w3f / schnorrkel

Schnorr VRFs and signatures on the Ristretto group
BSD 3-Clause "New" or "Revised" License
309 stars 93 forks source link

Remove serde? #73

Open burdges opened 3 years ago

burdges commented 3 years ago

Is serde too dangerous for key format stability? Should we remove the serde feature?

It's likely fine if someone uses bincode, etc. but if someone use something with unicode or utf-8 then maybe they'd output ligatures or something similarly messed up

Any thoughts @ordian or @becominginsane ?

Related:

burdges commented 1 year ago

https://github.com/dalek-cryptography/ed25519-dalek/pull/291/files

ghost commented 1 year ago

To be honest, I don't see a problem with the current implementation, or how UTF8 etc would impact this, but I suppose this comes into play when using a format that doesn't support raw bytes as "raw bytes" and instead tries to do something fancy with it?

I see in the PR that serdect was suggested, probably worth moving to that instead, so human readable always has hex, and non-human readable always has binary, so that should solve the issue

burdges commented 1 year ago

https://www.bleepingcomputer.com/news/security/rust-devs-push-back-as-serde-project-ships-precompiled-binaries/

https://twitter.com/eddyb_r/status/1693198713286541429 https://twitter.com/eddyb_r/status/1693215683449069981

ghost commented 1 year ago

The prebuilt binary has been removed since 1.0.184 but yeah, I didn't like it either