test: add SSZ test vectors by unnawut · Pull Request #1215 · ReamLabs/ream

unnawut · 2026-02-10T14:19:18Z

What was wrong?

Implement leanEthereum/leanSpec#380

Removed TreeHash derive from SignedAttestation because it's not used and it's in turn requiring Signature to implement TreeHash as well which requires custom logic but also not used.
Change Signature::inner from FixedBytes<SIGNATURE_SIZE> to LeanSigSignature so that the final Signature scheme type is used directly without needing FixedBytes in-between
Implemented Encode/Decode on Signature replacing the FixedBytes implementation
Because leanSpec is giving us the expected outputs in json and camelCase format, we have to parse it to Ream/Rust types. So I created passthrough_conversion, simple_conversion, nested_conversion, custom_conversion to handle the json camelCase conversion to Ream types.

To-Do

I have read CONTRIBUTING.md.
This PR title follows conventional commits.

unnawut · 2026-02-10T14:19:38Z

I need to do another round of self-review before marking this ready for review

unnawut · 2026-02-11T15:10:30Z

crates/crypto/post_quantum/src/leansig/signature.rs

-    pub inner: FixedBytes<SIGNATURE_SIZE>,
+    pub inner: LeanSigSignature,


Not sure if this is the right way to do it, but since FixedBytes is how Ream stores it internally and needs conversion to LeanSigSignature to do any operation anyway, so I'm thinking we could just remove this interim FixedBytes and use LeanSigSignature directly

Will LeanSigSignature encode to FixedBytes though is the question? I will read your code, but that would be the concern

In case it's useful this is what I recall. I think FixedBytes was introduced to the spec in devnet0 just because leanSig wasn't stable yet so we just used FixedBytes<SIGNATURE_SIZE> as the mock Signature type in the specs.

And then in devnet1 we started with Bytes3116 in the specs even with the actual signatures.

But towards end of devnet1, the class Signature(Bytes3116) was removed from the specs by leanEthereum/leanSpec#210 and the canonical Signature type becomes this class Signature(Container).

So since then, leanSig and leanSpec uses the Signature container type that's variable length, and the fixed bytes is no longer used. And so that's why I'm thinking we could just use LeanSigSignature directly without conversion to fixed bytes. But I dunno maybe I'm missing something...

Yeah I did some regression testing and this looks fine 👍

…ead of fixed bytes

KolbyML · 2026-02-17T06:59:31Z

testing/lean-spec-tests/src/types/ssz_test.rs

+/// Creates a passthrough JSON wrapper that deserializes directly to the inner type.
+macro_rules! passthrough_conversion {
+    ($name:ident, $inner:ty) => {
+        #[derive(Debug, Deserialize, Clone)]
+        #[serde(transparent)]
+        pub struct $name(pub $inner);
+
+        impl TryFrom<&$name> for $inner {
+            type Error = anyhow::Error;
+            fn try_from(value: &$name) -> anyhow::Result<Self> {
+                Ok(value.0.clone())
+            }
+        }
+    };
+}
+
+/// Creates a TryFrom impl where all fields are copied directly.
+macro_rules! simple_conversion {
+    ($json:ident => $target:ty { $($field:ident),+ }) => {
+        impl TryFrom<&$json> for $target {
+            type Error = anyhow::Error;
+            fn try_from(value: &$json) -> anyhow::Result<Self> {
+                Ok(Self {
+                    $($field: value.$field),+
+                })
+            }
+        }
+    };
+}
+
+/// Creates a TryFrom impl where all fields are converted via try_into().
+macro_rules! nested_conversion {
+    ($json:ident => $target:ty { $($field:ident),+ }) => {
+        impl TryFrom<&$json> for $target {
+            type Error = anyhow::Error;
+            fn try_from(value: &$json) -> anyhow::Result<Self> {
+                Ok(Self {
+                    $($field: (&value.$field).try_into()?),+
+                })
+            }
+        }
+    };
+}
+
+/// Creates a TryFrom impl with custom conversion expression per field.
+macro_rules! custom_conversion {
+    ($json:ident as $val:ident => $target:ty { $($field:ident: $conv:expr),+ $(,)? }) => {
+        impl TryFrom<&$json> for $target {
+            type Error = anyhow::Error;
+            #[allow(clippy::redundant_closure_call)]
+            fn try_from($val: &$json) -> anyhow::Result<Self> {
+                Ok(Self {
+                    $($field: $conv),+
+                })
+            }
+        }
+    };
+}
+


wouldn't the code just be easier to read without macro's? this kind of feels like we are abusing macros here without much benefits.

KolbyML · 2026-02-17T07:01:43Z

crates/crypto/post_quantum/src/leansig/signature.rs

    /// Create a mock signature for testing purposes.
+    ///
+    /// Note: This generates a real signature which is expensive. Prefer `blank()` when
+    /// you just need a placeholder signature.
    pub fn mock() -> Self {
        use rand::rng;


can you just remove this function as it isn't used

KolbyML · 2026-02-17T07:02:36Z

crates/crypto/post_quantum/src/leansig/signature.rs

+    /// Create a blank/placeholder signature.
+    ///
+    /// This decodes from minimal valid SSZ bytes, avoiding expensive key generation.
+    /// Only use in contexts where the signature won't be validated.
    pub fn blank() -> Self {
-        Self::new(Default::default())
+        // 40 bytes: offset_path(4) + rho(28 zeros) + offset_hashes(4) + path(4)
+        const BYTES: [u8; 40] = [
+            36, 0, 0, 0, // offset_path = 36
+            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // rho (28 zeros)
+            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, //
+            40, 0, 0, 0, // offset_hashes = 40
+            4, 0, 0, 0, // path: empty HashTreeOpening
+        ];
+        Self::from_ssz_bytes(&BYTES).expect("blank signature bytes are valid")


is this some universial constant? how did you derive it?

KolbyML · 2026-02-17T07:05:03Z

crates/crypto/post_quantum/src/leansig/signature.rs

+/// Wrapper around leansig's signature type.
+/// Uses leansig's built-in SSZ encoding for interoperability with other clients.
+#[derive(Clone, Serialize, Deserialize)]
 pub struct Signature {
-    pub inner: FixedBytes<SIGNATURE_SIZE>,
+    pub inner: LeanSigSignature,
+}
+
+impl std::fmt::Debug for Signature {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("Signature")
+            .field("inner", &"<LeanSigSignature>")
+            .finish()
+    }
+}
+
+impl PartialEq for Signature {
+    fn eq(&self, other: &Self) -> bool {
+        // Compare by SSZ encoding since LeanSigSignature doesn't implement PartialEq
+        self.inner.as_ssz_bytes() == other.inner.as_ssz_bytes()
+    }
+}
+
+impl Eq for Signature {}
+
+impl Encode for Signature {
+    fn is_ssz_fixed_len() -> bool {
+        <LeanSigSignature as Encode>::is_ssz_fixed_len()
+    }
+
+    fn ssz_bytes_len(&self) -> usize {
+        self.inner.ssz_bytes_len()
+    }
+
+    fn ssz_append(&self, buf: &mut Vec<u8>) {
+        self.inner.ssz_append(buf)
+    }
 }

-impl From<&[u8]> for Signature {
-    fn from(value: &[u8]) -> Self {
-        Self {
-            inner: FixedBytes::from_slice(value),
-        }
+impl Decode for Signature {
+    fn is_ssz_fixed_len() -> bool {
+        <LeanSigSignature as Decode>::is_ssz_fixed_len()
+    }
+
+    fn from_ssz_bytes(bytes: &[u8]) -> Result<Self, DecodeError> {
+        Ok(Self {
+            inner: LeanSigSignature::from_ssz_bytes(bytes)?,
+        })


Suggested change

/// Wrapper around leansig's signature type.

/// Uses leansig's built-in SSZ encoding for interoperability with other clients.

#[derive(Clone, Serialize, Deserialize)]

pub struct Signature {

pub inner: FixedBytes<SIGNATURE_SIZE>,

pub inner: LeanSigSignature,

}

impl std::fmt::Debug for Signature {

fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {

f.debug_struct("Signature")

.field("inner", &"<LeanSigSignature>")

.finish()

}

}

impl PartialEq for Signature {

fn eq(&self, other: &Self) -> bool {

// Compare by SSZ encoding since LeanSigSignature doesn't implement PartialEq

self.inner.as_ssz_bytes() == other.inner.as_ssz_bytes()

}

}

impl Eq for Signature {}

impl Encode for Signature {

fn is_ssz_fixed_len() -> bool {

<LeanSigSignature as Encode>::is_ssz_fixed_len()

}

fn ssz_bytes_len(&self) -> usize {

self.inner.ssz_bytes_len()

}

fn ssz_append(&self, buf: &mut Vec<u8>) {

self.inner.ssz_append(buf)

}

}

impl From<&[u8]> for Signature {

fn from(value: &[u8]) -> Self {

Self {

inner: FixedBytes::from_slice(value),

}

impl Decode for Signature {

fn is_ssz_fixed_len() -> bool {

<LeanSigSignature as Decode>::is_ssz_fixed_len()

}

fn from_ssz_bytes(bytes: &[u8]) -> Result<Self, DecodeError> {

Ok(Self {

inner: LeanSigSignature::from_ssz_bytes(bytes)?,

})

/// Wrapper around leansig's signature type.

/// Uses leansig's built-in SSZ encoding for interoperability with other clients.

#[derive(Clone, Serialize, Deserialize, Encode, Decode, PartialEq, Eq)]

#[ssz(struct_behaviour = "transparent")]

pub struct Signature {

pub inner: LeanSigSignature,

}

impl std::fmt::Debug for Signature {

fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {

f.debug_struct("Signature")

.field("inner", &"<LeanSigSignature>")

.finish()

}

}

We don't need to implement Encode, Decode, PartialEq ourselves, do we even need to implement Debug?

KolbyML · 2026-02-17T07:13:28Z

crates/crypto/post_quantum/src/leansig/signature.rs

-    pub inner: FixedBytes<SIGNATURE_SIZE>,
+    pub inner: LeanSigSignature,


Yeah I did some regression testing and this looks fine 👍

KolbyML

looks good

unnawut commented Feb 11, 2026

View reviewed changes

unnawut added 16 commits February 11, 2026 22:30

add ssz test vectors

6ef6284

change Signature.inner to hold leansig's signature type directly inst…

5a5df5b

…ead of fixed bytes

fix ssz tests

230e188

fix linting

68efe49

fix linting

32a7f8c

fix linting

aca1153

simplify ssz_test.rs

48aa070

remove unnecessary conversions

eef7fee

fix comments

afc6d4f

remove TreeHash checks from SSZ tests

bba40bb

simplify ssz test setup

e37e5ff

deduplicate conversions with macros

5b35536

fix linting

539f02d

better passthrough_conversion naming

1d28707

refactor to custom_conversion macro

3e554fe

lint lint lint

9e95a6d

unnawut force-pushed the ssz-test-vectors branch from 1d195d9 to 9e95a6d Compare February 11, 2026 15:30

unnawut added 2 commits February 11, 2026 23:11

fix slow Signature::blank()

22ba388

blank signatures

e89b286

unnawut marked this pull request as ready for review February 11, 2026 16:45

unnawut requested review from Kayden-ML, KolbyML, Patchoulis and shariqnaiyer as code owners February 11, 2026 16:45

Add regression tests

fb80946

KolbyML reviewed Feb 17, 2026

View reviewed changes

Clean up PR

4013c95

KolbyML approved these changes Feb 17, 2026

View reviewed changes

KolbyML merged commit 4bc9d69 into ReamLabs:master Feb 17, 2026
8 of 10 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comments

test: add SSZ test vectors#1215

test: add SSZ test vectors#1215
KolbyML merged 20 commits intoReamLabs:masterfrom
unnawut:ssz-test-vectors

unnawut commented Feb 10, 2026 •

edited

Loading

Uh oh!

unnawut commented Feb 10, 2026

Uh oh!

unnawut Feb 11, 2026

Uh oh!

KolbyML Feb 12, 2026

Uh oh!

unnawut Feb 13, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML Feb 17, 2026

Uh oh!

KolbyML left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		pub inner: FixedBytes<SIGNATURE_SIZE>,
		pub inner: LeanSigSignature,

Comments

Conversation

unnawut commented Feb 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What was wrong?

To-Do

Uh oh!

unnawut commented Feb 10, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

KolbyML left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

unnawut commented Feb 10, 2026 •

edited

Loading