Ref T13168. This is just a small quality-of-life fix: we can disclose which public key we're talking about because public keys are public.
- Hit public key error (through my own bumbling / not reading or following instructions). Specifically, I haven't associated the key with a device in Almanac.
- Before: vague error.
- After: more specific error with enough key material that I could grep for it.
Isn't the end of the public key way more useful for human readability in general? Most of my keys generated via ssh-keygen (which I assume is the practically universal method?) end in an identifier that I can plausibly use to identify the key, like this one:
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDzzrWAoWk0QFKtGz3860178qb5ue1cMSqrwYnBOJah7SXEUfwCOjXtvfvE3hyShS3SBdhm6qJZfSMngq7jO84Je1T7HU9FFjLg9HUPsfXaPtsojg4HhhN2n1dm+SysdmXScqIDb0g5/JWHenHTRh146t7ptP6+JzSHQGCWBzXsgXFCdlRrcrM+cUIGz9A0AIr9rZBgcwRtp8qK+TG9geA+Vcxf2Lw2T/GUS4nAeuT8KslAY6oC7ge2zT33VUhkvHyWV4iNlTdDHFneMvtdyXiCFa0oZwaQ7PDNw6Hrde7TDOkTi9Pppxawi8UYotsL7CzkeNIRWDNGXGTLyVsxHYBX mriley@mriley-C02TW3N0HTDD
I think structuring it as The public key ending in "...<xyz>". is at least a little weird, and potentially confusing even if it's technically better in most cases.
The comment may also be arbitrarily long and non-unique. (When we generate keys, I think we currently just use the comment "Generated", which in a German localization would of course be "Gensherkeymatthaffenuntermazerhosenauomattermachinen", making the last 64 characters nonunique and nonidentifying.)
Perhaps best would be "aaa ... bbb" but we don't have a convenient UTF8-aware truncation process for that today.
I suppose reversing phutil_utf8v_combined() is a reasonably good approximation of a UTF8 reverser. It's probably not perfect (e.g., RTL marks should be inverted, I guess?) but if we were to implement phutil_utf8_strrev(), reversing phutil_utf8v_combined() would probably be a reasonable v0.