Open codysoyland opened 4 weeks ago
There is still a test failure here, as the test expects that there is no returned error even if there are no valid timestamps (as the count of timestamps is used to test against the threshold, which may be zero). Should we update that test? Or is there a valid reason to expect that behavior?
as the test expects that there is no returned error even if there are no valid timestamps (as the count of timestamps is used to test against the threshold, which may be zero). Should we update that test? Or is there a valid reason to expect that behavior?
Thinking out loud, should VerifyTimestampAuthority
error if there are no entity.Timestamps() to verify?
On the one hand, I'm inclined to say "yes". It feels dangerous to be have a function named "Verify" that can return successfully when it has not actually verified anything. I probably had a hand in writing this function, so I will admit to this being a principle or vibe I only recently adopted; this behaviour got us in trouble in gh at verify
, where we returned successfully when supplied with an empty list of attestations.
On the other hand, as currently implemented there is some ambiguity around how VerifyObserverTimestamps
calls this function; as currently implemented, returning a zero-len array is expected behaviour. However I'm not a huge fan of how VerifyObserverTimestamps
ended up & would be willing to endorse a polite refactor - do we remember why the TlogInclusion
is verified independently of this function?
Summary
Release Note
Documentation