Use SHA256 Hash for Key Fingerprints

[srueg]

How would you feel about using the SHA256 hash instead of the current MD5 to calculate public key fingerprints?
https://github.com/jpillora/chisel/blob/v1.7.2/share/ccrypto/keys.go#L35

To remain backward compatible we could add a sha256: prefix to new fingerprints and evaluate old fingerprints (without the prefix) still as md5 hashes (at least until a new major version would allow the breaking change).

References regarding MD5 deprecation:

• https://tools.ietf.org/id/draft-lvelvindron-tls-md5-sha1-deprecate-01.html

[jpillora]

Definitely should move. I was thinking of switching to base64 and sha256, then we wouldn’t need a prefix, just a valid base64 check, and hex would fallback to md5 for legacy support

…

On Tue, 10 Nov 2020 at 12:28 am Simon Rüegg ***@***.***> wrote: How would you feel about using the SHA256 hash instead of the current MD5 to calculate public key fingerprints? https://github.com/jpillora/chisel/blob/v1.7.2/share/ccrypto/keys.go#L35 To remain backward compatible we could add a sha256: prefix to new fingerprints and evaluate old fingerprints (without the prefix) still as md5 hashes (at least until a new major version would allow the breaking change). References regarding MD5 deprecation: - https://tools.ietf.org/id/draft-lvelvindron-tls-md5-sha1-deprecate-01.html — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#214>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAE2X47NJHJKNILLJG54ZJDSO7U6NANCNFSM4TPK6ENA> .

[srueg]

I started implementing a PR for this but realized that this would break the feature of providing only a partial fingerprint in the clients --fingerprint flag: a partial fingerprint will most probably not be a valid base64 string, so it will fall back to MD5 which won't match either.
How would you like to proceed?

1. Only provide auto-fallback for fully specified fingerprints and deprecate partial fingerprints?
2. Go with the sha256: prefix?

[srueg]

I went with option 1 in PR #215