search

decentralized-identity / bbs-signature

The BBS Signature Scheme
https://identity.foundation/bbs-signature/draft-irtf-cfrg-bbs-signatures.html
Apache License 2.0
80 stars 26 forks source link

Unique fixture caseName #252

Closed christianpaquin closed 1 year ago

christianpaquin commented 1 year ago

It would be desirable to use unique caseName values in the fixtures, to differentiate them while running them in unit test framworks. For example, "multi-message signature, all messages revealed proof" appears 11 times in the proof fixture files, and all the signature ones are either name "single" or "multi-message signature".

Update: I just realized that the spec uses different names for each test, but the fixtures doesn't update all of them (copy-paste leftover, I assume)

BasileiosKal commented 1 year ago

Discussed on WG call on 13th of Mar.

It would be desirable to use unique caseName

I imagine this means unique per ciphersuite??

Update: I just realized that the spec uses different names for each test, but the fixtures doesn't update all of them (copy-paste leftover, I assume)

Not sure i follow 100%?? Is there places in the document that should be populated by test vectors but are not??

christianpaquin commented 1 year ago

I imagine this means unique per ciphersuite??

Yes.

Not sure i follow 100%?? Is there places in the document that should be populated by test vectors but are not??

The spec is ok, but I'd suggest changing the caseName of the signature.json and proof.json files to match the corresponding files. Otherwise the test output looks like you are running the same test over and over again (e.g, for my lib):

  ✓ signature001: single message signature (417 ms)
  ✓ signature002: single message signature (191 ms)
  ✓ signature003: single message signature (191 ms)
  ✓ signature004: multi-message signature (664 ms)
  ✓ signature005: multi-message signature (502 ms)
  ✓ signature006: multi-message signature (508 ms)
  ✓ signature007: multi-message signature (406 ms)
  ✓ signature008: multi-message signature (354 ms)
  ✓ signature009: multi-message signature (428 ms)
  ✓ proof001: single message signature, message revealed proof (251 ms)
  ✓ proof002: multi-message signature, all messages revealed proof (399 ms)
  ✓ proof003: multi-message signature, multiple messages revealed proof (512 ms)
  ✓ proof004: multi-message signature, all messages revealed proof (373 ms)
  ✓ proof005: multi-message signature, all messages revealed proof (427 ms)
  ✓ proof006: multi-message signature, all messages revealed proof (423 ms)
  ✓ proof007: multi-message signature, all messages revealed proof (29 ms)
  ✓ proof008: multi-message signature, all messages revealed proof (20 ms)
  ✓ proof009: multi-message signature, all messages revealed proof (292 ms)
  ✓ proof010: multi-message signature, all messages revealed proof (311 ms)
  ✓ proof011: multi-message signature, all messages revealed proof (15 ms)
  ✓ proof012: multi-message signature, all messages revealed proof (342 ms)
  ✓ proof013: multi-message signature, all messages revealed proof (348 ms)

So for signatures, sig001 is ok, I'd use "invalid single message signature (modified message)" for sig002, etc.

I could printout these descriptions by parsing the validity and reason (for non-valid tests), but that feels like a lot of trouble; I think it'd be nice for each case to have a different name (within a ciphersuite)