--- Log opened Thu Jul 08 00:00:00 2021
13:55 -!- BlueMatt [~BlueMatt@ircb.bluematt.me] has quit [Ping timeout: 268 seconds]
13:55 -!- BlueMatt [~BlueMatt@ircb.bluematt.me] has joined #secp256k1
14:17 < elichai2> https://usercontent.irccloud-cdn.com/file/a2GXwHdN/Screen%20Shot%202021-07-09%20at%200.17.05.png
14:17 < elichai2> It looks like github's copilot knows libsecp by heart haha
15:11 -!- jesseposner [~jesse@2601:647:0:89:616b:5a1b:543d:92b] has quit [Ping timeout: 240 seconds]
15:13 -!- jesseposner [~jesse@2601:647:0:89:6434:3ce9:af91:46cf] has joined #secp256k1
17:22 -!- reallll [~belcher@user/belcher] has joined #secp256k1
17:25 -!- belcher_ [~belcher@user/belcher] has quit [Ping timeout: 252 seconds]
17:59 < gmaxwell> elichai2: they probably don't eliminate duplication from the many bitcoin forks that include copies of it.
18:00 < gmaxwell> so it's probably massively overtrained on it.
18:00 < gmaxwell> probably the varrious bitcoin projects need contributor notes requesting people to not submit code authored that way (or at least disclose that it was used), out of concern that they'll inadvertantly create copyright problems for the project.
18:03 < roconnor> It's unclear to me that those snipits in the screen shot are even significant enough to fall under copyright.
18:04 < roconnor> with the possible exception of ge_set_gej, which is a bit longer, but has the virtue of being nonsense and not compiling.
18:05 < roconnor> oh wait ge_set_gej is there twice
18:05 < roconnor> and are both nonsense.
18:05 < roconnor> three times.
18:05 < roconnor> I'm not sure any of this code makes any sense.
18:06 < roconnor> okay non if this is libsecp256k1 code.
18:06 < roconnor> it is just vaguely libsecp256k1 shaped code.
18:07 < roconnor> maybe _ge_set_infinity is actual working code at the bottom there.
18:07 < roconnor> consisting of one line.
18:08 < gmaxwell> oh sure, but it certantly *can* emit big blocks that end up violating copyright, though it doesn't often.
18:09 < roconnor> it renamed the parameter in _ge_set_infinity.  Clever.
18:09 < gmaxwell> It has the other problem of emitting code that looks right but is nonsense at a rate much higher than humans. I imagine spammers powered by it flooding with small pull requests would make accepting changes from outsiders pretty hard (not a problem so much for libsecp256k1, but probably one for Bitcoin Core)
23:36 -!- reallll is now known as belcher
--- Log closed Fri Jul 09 00:00:01 2021