core_verify_pk_in_g1 fails to verify valid signature about blst

[dot-asm]

git pull, rebuild and re-test.

Just in case, Go and Rust bindings didn't suffer from this, because they parallelize even core verify, and didn't exercise the problematic code path. Which is how typo slipped through. I'm putting together a python script to keep swig in the loop...

[Nashatyrev]

Yeah, thanks for the prompt fix, but this case still fails for me (pulled, rebuilt and checked). I will double check tomorrow, may be something wrong on my side.
So the described case now passes on your side right? If so, may be you can share your test code?

[dot-asm]

It's in-the-middle-of-it, but you can git clone https://github.com/dot-asm/blst.git -b python.demo and run blst/bindings/python/demo.py. At the end of the script you'll see your inputs.

[dot-asm]

Oh! It's for Linux... Definitely won't work on other platforms...

[dot-asm]

Just in case, script doesn't actually call core_verify, but you can replace everything past "# vet the public key" with result = blst.core_verify_pk_in_g1(pk, sig, True, msg, DST). Note that it returns BLST_ERROR, and 0 stands for success. So in application code you'd write if blst.core_verify_pk_in_g1(pk, sig, True, msg, DST) != blst.BLST_SUCCESS panic.

[Nashatyrev]

Thanks! Your python code works for me both as is and with core_verify_pk_in_g1
Doing verification the same way as in your python demo (with pairing) in Java SWIG binding always results for me in 0 return from pairing_finalverify on both valid and invalid data :/
There must be something goes wrong on a Java wrapper side

[Nashatyrev]

Oh, just realized that pairing_finalverify returns bool. So verification also fails here

[Nashatyrev]

Will close this for now. Thanks for support 👍

[Nashatyrev]

Hurray! I was just incorrectly rebuilding the library. Everything works now 👍

[dot-asm]

python.demo branch switched to c++ class interface. I wonder how would it, class interface, look for you in Java. Keep in mind that this is work in progress...

[Nashatyrev]

Yep thanks! Will probably try C++ bindings a bit later 👍

[dot-asm]

The python.demo branch was updated to return Java byte[] from serialization methods. Check it out... [Not that I fully tested it, but it does compile...]