Test vectors

[cocowalla]

Any chance you could publish some test vectors for wyhash?

This would be really useful for verifying correctness of ports to other languages.

[wangyi-fudan]

Dear Cocowalla:
I am not sure about the meaning of "test vectors". I guess the simplest way to verify a ported code is to run the c version and c# version seperately, and compare them.
thanks!

[cocowalla]

Test vectors are inputs along with their expected outputs, and are commonly published for ciphers and hashes (although for some reason they seem to be more commonly published for cryptographic hashes, rather than non-cryptographic ones 🤷‍♂️).

For example, in the MD5 RFC some vectors are included in section A.5, some are published on the RIPEMD160 website, and NIST publishes test vectors for various algorithms.

I can of course run the C version myself, but it would be really useful to have an authoritative list published here. This might also help, for example, if different results are observed on different processor architectures, or if different compilers produce different results.

[wangyi-fudan]

thanks for your comment! I add test vectors in readme

[eldruin]

Historically at least the hash results have changed in between revisions. In my rust implementation of wyhash I added a small program that downloads the reference version of wyhash for my implementation automatically and generates the hashes and random numbers for a list of inputs which I later use in my tests.

[cocowalla]

@eldruin I had come across your rust port, but hadn't seen this test utility, so thanks for the pointer 😄

[Sahnvour]

Looks like they're gone, would you mind putting them back ?
I guess that the version 2 of wyhash also changed their results.