
• Please move the cbor helper to the helper module
• Is having a send_input_to_output really necessary if it's only used for one (debug) command?
• Please use constants to document "magic values" (for example why 3 in send_input_to_output?)
• It might be good to have a WebcryptClient trait that unites all the trait bounds, which are a bit verbose.
• Have you tested the OpenPGP import functionality? unsafe_inject_shared_key Doesn't work with anything other than symmetric keys.
• Wouldn't we be better off reusing opcard somehow for the OpenPGP stuff? I think users would expect to be able to use their "native" openpgp keys with wecrypt. If no I don't see why the OpenPGP compatibility can't be built into the JS side.
• In command.rs, a comment mention ChaCha20 but ChaCha8 is used

Edit: remaking this ticket into a task list

$ pipenv run pytest test_comm.py
================================================ test session starts =================================================
platform linux -- Python 3.11.5, pytest-7.4.3, pluggy-1.3.0
rootdir: /mnt/bigdrive/soso-nitrokey/Code/nitrokey-websmartcard-tests
configfile: pytest.ini
plugins: reporter-0.5.2, repeat-0.9.3, reporter-html1-0.8.3
collected 71 items                                                                                                   

test_comm.py .

Then the test hang, even if the device is detached (tested with usbip).

Use references in the OpenPGP request fields (e.g. CommandOpenPGPInfoResponse), similarly to the work done in #7

Use curve25519 for the derived keys, which would replace P256 usage completely.

Connected:

• trussed-dev/trussed#44 (comment)
• https://github.com/Nitrokey/trussed-staging/pull/8#discussion_r1246424807
• trussed-dev/trussed#43 (comment)