[serde_multipart] new crate: multipart codec for serde #831
+1,186
−0
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
meta-cla
bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
This was referenced Aug 18, 2025
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303723935 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303723935 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. Differential Revision: [D80108937](https://our.internmc.facebook.com/intern/diff/D80108937/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D80108937 |
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
shayne-fletcher
pushed a commit
to shayne-fletcher/monarch-1
that referenced
this pull request
Aug 18, 2025
Summary: Serde codec for multipart messages. //! Using [`serialize`] / [`deserialize`], fields typed [`Part`] are extracted //! from the main payload and appended to a list of `parts`. Each part is backed by //! [`bytes::Bytes`] for cheap, zero-copy sharing. //! //! On decode, the body and its parts are reassembled into the original value //! without copying. //! //! The on-the-wire form is a [`Message`] (body + parts). Your transport sends //! and receives [`Message`]s; the codec reconstructs the value, enabling //! efficient network I/O without compacting data into a single buffer. //! //! Implementation note: this crate uses Rust's min_specialization feature to enable //! the use of [`Part`]s with any Serde serializer or deserializer. This feature //! is fairly restrictive, and thus the API offered by [`serialize`] / [`deserialize`] //! is not customizable. If customization is needed, you need to add specialization //! implementations for these codecs. See [`part::PartSerializer`] and [`part::PartDeserializer`] //! for details. This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer. The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type). The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed. ghstack-source-id: 303813996 exported-using-ghexport Reviewed By: vidhyav, shayne-fletcher Differential Revision: D80108937
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):
Message#852Serde codec for multipart messages.
//! Using [
serialize] / [deserialize], fields typed [Part] are extracted//! from the main payload and appended to a list of
parts. Each part is backed by//! [
bytes::Bytes] for cheap, zero-copy sharing.//!
//! On decode, the body and its parts are reassembled into the original value
//! without copying.
//!
//! The on-the-wire form is a [
Message] (body + parts). Your transport sends//! and receives [
Message]s; the codec reconstructs the value, enabling//! efficient network I/O without compacting data into a single buffer.
//!
//! Implementation note: this crate uses Rust's min_specialization feature to enable
//! the use of [
Part]s with any Serde serializer or deserializer. This feature//! is fairly restrictive, and thus the API offered by [
serialize] / [deserialize]//! is not customizable. If customization is needed, you need to add specialization
//! implementations for these codecs. See [
part::PartSerializer] and [part::PartDeserializer]//! for details.
This will be the foundation for doing zero-copy I/O through our networking stack. With serde_multipart, we can avoid copying byte buffers, and delegate their efficient transmission to the network layer.
The representation -- a multipart message comprising a body and zero or more contiguous byte buffers -- lends itself to other optimizations as well. For example, we could perform transparent RDMA for large parts, and even control the allocation of the received bytes (through the Bytes data type).
The implementation relies on Rust's "min_specialization" feature, which provides a limited form of specialization. This is what allows us to override the serializer for "serde_multipart::Part" when using our multipart-aware encoder. This was quite finnicky to get to work right (the feature accepts only a limited subset of known-to-be-sound specializations), but appears to work well, at the expense of being slightly less generic than we might otherwise have managed.
Differential Revision: D80108937