Use Ryū's small feature to cut ~9kb #55
Merged
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
The ryu crate is used by serde_json to format floats when JSON encoding and also when printing floats on JSON parse errors. The `small` feature reduces its binary size (it embeds a smaller table) at the cost of some performance. 9kb for Functions is a lot, dumping Float probably isn't the most frequent operation, and the different is not massive at any rate: see the table below for the difference between the 2 implementations. This change makes it so the "small" feature is used in Functions. ``` $ cargo bench running 16 tests test bench_ryu::bench_0_f32 ... bench: 1 ns/iter (+/- 0) test bench_ryu::bench_0_f64 ... bench: 1 ns/iter (+/- 0) test bench_ryu::bench_e_f32 ... bench: 10 ns/iter (+/- 0) test bench_ryu::bench_e_f64 ... bench: 15 ns/iter (+/- 7) test bench_ryu::bench_max_f32 ... bench: 10 ns/iter (+/- 1) test bench_ryu::bench_max_f64 ... bench: 22 ns/iter (+/- 1) test bench_ryu::bench_short_f32 ... bench: 13 ns/iter (+/- 1) test bench_ryu::bench_short_f64 ... bench: 26 ns/iter (+/- 1) test bench_std_fmt::bench_0_f32 ... bench: 11 ns/iter (+/- 0) test bench_std_fmt::bench_0_f64 ... bench: 11 ns/iter (+/- 0) test bench_std_fmt::bench_e_f32 ... bench: 52 ns/iter (+/- 2) test bench_std_fmt::bench_e_f64 ... bench: 58 ns/iter (+/- 0) test bench_std_fmt::bench_max_f32 ... bench: 50 ns/iter (+/- 0) test bench_std_fmt::bench_max_f64 ... bench: 74 ns/iter (+/- 4) test bench_std_fmt::bench_short_f32 ... bench: 42 ns/iter (+/- 2) test bench_std_fmt::bench_short_f64 ... bench: 41 ns/iter (+/- 0) $ cargo bench --features small running 16 tests test bench_ryu::bench_0_f32 ... bench: 1 ns/iter (+/- 0) test bench_ryu::bench_0_f64 ... bench: 1 ns/iter (+/- 0) test bench_ryu::bench_e_f32 ... bench: 20 ns/iter (+/- 1) test bench_ryu::bench_e_f64 ... bench: 20 ns/iter (+/- 1) test bench_ryu::bench_max_f32 ... bench: 20 ns/iter (+/- 0) test bench_ryu::bench_max_f64 ... bench: 20 ns/iter (+/- 2) test bench_ryu::bench_short_f32 ... bench: 22 ns/iter (+/- 0) test bench_ryu::bench_short_f64 ... bench: 31 ns/iter (+/- 2) test bench_std_fmt::bench_0_f32 ... bench: 11 ns/iter (+/- 0) test bench_std_fmt::bench_0_f64 ... bench: 11 ns/iter (+/- 0) test bench_std_fmt::bench_e_f32 ... bench: 50 ns/iter (+/- 1) test bench_std_fmt::bench_e_f64 ... bench: 55 ns/iter (+/- 3) test bench_std_fmt::bench_max_f32 ... bench: 48 ns/iter (+/- 1) test bench_std_fmt::bench_max_f64 ... bench: 74 ns/iter (+/- 0) test bench_std_fmt::bench_short_f32 ... bench: 41 ns/iter (+/- 0) test bench_std_fmt::bench_short_f64 ... bench: 40 ns/iter (+/- 0) ```
adampetro
approved these changes
Mar 7, 2024
| # Use the `small` feature of ryu (transitive dependency through serde_json) | ||
| # to shave off ~9kb of the Wasm binary size. | ||
| [dependencies.ryu] | ||
| version = "*" |
There was a problem hiding this comment.
Is the thinking that by using *, cargo will use the same version that serde_json is using to avoid having multiple versions of ryu?
There was a problem hiding this comment.
Yeah, that's the intent / hope. I think it works, but I'm no Cargo expert.
Merged
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.
The ryu crate is used by serde_json to format floats when JSON encoding and when printing floats on JSON parse errors. The
smallfeature reduces its binary size (it embeds a smaller table) at the cost of some performance. This change makes it so the "small" feature is used for Functions using this crate, shaving ~9kb.Given dumping floats probably isn't the most frequent operation, and that the difference in performance is minimal (see below), this seems reasonable.
The caveat is that it doesn't look users can opt-in using the default features (i.e.
[]) of ryu, or at least, I couldn't figure out how. If we think this is problematic, we could updatefunction-examples'Cargo.tomls to include the feature override.The benchmark