Closed Demi-Marie closed 4 years ago
Thanks!
As written, this should not build, or at least not build for tests, because rand_os 0.2.2 needs rand_core 0.5 https://docs.rs/crate/rand_os/0.2.2 but dalek refuses to upgrade due to semver.
Is substrate on rand_core 0.5 aka rand 0.7 yet? I could just merge https://github.com/w3f/schnorrkel/commit/8150ef6333df6238d8d861aec74bf7ccd87119aa maybe?
I think my only caveats reasons for not merging that yet were:
cargo-features = ["rename-dependencies"]
requires nightly, not sure.I think 1 can be avoided by splitting RngCore5As4
out into a separate crate, which then substrate can use directly to invoke ed25519-dalek. If 2 is even an issue, then it could be addressed by splitting RngCore5As4
out with a second separate crates for its old_rand_core.
What do you think? Should we just split out RngCore5As4
into a subcrate and be done with it?
@burdges Substrate uses both 0.4 and 0.5. I think your answer is a good one.
zeroize
v1.0.0 has been released.
curve25519-dalek v2.0.0-alpha.0 has upgraded rand_core in https://github.com/dalek-cryptography/curve25519-dalek/pull/298 so merlin should soon follow
@burdges do you want me to bump curve25519-dalek to v2.0.0-alpha.0, or should I wait for a release?
I'll do something on this soon, just busy lately, but sure whatever. :)
It's actually more merlin updating rand_core that I care about than curve25519-dalek https://github.com/dalek-cryptography/merlin/blob/master/Cargo.toml#L24 but they've no issue, so maybe I cannot escape the RandCore5As4 thing.
Anyone know what happened with bincode here? I merged this despite the CI failing I guess.
Easy to fix, never mind.
Also remove spurious braces.