[blackshirt]

This PR contains small fix bug when signing message with .with_no_hash options. The bug more clearly self-explained in this snippet

import crypto.ecdsa

fn main() {
        // default prime256v1 key was 32 bytes length
	pv := ecdsa.PrivateKey.new()!
        // creates message with length over key size
	msg_a := 'a'.repeat(34).bytes()
	opt := ecdsa.SignerOpts{
		hash_config: .with_no_hash
	}
	signed := pv.sign(msg_a, opt)!
	pb := pv.public_key()!
	st := pb.verify(msg_a, signed, opt)!
        // this should be true
	dump(st) 

	// different msg should not be verified under the same signature.
	msg_b := 'a'.repeat(392).bytes()
	ds := pb.verify(msg_b, signed, opt)!

	// should false
	dump(ds)
}

but the output was

[code.v:14] st: true
[code.v:19] ds: true

Its changed .with_no_hash to have the same behaviour with .with_recommended_hash. Its can be deprecated on next cycled.

Thanks

[huly-for-github]

Connected to Huly®: V_0.6-22037

[blackshirt]

@spytheman : why the old commit 23612 thats already merged still included in this queue of PR ?

[spytheman]

@spytheman : why the old commit 23612 thats already merged still included in this queue of PR ?

Most likely because you continued your development over the same branch, or on a new branch, based on the old.

You can try doing: git pull --rebase origin master, then git push -f, while being on your ext_ecdsa branch locally, to fix it.