[noise] New test vectors

[Rhys Weatherley]

On Fri, May 20, 2016 at 5:30 PM, Trevor Perrin <trevp at trevp.net> wrote:

> To keep it simpler:  What if the presence of any unknown tag (e.g.
> init_ssk or resp_ssk) just causes the test to be skipped?
>
> We might also have to skip tests with unknown patterns, because there will
> be more patterns defined, and not every library will support all of them.
>

I'm OK with this.

I have now created some tests for fallback scenarios.  They are tagged with
a simple "fallback": true in the test case.  I have updated the wiki page
with a description of how Noise Pipes can be tested plus a link to the new
test cases from Noise C.

Feedback is appreciated.  I would also appreciate it if others could try
out these new vectors as the IK to XXfallback transition and role reversal
is tricky and I might have messed up somewhere.

The current method for defining fallback test vectors is very primitive.
As discussed elsewhere we need a more general method.  The test vector
format can be updated later once we figure out what that may be.

Cheers,

Rhys.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://moderncrypto.org/mail-archive/noise/attachments/20160521/69c984ac/attachment.html>