KATs with data please

[Exaelitus]

Hi! I wrote an implementation in c++ of CircleHash, but I need to ensure that it is correct.

I extracted out 3-4 simple KATs from circlehash64_test.go but those have no data vectors. They are only a seed with a null vector/msg payload. This is great for testing Reset/Init/Seed functionality, and my implementation passes those fine. However, none of the Update/Hash data functionality is exercised.

Could you please publish KATs with seed/iv, data/msg, and expecteds? Even 3-4 more with data vectors would be GREATLY appreciated!

Thanks in advance!

[Exaelitus]

To be specific, I mean simple vectors like "abc". "cba", "cab", "HelloWorld" etc. Rather than the 16k subroutine for data which I would then also have to implement and not know if I have correct, etc.

For example, here are my KATs for Murmur3, in files for my testing of my implementation of it. Obviously, KATs for CircleHash could be anything, and in the format currently used in *_test.go, just illustrating what I mean...

HASH_MURMUR3_32.testvector.txt
HASH_MURMUR3_32.expected.txt
HASH_MURMUR3_32.iv.txt

[fxamacker]

Hi @Exaelitus

To be specific, I mean simple vectors like "abc". "cba", "cab", "HelloWorld" etc. Rather than the 16k subroutine for data which I would then also have to implement and not know if I have correct, etc.

Great point!

BTW, I've been working on publishing CircleHash64fx, which will replace CircleHash64f as the default 64-bit hash in CircleHash.

The new hash will have an improved API and I'll be sure to take your suggestion to include a few simpler-to-implement KATs.

In addition to adding a few simple test vectors, I'll also include an intermediate test like what BLAKE3 uses (which is simpler than the current test that consolidates many thousands of digests into one KAT).
https://github.com/BLAKE3-team/BLAKE3/blob/master/test_vectors/test_vectors.json

[Exaelitus]

That would be perfect, thanks!

In the meantime, do you have 2-3 that I could use to test my current implementation of CircleHash64f ? This would help me know if I'm porting things correctly and be in better position for when you publish CircleHash64fx. No worries if not, thanks!