Add support for bincode v2.0 #821

francisbr · 2025-03-27T18:34:45Z

Following the latest major release of Bincode, the crate now relies on its own traits (Encode/Decode) it switched away from serde's De/Serialize. This PR is to add a feature flag here to optionally implement these traits on the Uuid structs.

The 2.0 bincode release: https://github.com/bincode-org/bincode/releases/tag/v2.0.0

KodrAus

Thanks for working on this @francisbr! As much as I'd love to fix the redundant length bytes, we need to maintain compatibility with the old serde and bincode 1.x format. If the new bincode derive macros have a similar concept to serde's #[with] for customizing serialization then we could add support for the more compact format that way.

KodrAus · 2025-03-27T21:29:32Z

Cargo.toml

 readme = "README.md"
 repository = "https://github.com/uuid-rs/uuid"
-version = "1.16.0" # remember to update html_root_url in lib.rs
+version = "1.16.1" # remember to update html_root_url in lib.rs


I have a script that updates this so we can leave it as 1.16.0 here. I’ll end up releasing it as 1.17.0

Got it, reverted!

KodrAus · 2025-03-28T21:47:14Z

src/external/bincode_support.rs

+
+impl Encode for Uuid {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        Encode::encode(self.as_bytes(), encoder)


Unfortunately, in order to maintain the original serde 1.x + bincode 1.x encoding, we'll need to include the redundant length bytes here too. Here's an example original format UUID we can use to test equivalence:

[16, 0, 0, 0, 0, 0, 0, 0, 232, 18, 203, 187, 159, 207, 72, 46, 151, 197, 95, 187, 59, 106, 126, 5]

I am not 100% sure I understand how that was working. The example you gave me is 24 bytes (more than the 16 or 17 with the length byte).

I added a check in the encode/decode function to check if the config passed is legacy. If so I simply prepend the '16' byte before the UUID. Is that what is needed?
You can see the test_legacy() test I added, when passing config::legacy() to the functions, 17 bytes are encoded/decoded.

Sorry if that is not what is needed and I missunderstood.

Yes I thought our original encoding gave us a 1 byte length, but running bincode::serialize on 1.x gave me what appeared to be a usize length prefix

src/external/bincode_support.rs

KodrAus

Thanks @francisbr! This looks good to me.

KodrAus · 2025-04-11T23:19:37Z

Looks like we've got a few build failures here:

failures:

---- external::bincode_support::bincode_tests::test_decode_failure stdout ----

thread 'external::bincode_support::bincode_tests::test_decode_failure' panicked at src/external/bincode_support.rs:221:18:
Unexpected error
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

---- external::bincode_support::bincode_tests::test_decode_non_nil_uuid stdout ----

thread 'external::bincode_support::bincode_tests::test_decode_non_nil_uuid' panicked at src/external/bincode_support.rs:239:9:
assertion `left == right` failed
  left: 16
 right: 17

loarca · 2025-05-01T20:21:49Z

src/external/bincode_support.rs

+impl Encode for NonNilUuid {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        self.get().encode(encoder)
+    }
+}
+
+impl Encode for Hyphenated {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        self.as_uuid().encode(encoder)
+    }
+}
+
+impl Encode for Simple {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        self.as_uuid().encode(encoder)
+    }
+}
+
+impl Encode for Urn {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        self.as_uuid().encode(encoder)
+    }
+}
+
+impl Encode for Braced {
+    fn encode<E: Encoder>(&self, encoder: &mut E) -> Result<(), EncodeError> {
+        self.as_uuid().encode(encoder)
+    }
+}


Wouldn't these differ from the serde impl?

serde would output strings for these, therefore I believe bincode v1 encodes a string rather than the bytes.

So, if we care that the Uuid serde impl is the same as the bincode one, these should encode the string representations

Ah well spotted! We definitely should maintain compatibility on these adapter types too

loarca · 2025-05-01T20:30:36Z

src/external/bincode_support.rs

+        if encoder.config().int_encoding() == IntEncoding::Fixed {
+            usize::encode(&16, encoder)?;
+        } else {
+            u8::encode(&16, encoder)?;
+        }
+
+        <&[u8; 16]>::encode(&self.as_bytes(), encoder)?;
+
+        Ok(())


I believe this is sufficient and delegates the int encoding implementation to bincode itself

Suggested change

if encoder.config().int_encoding() == IntEncoding::Fixed {

usize::encode(&16, encoder)?;

} else {

u8::encode(&16, encoder)?;

}

<&[u8; 16]>::encode(&self.as_bytes(), encoder)?;

Ok(())

<[u8]>::encode(self.as_bytes().as_slice(), encoder)

Leaning on the impl provided by bincode itself seems like a good idea. That way we should naturally respect any new options it introduces that might affect encoding that end users may want.

loarca · 2025-05-01T21:22:44Z

src/external/bincode_support.rs

+}
+
+#[cfg(test)]
+mod bincode_tests {


Besides testing the roundtrip of encode->decode, we should probably also test the output binary representation of encode. Just to ensure we know what to expect from a Uuid and its variants (Simple, etc)

Found some incompatibilities that we need to fix up

francisbr force-pushed the bincode branch 2 times, most recently from e619423 to 2d40b0f Compare March 27, 2025 18:48

francis-spliced mentioned this pull request Mar 27, 2025

Support bincode 2 #820

Closed

Add support for bincode v2

ce92df8

francisbr force-pushed the bincode branch 2 times, most recently from 168e645 to de93dbe Compare March 28, 2025 01:15

Cleanup

621c89a

francisbr force-pushed the bincode branch from de93dbe to 621c89a Compare March 28, 2025 01:20

KodrAus reviewed Mar 28, 2025

View reviewed changes

francisbr added 2 commits March 29, 2025 01:31

Add support for legacy bincode 1

61124e3

Add legacy test with container & derive macro

33a4ea2

KodrAus reviewed Mar 29, 2025

View reviewed changes

src/external/bincode_support.rs Outdated Show resolved Hide resolved

KodrAus reviewed Mar 29, 2025

View reviewed changes

src/external/bincode_support.rs Outdated Show resolved Hide resolved

Use bincodev1 for test

9b7eddd

KodrAus previously approved these changes Apr 9, 2025

View reviewed changes

loarca reviewed May 1, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add support for bincode v2.0 #821

Add support for bincode v2.0 #821

Uh oh!

francisbr commented Mar 27, 2025 •

edited

Loading

Uh oh!

KodrAus left a comment

Uh oh!

KodrAus Mar 27, 2025

Uh oh!

francisbr Mar 29, 2025

Uh oh!

KodrAus Mar 28, 2025

Uh oh!

francisbr Mar 29, 2025

Uh oh!

KodrAus Mar 29, 2025

Uh oh!

Uh oh!

Uh oh!

KodrAus left a comment

Uh oh!

KodrAus commented Apr 11, 2025

Uh oh!

loarca May 1, 2025 •

edited

Loading

Uh oh!

KodrAus May 13, 2025

Uh oh!

loarca May 1, 2025 •

edited

Loading

Uh oh!

KodrAus May 13, 2025

Uh oh!

loarca May 1, 2025

Uh oh!

Uh oh!

Add support for bincode v2.0 #821

Are you sure you want to change the base?

Add support for bincode v2.0 #821

Uh oh!

Conversation

francisbr commented Mar 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

KodrAus left a comment

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!

Uh oh!

Uh oh!

KodrAus left a comment

Choose a reason for hiding this comment

Uh oh!

KodrAus commented Apr 11, 2025

Uh oh!

loarca May 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

loarca May 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

Uh oh!

francisbr commented Mar 27, 2025 •

edited

Loading

loarca May 1, 2025 •

edited

Loading

loarca May 1, 2025 •

edited

Loading