--- Log opened Fri Jan 21 00:00:36 2022
02:57 -!- meshcollider [meshcollid@jujube.ircnow.org] has joined #secp256k1
04:48 -!- stickies-v_ [~stickies-@host-92-12-67-127.as13285.net] has joined #secp256k1
04:49 -!- stickies-v [~stickies-@81.178.231.129] has quit [Ping timeout: 256 seconds]
05:58 -!- halosghost [~halosghos@user/halosghost] has joined #secp256k1
10:30 < andytoshi> when updating rust-secp to the latest upstream libsecp, i find that a fixed signing test vector fails
10:30 < andytoshi> this is a bit of a silly test, obviously we don't promise unique signatures, but it's eyebrow-raising
10:30 < andytoshi> did our interpretation of aux data change? i am using secp256k1_nonce_function_bip340, not the default
10:31 < sipa> Yes. https://github.com/bitcoin-core/secp256k1/pull/1002
10:32 < andytoshi> thanks!
10:32 < andytoshi> ah -- i found that, but in this case i am giving a fixed 32-byte aux_rand value
10:32 < sipa> That's strange then.
10:33 < sipa> Is this for BIP340 signing or ECDSA?
10:33 < andytoshi> BIP340
10:34 < andytoshi> valgrind runs clean, i'm pretty sure i don't have API incompatibilites (i think the only API thing that changed is that schnorrsig_verify has a msglen field now)
10:37 < andytoshi> oh lol, no i definitely do have an API mismatch
10:37 < andytoshi> i am giving a custom nonce function to schnorrsig_sign, which no longer takes that (i need to use sign_custom)
10:38 < andytoshi> i am surprised valgrind didn't yell at me for that
10:38 < andytoshi> it appears that i am passing a function pointer where schnorrsig_sign expects aux_random, so (presumably) the signing code is hashing the first 32 bytes of the function code or something
10:38 < sipa> Ouch.
10:39 < andytoshi> lol
10:39 < andytoshi> ok, easy enough to fix, but i'm concerned about how to find all of these
10:39 < andytoshi> i guess i should just diff the header files and manually check that every change is reflected in the bindings
10:40 < andytoshi> but i was really hoping the tests would be able to catch shit like this :(
13:14 < nickler> arguably your test caught this
13:15 < nickler> releases with a changelog would help with this *cough* https://github.com/bitcoin-core/secp256k1/pull/1055
13:15 < sipa> Once we have releases I'd hope we don't make such API changes anymore in the first place ;)
13:19 < nickler> if we continue to have experimental modules we might have to
13:27 < sipa> That's fair... though I also hope that we can make the schnorrsig module non-experimental before that happens.
13:36 < nickler> hm, a 0.x release with a still experimental schnorrsig module wouldn't hurt afaict and immediately benefit users like rust-secp
14:17 -!- halosghost [~halosghos@user/halosghost] has quit [Ping timeout: 268 seconds]
14:17 -!- halosghost [~halosghos@user/halosghost] has joined #secp256k1
14:50 -!- meshcollider [meshcollid@jujube.ircnow.org] has quit [Changing host]
14:50 -!- meshcollider [meshcollid@user/meshcollider] has joined #secp256k1
16:13 -!- stickies-v_ [~stickies-@host-92-12-67-127.as13285.net] has quit [Ping timeout: 256 seconds]
16:15 -!- stickies-v [~stickies-@81.178.226.109] has joined #secp256k1
--- Log closed Sat Jan 22 00:00:37 2022