Re: [openpgp] Expected client behaviour ambiguity in signature verification

[Justus Winter <[e-mail]>]

Peter Gutmann <[e-mail]> writes:

> Justus Winter <[e-mail]> writes:
>
>>I wrote a test to explore how the different implementations behave:
>>
>>  https://tests.sequoia-pgp.org/#Signature_digest_prefix
>>
>>Every implementation I test sets the digest prefix correctly (GPGME's support
>>for gpg1.4 seems to be faulty, we see a lot of unexpected failures for
>>gpg1.4.)
>
> You may want to indicate in the text whether a check means "handles an
> incorrect prefix" or "checks for and rejects an incorrect prefix".

In the test suite, a checkmark means that the test suite did an
operation, the operation succeeded, and the result of the operation was
not obviously incorrect (e.g. the test suite may check for a decryption
operation that the expected plaintext is returned to prevent
implementations from claiming success while producing no output at all).

This is orthogonal to whether we think the operation should succeed or
not.  Some test vectors have an expected outcome.  This is indicated in
the two rightmost columns "expectation" and "comment".  For the test
vectors that do have such an expectation, we check whether the
implementations match the expectation.  This is indicated by the green
and red background (and the left and right earmark).

See https://tests.sequoia-pgp.org/#how-to for a more verbose description
and a sample test result illustrating how the results are presented.

Best,
Justus