v0.10.0

Quick Summary

snow is now no_std-friendly, with only one required dependency (subtle) now if you bring your own cryptographic provider, thank you @complexspaces!

This release also sees some breaking API changes in the builder to further prevent misuse/abuse - more builder functions now return Result<Self, Error> instead of Self.

The P-256 curve is now also available as a non-default feature for those of you needing more curves in your life.

Finally, this release transitioned the snow codebase to Rust 2024 and has a new MSRV of 1.85.

New

• Add support for NIST P-256 curve by @AlfioEmanueleFresta in #185
• no_std support with alloc by @jmlepisto #183

Improvements

• Removed dependency on rand by @complexspaces in #193
• Update dead docs link for forming NoiseParams strings by @thombles in #129
• Fix documentation reference to parameters in examples/simple.rs by @okdistribute in #113
• Update various links by @robyoder in #151
• Fix nonce incrementing in stateful transport to match the specification by @complexspaces in #152
• s/Ed448/Curve448/ by @dconnolly in #155
• curve25519_dalek 4 by @kayabaNerve in #161
• Bump aes-gcm and chacha20poly1305 by @kayabaNerve in #166
• Simplify example code by @BlackHoleFox in #175
• added example for oneway pattern Xpsk1 by @bluen in #186

Migrating from 0.9

The main change you'll likely notice is the need to add expect(...) or ? to a few builder functions that have been updated to further limit the chance of misuse.

Here are the semver-breaking changes:

• Builder function fallibility: more functions in the builder now return a Result to more explicitly prohibit calling the same setter twice.
• DHChoice::Ed448 has been renamed to DHChoice::Curve448
• Multiple new error enum variants have been added.

PRs

• Update dead docs link for forming NoiseParams strings by @thombles in #129
• Fix documentation reference to parameters in examples/simple.rs by @okdistribute in #113
• Bump curve25519-dalek to v4.0.0-rc.0 by @tarcieri in #148
• Update various links by @robyoder in #151
• Fix nonce incrementing in stateful transport to match the specification by @complexspaces in #152
• s/Ed448/Curve448/ by @dconnolly in #155
• curve25519_dalek 4 by @kayabaNerve in #161
• Bump aes-gcm and chacha20poly1305 by @kayabaNerve in #166
• Simplify example code by @BlackHoleFox in #175
• Add support for NIST P-256 curve by @AlfioEmanueleFresta in #185
• added example for oneway pattern Xpsk1 by @bluen in #186
• Remove direct dependency on rand and rand_core by @complexspaces in #193

Full Changelog: v0.9.6...v0.10.0

v0.10.0-beta.2

This latest beta cleans out the dependency graph, such that snow now only has one required dependency (on subtle)! Shoutout to @complexspaces for the work.

changes since v0.10.0-beta.1

• Updated Rust edition to 2024
• Remove dependency on rand, and feature-gated dependency on getrandom
• Removed unused byteorder dependency

Quick Summary

snow is now no_std-friendly, with only one required dependency now if you bring your own cryptographic provider!

This release also sees some breaking API changes in the builder to further prevent misuse/abuse - more builder functions now return Result<Self, Error> instead of Self.

The P-256 curve is now also available as a non-default feature for those of you needing more curves in your life.

New

• Add support for NIST P-256 curve by @AlfioEmanueleFresta in #185
• no_std support with alloc by @jmlepisto #183

Improvements

• Removed dependency on rand by @complexspaces in #193
• Update dead docs link for forming NoiseParams strings by @thombles in #129
• Fix documentation reference to parameters in examples/simple.rs by @okdistribute in #113
• Update various links by @robyoder in #151
• Fix nonce incrementing in stateful transport to match the specification by @complexspaces in #152
• s/Ed448/Curve448/ by @dconnolly in #155
• curve25519_dalek 4 by @kayabaNerve in #161
• Bump aes-gcm and chacha20poly1305 by @kayabaNerve in #166
• Simplify example code by @BlackHoleFox in #175
• added example for oneway pattern Xpsk1 by @bluen in #186

Migrating from 0.9

The main change you'll likely notice is the need to add expect(...) or ? to a few builder functions that have been updated to further limit the chance of misuse.

Here are all the semver-breaking changes:

• Builder function fallibility: more functions in the builder now return a Result to more explicitly prohibit calling the same setter twice.
• DHChoice::Ed448 has been renamed to DHChoice::Curve448
• Multiple new error enum variants have been added.

New Contributors

• @thombles made their first contribution in #129
• @okdistribute made their first contribution in #113
• @robyoder made their first contribution in #151
• @complexspaces made their first contribution in #152
• @dconnolly made their first contribution in #155
• @kayabaNerve made their first contribution in #161
• @AlfioEmanueleFresta made their first contribution in #185
• @bluen made their first contribution in #186
• @jmlepisto made their first contribution in #183

Full Changelog: v0.9.0...v0.10.0-beta.1

v0.10.0-beta.1

v0.10.0-beta.1

This change brings in some very welcome additions: no_std support, finally, and unofficial support for the P-256 curve.

changes since v0.10.0-alpha.1

• Updated rand_core to 0.9
• Updated criterion to 0.6
• Updated ring and curve25519-dalek versions to the latest version that doesn't have a RUSTSEC warning

New

• Add support for NIST P-256 curve by @AlfioEmanueleFresta in #185
• no_std support with alloc by @jmlepisto #183

Improvements

• Update dead docs link for forming NoiseParams strings by @thombles in #129
• Fix documentation reference to parameters in examples/simple.rs by @okdistribute in #113
• Update various links by @robyoder in #151
• Fix nonce incrementing in stateful transport to match the specification by @complexspaces in #152
• s/Ed448/Curve448/ by @dconnolly in #155
• curve25519_dalek 4 by @kayabaNerve in #161
• Bump aes-gcm and chacha20poly1305 by @kayabaNerve in #166
• Simplify example code by @BlackHoleFox in #175
• added example for oneway pattern Xpsk1 by @bluen in #186

Migrating from 0.9

The main change you'll likely notice is the need to add expect(...) or ? to a few builder functions that have been updated to further limit the chance of misuse.

Here are all the semver-breaking changes:

• Builder function fallibility: more functions in the builder now return a Result to more explicitly prohibit calling the same setter twice.
• DHChoice::Ed448 has been renamed to DHChoice::Curve448
• Multiple new error enum variants have been added.

New Contributors

• @thombles made their first contribution in #129
• @okdistribute made their first contribution in #113
• @robyoder made their first contribution in #151
• @complexspaces made their first contribution in #152
• @dconnolly made their first contribution in #155
• @kayabaNerve made their first contribution in #161
• @AlfioEmanueleFresta made their first contribution in #185
• @bluen made their first contribution in #186
• @jmlepisto made their first contribution in #183

Full Changelog: v0.9.0...v0.10.0-beta.1

v0.10.0-alpha.1

v0.10.0-alpha.1

This change brings in some very welcome additions: no_std support, finally, and unofficial support for the P-256 curve.

New

• Add support for NIST P-256 curve by @AlfioEmanueleFresta in #185
• no_std support with alloc by @jmlepisto #183

Improvements

• Update dead docs link for forming NoiseParams strings by @thombles in #129
• Fix documentation reference to parameters in examples/simple.rs by @okdistribute in #113
• Update various links by @robyoder in #151
• Fix nonce incrementing in stateful transport to match the specification by @complexspaces in #152
• s/Ed448/Curve448/ by @dconnolly in #155
• curve25519_dalek 4 by @kayabaNerve in #161
• Bump aes-gcm and chacha20poly1305 by @kayabaNerve in #166
• Simplify example code by @BlackHoleFox in #175
• added example for oneway pattern Xpsk1 by @bluen in #186

Migrating from 0.9

This should be a zero-diff change for most users of Snow, but there are small semver incompatibilities.

• DHChoice::Ed448 has been renamed to DHChoice::Curve448
• Multiple new error enum variants have been added.

New Contributors

• @thombles made their first contribution in #129
• @okdistribute made their first contribution in #113
• @robyoder made their first contribution in #151
• @complexspaces made their first contribution in #152
• @dconnolly made their first contribution in #155
• @kayabaNerve made their first contribution in #161
• @AlfioEmanueleFresta made their first contribution in #185
• @bluen made their first contribution in #186
• @jmlepisto made their first contribution in #183

Full Changelog: v0.9.0...v0.10.0-alpha.1

v0.9.6

• Validate invalid PSK positions when building a Noise protocol.
• Raise errors in various typos/mistakes in Noise patterns when parsing.
• Deprecate the sodiumoxide backend, as that crate is no longer maintained. We may eventually migrate it to a maintaned version of the crate, but for now it's best to warn users.
• Set a hard limit in read_message() in transport mode to 65535 to be fully compliant with the Noise specification.

Full Changelog: v0.9.5...v0.9.6

v0.9.5

This is a security release that fixes a logic flaw in decryption in TransportState (i.e. the stateful one), where the nonce could increase even when decryption failed, which can cause a desync between the sender and receiver, opening this up as a denial of service vector if the attacker has the ability to inject packets in the channel Noise is talking over.

More details can be found in the advisory: GHSA-7g9j-g5jg-3vv3

All users are encouraged to update.

v0.9.4

This is a dependency version bump release because a couple of important dependencies released new versions that needed a Cargo.toml bump:

• ring 0.17
• pqcrypto-kyber 0.8
• aes-gcm 0.10
• chacha20poly1305 0.10

v0.9.3

This is a quick patch release to use the stable 4.0 version of curve25519-dalek.

v0.9.2

This is a patch release to address a correctness issue for compliance with the Noise specification: the nonce $2^{64} - 1$ is reserved for rekeying, and CipherState and StatelessCipherState did not check that, instead just making sure that there was no integer overflow.

Thanks to @kjvalencik for reporting the issue and @complexspaces for contributing the fix PR (#152).

Thanks to @robyoder as well for fixing broken links and making sure all links were HTTPS (#151).

Full Changelog: v0.9.1...v0.9.2

v0.9.1

This is a patch release to fix build breakages due to not pinning curve25519-dalek to a specific pre-release version.

Thanks to @Kofituo and @thomaseizinger for bringing it to attention and @tarcieri for the fix PR (#148).