Implements reading/writing FlexInt, FlexUInt
#690
Conversation
| @@ -11,7 +11,6 @@ jobs: | |||
| if: github.event_name == 'push' || github.event.pull_request.head.repo.full_name != 'amazon-ion/ion-rust' | |||
| strategy: | |||
| matrix: | |||
| # See https://github.com/amazon-ion/ion-rust/issues/353 | |||
| - name: Remove MSys64 MingW64 Binaries | ||
| if: runner.os == 'Windows' | ||
| # remove this because there is a bad libclang.dll that confuses bindgen | ||
| run: Remove-Item -LiteralPath "C:\msys64\mingw64\bin" -Force -Recurse |
There was a problem hiding this comment.
🗺️ This was necessary for ion-c-sys, which was removed.
| use criterion::{black_box, criterion_group, criterion_main, Criterion}; | ||
| use rand::prelude::StdRng; | ||
| use rand::{distributions::Uniform, Rng, SeedableRng}; | ||
| use std::io; |
There was a problem hiding this comment.
🗺️ This file is a suite of benchmarks comparing the time needed to read or write 10k random integers using VarUInt, VarInt, FlexUInt, or FlexInt encoding.
| @@ -9,7 +9,7 @@ use crate::IonType; | |||
| /// [Typed Value Formats](https://amazon-ion.github.io/ion-docs/docs/binary.html#typed-value-formats) | |||
| /// section of the binary Ion spec. | |||
| #[derive(Copy, Clone, Debug, PartialEq)] | |||
| pub(crate) struct TypeDescriptor { | |||
| pub struct TypeDescriptor { | |||
There was a problem hiding this comment.
🗺️ This PR makes a couple of types pub while leaving their module pub(crate) or even private. A later change in this diff re-exports these pub types if the experimental-lazy-reader feature is enabled, allowing the benchmarks to use some of these implementation details.
| lazy::encoder::binary::v1_1::flex_int::FlexInt, | ||
| lazy::encoder::binary::v1_1::flex_uint::FlexUInt, | ||
| }; | ||
|
|
There was a problem hiding this comment.
🗺️ Feature-gated re-exports of encoding-primitive-related types.
src/lazy/binary/immutable_buffer.rs
Outdated
| // TODO: Should this be a const cache of num_encoded_bits -> mask? | ||
| let mask = 1u64 | ||
| .checked_shl(num_encoded_bits as u32) | ||
| .map(|v| v - 1) | ||
| .unwrap_or(u64::MAX); |
There was a problem hiding this comment.
🗺️ I think making a const cache for this would get us a modest speedup, but the performance is already much better than VarUInt.
Implements reading and writing of Ion 1.1 encoding primitives (
FlexInt,FlexUInt).Also adds benchmarks comparing the time needed to read/write 10k integers with that of the corresponding Ion 1.0 encoding primitives (
VarInt,VarUInt). Benchmark results on my 2019 16" Intel MBP:By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.