support 32bit and expand CI #30
Conversation
Codecov Report
@@ Coverage Diff @@
## master #30 +/- ##
==========================================
+ Coverage 86.81% 86.82% +0.01%
==========================================
Files 14 14
Lines 6788 6794 +6
==========================================
+ Hits 5893 5899 +6
Misses 895 895
Continue to review full report at Codecov.
|
|
@dot-asm if you have a moment to sanity check this, I would appreciate it! |
| impl PrimeFieldBits for Scalar { | ||
| // Representation in non-Montgomery form. | ||
| type ReprBits = [u64; 4]; |
There was a problem hiding this comment.
What prevents you from type ReprBits = [u8; 32]; and then
impl PrimeFieldBits for Scalar {
// Representation in non-Montgomery form.
type ReprBits = [u8; 32];
fn to_le_bits(&self) -> FieldBits<Self::ReprBits> {
let mut scalar = blst_scalar::default();
unsafe { blst_scalar_from_fr(&mut scalar, &self.0) };
FieldBits::new(scalar.b)
}
fn char_le_bits() -> FieldBits<Self::ReprBits> {
FieldBits::new(MODULUS_REPR)
}
}
In other words without dependency on target_pointer_width.
There was a problem hiding this comment.
Not exactly related to this, but I've spotted
/// Converts an element of `Scalar` into a byte representation in
/// little-endian byte order.
#[inline]
pub fn to_bytes_le(&self) -> [u8; 32] {
let mut out = [0u64; 4];
unsafe { blst_uint64_from_fr(out.as_mut_ptr(), &self.0) };
out.as_byte_slice().try_into().unwrap()
}
This can also be rewritten using blst_scalar_from_fr. This way you'll remove dependency on [target platform's] endiannes. [At least in this case of course. [Not that I looked for other cases.]]
There was a problem hiding this comment.
What prevents you from type ReprBits = [u8; 32]; and then
The point of this representation is to be u64 or u32s due to how it is used
There was a problem hiding this comment.
The FieldBits accepts a range of types, u8 included, and blst_scalar is in right byte order. Note that cargo test passes...
dignifiedquire
force-pushed
the
expand-ci
branch
from
b84ddb0
to
1d37b87
Compare
|
Docs failure is because of |
Closes #29