winnow/_topic/
performance.rs

1//! # Performance
2//!
3//! ## Runtime Performance
4//!
5//! See also the general Rust [Performance Book](https://nnethercote.github.io/perf-book/)
6//!
7//! Tips
8//! - Try `cargo add winnow -F simd`. For some it offers significant performance improvements
9//! - When enough cases of an [`alt`] have unique prefixes, prefer [`dispatch`]
10//! - When parsing text, try to parse as bytes (`u8`) rather than `char`s ([`BStr`] can make
11//!   debugging easier)
12//! - Find simplified subsets of the grammar to parse, falling back to the full grammar when it
13//!   doesn't work. For example, when parsing json strings, parse them without support for escapes,
14//!   falling back to escape support if it fails.
15//! - Watch for large return types. A surprising place these can show up is when chaining parsers
16//!   with a tuple.
17//!
18//! ## Build-time Performance
19//!
20//! Returning complex types as `impl Trait` can negatively impact build times. This can hit in
21//! surprising cases like:
22//! ```rust
23//! # use winnow::prelude::*;
24//! fn foo<I, O, E>() -> impl Parser<I, O, E>
25//! # where
26//! #    I: winnow::stream::Stream<Token=O>,
27//! #    I: winnow::stream::StreamIsPartial,
28//! #    E: winnow::error::ParserError<I>,
29//! {
30//!     // ...some chained combinators...
31//! # winnow::token::any
32//! }
33//! ```
34//!
35//! Instead, wrap the combinators in a closure to simplify the type:
36//! ```rust
37//! # use winnow::prelude::*;
38//! fn foo<I, O, E>() -> impl Parser<I, O, E>
39//! # where
40//! #    I: winnow::stream::Stream<Token=O>,
41//! #    I: winnow::stream::StreamIsPartial,
42//! #    E: winnow::error::ParserError<I>,
43//! {
44//!     move |input: &mut I| {
45//!         // ...some chained combinators...
46//! # winnow::token::any
47//!             .parse_next(input)
48//!     }
49//! }
50//! ```
51
52#![allow(unused_imports)]
53use crate::combinator::alt;
54use crate::combinator::dispatch;
55use crate::stream::BStr;