perf: use a faster hasher in compiler and LSP

By applying the same hasher as in Rust's compiler, performance can be
improved wherever a hash map or set is used.

This currently has the most visible impact on the compiler's validation
step as it internally makes heavy use of hash maps.

Cargo.toml

@@ -34,6 +34,7 @@ miette = "5.10.0"
 mimalloc = "0.1.39"
 proc-macro2 = { version = "1.0.76", default-features = false }
 quote = { version = "1.0.35", default-features = false }
+rustc-hash = "1.1.0"
 serde = { version = "1.0.195", features = ["derive"] }
 serde_json = "1.0.111"
 syn = "2.0.48"

crates/mabo-compiler/Cargo.toml

@@ -14,6 +14,7 @@ anstream.workspace = true
 anstyle.workspace = true
 mabo-parser = { path = "../mabo-parser" }
 miette.workspace = true
+rustc-hash.workspace = true
 schemars = { version = "0.8.16", optional = true }
 serde = { workspace = true, optional = true }
 serde_json = { workspace = true, optional = true }

crates/mabo-compiler/src/resolve/mod.rs

@@ -1,12 +1,11 @@
 //! Ensure all referenced types within a schema itself, aswell as between schemas exist and are
 //! correct.
 
-use std::collections::HashMap;
-
 use mabo_parser::{
     DataType, Definition, ExternalType, Fields, Generics, Import, Name, Schema, Spanned, Type,
 };
 use miette::NamedSource;
+use rustc_hash::FxHashMap;
 
 pub use self::error::{
     Error, GenericsCount, InvalidKind, MissingDefinition, MissingImport, MissingModule,
@@ -83,7 +82,7 @@ pub(crate) struct Module<'a> {
     /// List of types that are declared in this module.
     types: Vec<Declaration<'a>>,
     /// Direct submodules located in this module.
-    modules: HashMap<&'a str, Module<'a>>,
+    modules: FxHashMap<&'a str, Module<'a>>,
     /// List of original definitions.
     definitions: &'a [Definition<'a>],
 }
@@ -411,7 +410,7 @@ fn visit_module_tree<'a>(
         path,
         imports: Vec::new(),
         types: Vec::new(),
-        modules: HashMap::new(),
+        modules: FxHashMap::default(),
         definitions: defs,
     };
 

crates/mabo-compiler/src/validate/generics.rs

@@ -1,7 +1,8 @@
-use std::{collections::HashMap, ops::Range};
+use std::{hash::BuildHasherDefault, ops::Range};
 
 use mabo_parser::{DataType, Enum, ExternalType, Fields, Generics, Span, Spanned, Struct, Type};
 use miette::Diagnostic;
+use rustc_hash::FxHashMap;
 use thiserror::Error;
 
 /// Generic type parameters are considered invalid.
@@ -53,7 +54,7 @@ pub fn validate_struct_generics(value: &Struct<'_>) -> Result<(), InvalidGeneric
         .0
         .iter()
         .map(|gen| (gen.get(), gen.span()))
-        .collect::<HashMap<_, _>>();
+        .collect::<FxHashMap<_, _>>();
 
     validate_field_generics(&value.fields, &mut unvisited);
 
@@ -75,7 +76,7 @@ pub fn validate_enum_generics(value: &Enum<'_>) -> Result<(), InvalidGenericType
         .0
         .iter()
         .map(|gen| (gen.get(), gen.span()))
-        .collect::<HashMap<_, _>>();
+        .collect::<FxHashMap<_, _>>();
 
     for variant in &value.variants {
         validate_field_generics(&variant.fields, &mut unvisited);
@@ -92,7 +93,8 @@ pub fn validate_enum_generics(value: &Enum<'_>) -> Result<(), InvalidGenericType
 
 /// Ensure all generic type arguments are unique within a struct or enum.
 fn validate_duplicate_generics(value: &Generics<'_>) -> Result<(), DuplicateGenericName> {
-    let mut visited = HashMap::with_capacity(value.0.len());
+    let mut visited =
+        FxHashMap::with_capacity_and_hasher(value.0.len(), BuildHasherDefault::default());
     value
         .0
         .iter()
@@ -110,7 +112,7 @@ fn validate_duplicate_generics(value: &Generics<'_>) -> Result<(), DuplicateGene
 
 /// Iterate over all the fields and mark any generic types as used when disvored as type for a
 /// field.
-fn validate_field_generics(value: &Fields<'_>, unvisited: &mut HashMap<&str, Span>) {
+fn validate_field_generics(value: &Fields<'_>, unvisited: &mut FxHashMap<&str, Span>) {
     match &value {
         Fields::Named(named) => {
             for field in named {

crates/mabo-compiler/src/validate/ids.rs

@@ -1,7 +1,8 @@
-use std::{collections::HashMap, ops::Range};
+use std::{hash::BuildHasherDefault, ops::Range};
 
 use mabo_parser::{Enum, Fields, Id, Spanned, Struct};
 use miette::Diagnostic;
+use rustc_hash::FxHashMap;
 use thiserror::Error;
 
 use crate::IdGenerator;
@@ -97,7 +98,8 @@ pub(crate) fn validate_struct_ids(value: &Struct<'_>) -> Result<(), DuplicateFie
 /// Ensure all IDs inside an enum are unique, which means all variants have a unique ID, plus all
 /// potential fields in a variant are unique (within that variant).
 pub(crate) fn validate_enum_ids(value: &Enum<'_>) -> Result<(), DuplicateId> {
-    let mut visited = HashMap::with_capacity(value.variants.len());
+    let mut visited =
+        FxHashMap::with_capacity_and_hasher(value.variants.len(), BuildHasherDefault::default());
     let mut id_gen = IdGenerator::new();
 
     value
@@ -131,7 +133,8 @@ pub(crate) fn validate_enum_ids(value: &Enum<'_>) -> Result<(), DuplicateId> {
 fn validate_field_ids(value: &Fields<'_>) -> Result<(), DuplicateFieldId> {
     match value {
         Fields::Named(named) => {
-            let mut visited = HashMap::with_capacity(named.len());
+            let mut visited =
+                FxHashMap::with_capacity_and_hasher(named.len(), BuildHasherDefault::default());
             let mut id_gen = IdGenerator::new();
 
             named
@@ -152,7 +155,8 @@ fn validate_field_ids(value: &Fields<'_>) -> Result<(), DuplicateFieldId> {
                 .map_or(Ok(()), Err)?;
         }
         Fields::Unnamed(unnamed) => {
-            let mut visited = HashMap::with_capacity(unnamed.len());
+            let mut visited =
+                FxHashMap::with_capacity_and_hasher(unnamed.len(), BuildHasherDefault::default());
             let mut id_gen = IdGenerator::new();
 
             unnamed

crates/mabo-compiler/src/validate/names.rs

@@ -1,7 +1,8 @@
-use std::{collections::HashMap, ops::Range};
+use std::{hash::BuildHasherDefault, ops::Range};
 
 use mabo_parser::{Definition, Enum, Fields, Import, Spanned, Struct};
 use miette::Diagnostic;
+use rustc_hash::FxHashMap;
 use thiserror::Error;
 
 /// Duplicate name was encountered for two elements in the same scope.
@@ -76,7 +77,8 @@ pub(crate) fn validate_struct_names(value: &Struct<'_>) -> Result<(), DuplicateF
 /// Ensure all names inside an enum are unique, which means all variants have a unique name, plus
 /// all potential fields in a variant are unique (within that variant).
 pub(crate) fn validate_enum_names(value: &Enum<'_>) -> Result<(), DuplicateName> {
-    let mut visited = HashMap::with_capacity(value.variants.len());
+    let mut visited =
+        FxHashMap::with_capacity_and_hasher(value.variants.len(), BuildHasherDefault::default());
     value
         .variants
         .iter()
@@ -104,7 +106,8 @@ pub(crate) fn validate_enum_names(value: &Enum<'_>) -> Result<(), DuplicateName>
 fn validate_field_names(value: &Fields<'_>) -> Result<(), DuplicateFieldName> {
     match value {
         Fields::Named(named) => {
-            let mut visited = HashMap::with_capacity(named.len());
+            let mut visited =
+                FxHashMap::with_capacity_and_hasher(named.len(), BuildHasherDefault::default());
             named
                 .iter()
                 .find_map(|field| {
@@ -125,7 +128,8 @@ fn validate_field_names(value: &Fields<'_>) -> Result<(), DuplicateFieldName> {
 }
 
 pub(crate) fn validate_names_in_module(value: &[Definition<'_>]) -> Result<(), DuplicateName> {
-    let mut visited = HashMap::with_capacity(value.len());
+    let mut visited =
+        FxHashMap::with_capacity_and_hasher(value.len(), BuildHasherDefault::default());
     value
         .iter()
         .find_map(|definition| {

crates/mabo-lsp/Cargo.toml

@@ -24,6 +24,7 @@ mabo-project = { path = "../mabo-project" }
 ouroboros = "0.18.2"
 parking_lot = "0.12.1"
 ropey = "1.6.1"
+rustc-hash.workspace = true
 serde_json.workspace = true
 serde.workspace = true
 time = { version = "0.3.31", features = ["formatting", "local-offset", "macros"] }

crates/mabo-lsp/src/main.rs

@@ -2,7 +2,7 @@
 
 #![warn(clippy::expect_used, clippy::unwrap_used)]
 
-use std::{collections::HashMap, net::Ipv4Addr, time::Instant};
+use std::{net::Ipv4Addr, time::Instant};
 
 use anyhow::{bail, Result};
 use log::{as_debug, debug, error, info, warn};
@@ -18,6 +18,7 @@ use lsp_types::{
     },
     DocumentSymbol, InitializeParams, SemanticTokens,
 };
+use rustc_hash::FxHashMap;
 
 use self::{cli::Cli, client::Client};
 use crate::state::GlobalState;
@@ -45,7 +46,7 @@ fn main() -> Result<()> {
 
     let mut state = GlobalState {
         client: Client::new(&connection),
-        files: HashMap::default(),
+        files: FxHashMap::default(),
         settings: config::Global::default(),
     };
 

crates/mabo-lsp/src/state.rs

@@ -1,19 +1,18 @@
-use std::collections::HashMap;
-
 use anyhow::{ensure, Context, Result};
 use line_index::LineIndex;
 use log::{as_debug, debug};
 use lsp_types::{ConfigurationItem, Diagnostic, Url};
 use mabo_parser::Schema;
 use ouroboros::self_referencing;
 use ropey::Rope;
+use rustc_hash::FxHashMap;
 
 use crate::{client::Client, config};
 
 #[allow(clippy::module_name_repetitions)]
 pub struct GlobalState<'a> {
     pub client: Client<'a>,
-    pub files: HashMap<Url, File>,
+    pub files: FxHashMap<Url, File>,
     pub settings: config::Global,
 }