NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
crates.io
crates.io, #3570 most downloaded on crates.io
Implementation of the Pallas and Vesta (Pasta) curve cycle
Latest stable
0.6.1
Last release 1 months ago
09 Oct 2026
Ships unpredictably
gaps range from 2 weeks to 3.4 years
Nearly every release is documented
notes for 14 of 15 stable releases
Nothing withdrawn
no release was ever pulled
6 years old
15 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
asm feature flag, which enables assembly field arithmetic on supported targets. Passing the compiler flag --cfg pasta_curves_noasm compiles the assemb
asm feature flag, which enables assembly field arithmetic on supported--cfg pasta_curves_noasm compiles thepasta_curves::BACKEND, a diagnostic constant identifying the selected"aarch64", "x86-64", or "portable").CurveExt::batch_mul_same_scalar_vartime, for multiplying a batch of affineFp::invert and Fq::invert are faster, while remaining constant-time.sqrt-table feature is faster, improving the performance of{Fp, Fq}::{sqrt, sqrt_alt, sqrt_ratio}.CurveExt::hash_to_curve is faster in release builds.zeroize feature flag, which enables impl zeroize::DefaultIsZeroes for Fp , Fq , Ep , EpAffine , Eq and EqAffine . Zeroizing a field element sets it to
zeroize feature flag, which enables impl zeroize::DefaultIsZeroes forFp, Fq, Ep, EpAffine, Eq and EqAffine. Zeroizing a field elementpasta_curves::{EpAffine, EqAffine}::from_xy_unchecked, a constconst or static itemsCurveAffine::from_xy on every use.pasta_curves::arithmetic:
VartimeField, an extension trait for ff::Field that exposesVartimeBatchInvert, a variable-time equivalent of ff::BatchInvert.impl VartimeField for pasta_curves::{Fp, Fq}.ff 0.14, group 0.14, rand 0.10.pasta_curves::arithmetic:
Base and ScalarExt associated types of CurveExt and CurveAffineVartimeField bound, enabling downstream genericCurveExt trait:
CurveExt::to_affine_vartimeCurveExt::batch_normalize_vartimeYour coding agent can read these notes before it upgrades. Set up the MCP server →
pasta_curves::deferred module, behind the new deferred feature flag. This provides the DeferredField trait and a wide Product accumulator, which toget
pasta_curves::deferred module, behind the new deferred feature flag. This
provides the DeferredField trait and a wide Product accumulator, which
together allow summing many field multiplications (e.g. an inner product)
with a single Montgomery reduction at the end.pasta_curves::glv module, behind the new glv feature flag. This provides
variable-time scalar multiplication for Pallas and Vesta via their cube-root
endomorphism, for use where scalars are not secret (e.g. in verifiers); its
precomputations can be reused across multiplications that share a point or a
scalar.Fix a bug on 32-bit platforms that could cause the square root implementation to return an incorrect result.
sqrt-table feature now works without std and only requires alloc.serde feature flag, which enables Serde compatibility to the crate types. Field elements and points are serialized to their canonical byte encoding (e
serde feature flag, which enables Serde compatibility to the crate types.
Field elements and points are serialized to their canonical byte encoding
(encoded as hexadecimal if the data format is human readable).ff 0.13, group 0.13, ec-gpu 0.2.pasta_curves::arithmetic:
FieldExt bounds on associated types of CurveExt and CurveAffine have
been replaced by bounds on ff::WithSmallOrderMulGroup<3> (and Ord in the
case of CurveExt).pasta_curves::hashtocurve:
FieldExt bounds on the module functions have been replaced by equivalent
ff trait bounds.pasta_curves::arithmetic:
FieldExt (use ff::PrimeField or ff::WithSmallOrderMulGroup instead).GroupSqrtRatio (use ff::Field::{sqrt_ratio, sqrt_alt} instead).SqrtTables (from public API, as it isn't suitable for generic usage).uninline-portable feature flag, which disables inlining of some functions. This is useful for tiny microchips (such as ARM Cortex-M0), where inlining
uninline-portable feature flag, which disables inlining of some functions.
This is useful for tiny microchips (such as ARM Cortex-M0), where inlining
can hurt performance and blow up binary size.Migrated to ff 0.12, group 0.12.
ff 0.12, group 0.12.gpu feature flag, which exposes implementations of the GpuField trait from the ec-gpu crate for pasta_curves::{Fp, Fq}. This flag will eventually cont
gpu feature flag, which exposes implementations of the GpuField trait from
the ec-gpu crate for pasta_curves::{Fp, Fq}. This flag will eventually
control all GPU functionality.repr-c feature flag, which helps to facilitate usage of this crate's types
across FFI by conditionally adding repr(C) attribute to point structures.pasta_curves::arithmetic::Coordinates::from_xypasta_curves::{Fp, Fq} are now declared as repr(transparent), to enable
their use across FFI. They remain opaque structs in Rust code.Support for no-std builds, via two new (default-enabled) feature flags:
no-std builds, via two new (default-enabled) feature flags:
alloc enables the pasta_curves::arithmetic::{CurveAffine, CurveExt}
traits, as well as implementations of traits like group::WnafGroup.sqrt-table depends on alloc, and enables the large precomputed tables
(stored on the heap) that speed up square root computation.pasta_curves::arithmetic::SqrtRatio trait, extending ff::PrimeField with
square roots of ratios. This trait is likely to be moved into the ff crate
in a future release (once we're satisfied with it).pasta_curves::arithmetic:
Field re-export (pasta_curves::group::ff::Field is equivalent).FieldExt::ROOT_OF_UNITY (use ff::PrimeField::root_of_unity instead).FieldExt::{T_MINUS1_OVER2, pow_by_t_minus1_over2, get_lower_32, sqrt_alt,
sqrt_ratio} (moved to SqrtRatio trait).FieldExt::{RESCUE_ALPHA, RESCUE_INVALPHA}FieldExt::from_u64 (use From<u64> for ff::PrimeField instead).FieldExt::{from_bytes, read, to_bytes, write}
(use ff::PrimeField::{from_repr, to_repr} instead).FieldExt::rand (use ff::Field::random instead).CurveAffine::{read, write}
(use group::GroupEncoding::{from_bytes, to_bytes} instead).The crate is now licensed as MIT OR Apache-2.0.
MIT OR Apache-2.0.Migrated to ff 0.11, group 0.11.
ff 0.11, group 0.11.Implementation of group::WnafGroup for Pallas and Vesta, enabling them to be used with group::Wnaf for targeted performance improvements.
group::WnafGroup for Pallas and Vesta, enabling them to be
used with group::Wnaf for targeted performance improvements.Implementations of group::{CofactorCurve, CofactorCurveAffine} for Pallas and Vesta, enabling them to be used in cofactor-aware protocols that also wa
group::cofactor::{CofactorCurve, CofactorCurveAffine} for
Pallas and Vesta, enabling them to be used in cofactor-aware protocols that
also want to leverage the affine point representation.Initial release!
Initial release!
Nothing published for this version
One column per quarter.