Defined the SignerMut trait

[rozbb]

As per our discussion on the RCIG Zulip, there seems to be a need for a stateful version of the Signer trait. This lets us support "Key Evolving Signature Schemes" such as XMSS.

One thing that is unclear so far is whether we also want to do stateful versions of DigestSigner, RandomizedSigner, and AsyncSigner. I think the latter two won't be an issue for now, since XMSS is a deterministic scheme, and HSMs do not support XMSS.

[tarcieri]

Our other stateful types are named *Mut, for example:

• https://docs.rs/aead/0.4.3/aead/trait.AeadMut.html
• https://docs.rs/cipher/0.3.0/cipher/trait.BlockEncryptMut.html
• https://docs.rs/cipher/0.3.0/cipher/trait.BlockDecryptMut.html

[rozbb]

Fixed

[tarcieri]

This looks fine to me.

Our other *Mut traits also have blanket impls of the stateful *Mut trait for types which impl the stateless trait, which might also make sense here.

[tarcieri]

Thank you!

[rozbb]

Sorry one more thing: I just pushed another commit to this branch. It wasn't 100% clear to me that my blanket impl wasn't trivially gonna loop forever. Feel free to copy into the tree

[tarcieri]

Oh huh, can you make sure one way or another and open another PR if it's broken?

[rozbb]

Will do

…

El sep. 5, 2021, a la(s) 10:41, Tony Arcieri ***@***.***> escribió:  Oh huh, can you make sure one way or another and open another PR if it's broken? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[rozbb]

Update: it's not broken. I still don't love how it looks (under the hood, the steps it takes are x := &&mut self, y := &**x, y.sign(...)), but you can decide if it's worth changing.