Hello Mikhail, Am 2019-02-22 um 12:53 schrieb Mikhail Glushenkov:
Would you as developers of hackage-security / Cabal / stack be interested in adopting such a replacement? And, if yes, what conditions would such a library have to fulfill?
If it doesn't result in too much of a slowdown, I think it would make sense to do this change. I'd recommend making the choice between ed25519/eccrypto configurable via a flag so that we could compare the two. I'm cc:ing Edsko, who is the main author of hackage-security.
That sounds prudent. I included some basic benchmarks and depending on the CPU used I get some triple digit signatures/verifications per second per core (even on Sandy Bridge; only low-hanging fruit optimizations used until now). ed25519 is faster, but I am unsure what your intended performance profile is.