Replies: 1 comment
-
Hi,
This is a pretty heavy protocol, likely introducing multiple new dependencies to libfido2 and, as you point out, it's not obvious how these map to some libfido2 concepts such as Our general recommendation is that you either try to (a) make it a separate daemon that can synthesize a hidraw device that libfido2 could pick up automatically, or (b) write the transport layer in your application and use That said, here's a couple quick thoughts to your initial questions:
We would probably prefer to not bring in dependencies for encoding a QR code and instead have a function that the application can use to retrieve the data that needs to be encoded. It is then up to the application to do that and to display it to the user.
Unfortunately, I don't have a definite answer to that question. Just like the linked BLE transport implementation, it's not a great fit to our current API.
This is probably a question best answered after having completed the QR-initiated implementation. |
Beta Was this translation helpful? Give feedback.
-
I was looking into giving it a go to implement hybrid transports in particular QR-initiated Transactions on Linux to be able to also using a phone e.g. for SSH logins and maybe git signing.
The main communication flow would be to start the command and then
Does this sound reasonable?
To enable State Assisted Transactions, some of the state could be saved to a file.
I imagine basic usage to work kind of like
Some of the open questions are:
stdout
. However, when this is integrated into another app, that's probably not the cleanest way.token_list
-->fido_dev_info_manifest
)Would there be interest for this type of thing here?
Beta Was this translation helpful? Give feedback.
All reactions