1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-11-24 15:19:26 -05:00
denoland-deno/serde_v8
Aaron O'Mullan fec1b2a5a4
refactor: new optimized op-layer using serde_v8 (#9843)
- Improves op performance.
- Handle op-metadata (errors, promise IDs) explicitly in the op-layer vs
  per op-encoding (aka: out-of-payload).
- Remove shared queue & custom "asyncHandlers", all async values are
  returned in batches via js_recv_cb.
- The op-layer should be thought of as simple function calls with little
  indirection or translation besides the conceptually straightforward
  serde_v8 bijections.
- Preserve concepts of json/bin/min as semantic groups of their
  inputs/outputs instead of their op-encoding strategy, preserving these
  groups will also facilitate partial transitions over to v8 Fast API for the
  "min" and "bin" groups
2021-03-31 10:37:38 -04:00
..
benches serde_v8: restore ser/de benches (#9939) 2021-03-30 22:24:49 -04:00
examples chore: add Deno copyright headers to all rust files (#9909) 2021-03-27 01:26:27 +01:00
src refactor: new optimized op-layer using serde_v8 (#9843) 2021-03-31 10:37:38 -04:00
tests refactor: new optimized op-layer using serde_v8 (#9843) 2021-03-31 10:37:38 -04:00
Cargo.toml serde_v8: restore ser/de benches (#9939) 2021-03-30 22:24:49 -04:00
README.md chore: publish serde_v8, bump version (#9898) 2021-03-26 08:51:38 +01:00

serde_v8

Author: Aaron O'Mullan aaron.omullan@gmail.com

Serde support for encoding/decoding (rusty_)v8 values.

Broadly serde_v8 aims to provide an expressive but ~maximally efficient encoding layer to biject rust & v8/js values. It's a core component of deno's op-layer and is used to encode/decode all non-buffer values.

Original issue: denoland/deno#9540

Quickstart

serde_v8 fits naturally into the serde ecosystem, so if you've already used serde or serde_json, serde_v8's API should be very familiar.

serde_v8 exposes two key-functions:

  • to_v8: maps rust->v8, similar to serde_json::to_string, ...
  • from_v8: maps v8->rust, similar to serde_json::from_str, ...

Best practices

Whilst serde_v8 is compatible with serde_json::Value it's important to keep in mind that serde_json::Value is essentially a loosely-typed value (think nested HashMaps), so when writing ops we recommend directly using rust structs/tuples or primitives, since mapping to serde_json::Value will add extra overhead and result in slower ops.

I also recommend avoiding unecessary "wrappers", if your op takes a single-keyed struct, consider unwrapping that as a plain value unless you plan to add fields in the near-future.

Instead of returning "nothing" via Ok(json!({})), change your return type to rust's unit type () and returning Ok(()), serde_v8 will efficiently encode that as a JS null.

Advanced features

If you need to mix rust & v8 values in structs/tuples, you can use the special serde_v8::Value type, which will passthrough the original v8 value untouched when encoding/decoding.

TODO

  • Experiment with KeyCache to optimize struct keys
  • Experiment with external v8 strings
  • Explore using json-stringifier.cc's fast-paths for arrays
  • Improve tests to test parity with serde_json (should be mostly interchangeable)
  • Consider a Payload type that's deserializable by itself (holds scope & value)
  • Ensure we return errors instead of panicking on .unwrap()s