rust/src/librustc/lint/levels.rs

534 lines
20 KiB
Rust
Raw Normal View History

rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
// Copyright 2017 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
use std::cmp;
suggestion applicabilities for libsyntax and librustc, run-rustfix tests Consider this a down payment on #50723. To recap, an `Applicability` enum was recently (#50204) added, to convey to Rustfix and other tools whether we think it's OK for them to blindly apply the suggestion, or whether to prompt a human for guidance (because the suggestion might contain placeholders that we can't infer, or because we think it has a sufficiently high probability of being wrong even though it's— presumably—right often enough to be worth emitting in the first place). When a suggestion is marked as `MaybeIncorrect`, we try to use comments to indicate precisely why (although there are a few places where we just say `// speculative` because the present author's subjective judgement balked at the idea that the suggestion has no false positives). The `run-rustfix` directive is opporunistically set on some relevant UI tests (and a couple tests that were in the `test/ui/suggestions` directory, even if the suggestions didn't originate in librustc or libsyntax). This is less trivial than it sounds, because a surprising number of test files aren't equipped to be tested as fixed even when they contain successfully fixable errors, because, e.g., there are more, not-directly-related errors after fixing. Some test files need an attribute or underscore to avoid unused warnings tripping up the "fixed code is still producing diagnostics" check despite the fixes being correct; this is an interesting contrast-to/inconsistency-with the behavior of UI tests (which secretly pass `-A unused`), a behavior which we probably ought to resolve one way or the other (filed issue #50926). A few suggestion labels are reworded (e.g., to avoid phrasing it as a question, which which is discouraged by the style guidelines listed in `.span_suggestion`'s doc-comment).
2018-05-19 21:52:24 +00:00
use errors::{Applicability, DiagnosticBuilder};
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
use hir::HirId;
use ich::StableHashingContext;
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
use lint::builtin;
use lint::context::CheckLintNameResult;
use lint::{self, Lint, LintId, Level, LintSource};
use rustc_data_structures::stable_hasher::{HashStable, ToStableHashKey,
StableHasher, StableHasherResult};
use session::Session;
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
use syntax::ast;
use syntax::attr;
2018-08-18 10:14:03 +00:00
use syntax::source_map::MultiSpan;
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
use syntax::symbol::Symbol;
use util::nodemap::FxHashMap;
pub struct LintLevelSets {
list: Vec<LintSet>,
lint_cap: Level,
}
enum LintSet {
CommandLine {
// -A,-W,-D flags, a `Symbol` for the flag itself and `Level` for which
// flag.
specs: FxHashMap<LintId, (Level, LintSource)>,
},
Node {
specs: FxHashMap<LintId, (Level, LintSource)>,
parent: u32,
},
}
impl LintLevelSets {
pub fn new(sess: &Session) -> LintLevelSets {
let mut me = LintLevelSets {
list: Vec::new(),
lint_cap: Level::Forbid,
};
me.process_command_line(sess);
return me
}
pub fn builder(sess: &Session) -> LintLevelsBuilder<'_> {
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
LintLevelsBuilder::new(sess, LintLevelSets::new(sess))
}
fn process_command_line(&mut self, sess: &Session) {
let store = sess.lint_store.borrow();
let mut specs = FxHashMap();
self.lint_cap = sess.opts.lint_cap.unwrap_or(Level::Forbid);
for &(ref lint_name, level) in &sess.opts.lint_opts {
store.check_lint_name_cmdline(sess, &lint_name, level);
// If the cap is less than this specified level, e.g. if we've got
// `--cap-lints allow` but we've also got `-D foo` then we ignore
// this specification as the lint cap will set it to allow anyway.
let level = cmp::min(level, self.lint_cap);
let lint_flag_val = Symbol::intern(lint_name);
let ids = match store.find_lints(&lint_name) {
Ok(ids) => ids,
Err(_) => continue, // errors handled in check_lint_name_cmdline above
};
for id in ids {
let src = LintSource::CommandLine(lint_flag_val);
specs.insert(id, (level, src));
}
}
self.list.push(LintSet::CommandLine {
specs: specs,
});
}
fn get_lint_level(&self,
lint: &'static Lint,
idx: u32,
aux: Option<&FxHashMap<LintId, (Level, LintSource)>>,
sess: &Session)
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
-> (Level, LintSource)
{
let (level, mut src) = self.get_lint_id_level(LintId::of(lint), idx, aux);
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
// If `level` is none then we actually assume the default level for this
// lint.
let mut level = level.unwrap_or(lint.default_level(sess));
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
// If we're about to issue a warning, check at the last minute for any
// directives against the warnings "lint". If, for example, there's an
// `allow(warnings)` in scope then we want to respect that instead.
if level == Level::Warn {
let (warnings_level, warnings_src) =
self.get_lint_id_level(LintId::of(lint::builtin::WARNINGS),
idx,
aux);
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
if let Some(configured_warning_level) = warnings_level {
if configured_warning_level != Level::Warn {
level = configured_warning_level;
src = warnings_src;
}
}
}
// Ensure that we never exceed the `--cap-lints` argument.
level = cmp::min(level, self.lint_cap);
if let Some(driver_level) = sess.driver_lint_caps.get(&LintId::of(lint)) {
// Ensure that we never exceed driver level.
level = cmp::min(*driver_level, level);
}
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
return (level, src)
}
fn get_lint_id_level(&self,
id: LintId,
mut idx: u32,
aux: Option<&FxHashMap<LintId, (Level, LintSource)>>)
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
-> (Option<Level>, LintSource)
{
if let Some(specs) = aux {
if let Some(&(level, src)) = specs.get(&id) {
return (Some(level), src)
}
}
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
loop {
match self.list[idx as usize] {
LintSet::CommandLine { ref specs } => {
if let Some(&(level, src)) = specs.get(&id) {
return (Some(level), src)
}
return (None, LintSource::Default)
}
LintSet::Node { ref specs, parent } => {
if let Some(&(level, src)) = specs.get(&id) {
return (Some(level), src)
}
idx = parent;
}
}
}
}
}
pub struct LintLevelsBuilder<'a> {
sess: &'a Session,
sets: LintLevelSets,
id_to_set: FxHashMap<HirId, u32>,
cur: u32,
warn_about_weird_lints: bool,
}
pub struct BuilderPush {
prev: u32,
}
impl<'a> LintLevelsBuilder<'a> {
pub fn new(sess: &'a Session, sets: LintLevelSets) -> LintLevelsBuilder<'a> {
assert_eq!(sets.list.len(), 1);
LintLevelsBuilder {
sess,
sets,
cur: 0,
id_to_set: FxHashMap(),
warn_about_weird_lints: sess.buffered_lints.borrow().is_some(),
}
}
/// Pushes a list of AST lint attributes onto this context.
///
/// This function will return a `BuilderPush` object which should be be
/// passed to `pop` when this scope for the attributes provided is exited.
///
/// This function will perform a number of tasks:
///
/// * It'll validate all lint-related attributes in `attrs`
/// * It'll mark all lint-related attriutes as used
/// * Lint levels will be updated based on the attributes provided
/// * Lint attributes are validated, e.g. a #[forbid] can't be switched to
/// #[allow]
///
/// Don't forget to call `pop`!
pub fn push(&mut self, attrs: &[ast::Attribute]) -> BuilderPush {
let mut specs = FxHashMap();
let store = self.sess.lint_store.borrow();
let sess = self.sess;
let bad_attr = |span| {
span_err!(sess, span, E0452,
"malformed lint attribute");
};
for attr in attrs {
let level = match Level::from_str(&attr.name().as_str()) {
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
None => continue,
Some(lvl) => lvl,
};
let meta = unwrap_or!(attr.meta(), continue);
attr::mark_used(attr);
let metas = if let Some(metas) = meta.meta_item_list() {
metas
} else {
bad_attr(meta.span);
continue
};
for li in metas {
let word = match li.word() {
Some(word) => word,
None => {
bad_attr(li.span);
continue
}
};
let tool_name = if let Some(lint_tool) = word.is_scoped() {
2018-10-02 08:59:41 +00:00
if !attr::is_known_lint_tool(lint_tool) {
2018-07-03 11:50:48 +00:00
span_err!(
sess,
lint_tool.span,
2018-07-03 11:50:48 +00:00
E0710,
"an unknown tool name found in scoped lint: `{}`",
2018-07-03 11:50:48 +00:00
word.ident
);
continue;
}
Some(lint_tool.as_str())
} else {
None
};
2018-04-17 13:33:39 +00:00
let name = word.name();
match store.check_lint_name(&name.as_str(), tool_name) {
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
CheckLintNameResult::Ok(ids) => {
let src = LintSource::Node(name, li.span);
for id in ids {
specs.insert(*id, (level, src));
}
}
CheckLintNameResult::Tool(result) => {
match result {
Ok(ids) => {
let complete_name = &format!("{}::{}", tool_name.unwrap(), name);
let src = LintSource::Node(Symbol::intern(complete_name), li.span);
for id in ids {
specs.insert(*id, (level, src));
}
}
Err((Some(ids), new_lint_name)) => {
let lint = builtin::RENAMED_AND_REMOVED_LINTS;
let (lvl, src) =
self.sets
.get_lint_level(lint, self.cur, Some(&specs), &sess);
2018-08-29 22:12:47 +00:00
let msg = format!(
"lint name `{}` is deprecated \
2018-08-30 08:28:18 +00:00
and may not have an effect in the future. \
2018-08-29 22:12:47 +00:00
Also `cfg_attr(cargo-clippy)` won't be necessary anymore",
name
);
let mut err = lint::struct_lint_level(
self.sess,
lint,
lvl,
src,
Some(li.span.into()),
2018-08-29 22:12:47 +00:00
&msg,
);
err.span_suggestion_with_applicability(
li.span,
"change it to",
new_lint_name.to_string(),
Applicability::MachineApplicable,
).emit();
2018-08-29 22:12:47 +00:00
let src = LintSource::Node(Symbol::intern(&new_lint_name), li.span);
for id in ids {
specs.insert(*id, (level, src));
}
}
Err((None, _)) => {
// If Tool(Err(None, _)) is returned, then either the lint does not
// exist in the tool or the code was not compiled with the tool and
// therefore the lint was never added to the `LintStore`. To detect
// this is the responsibility of the lint tool.
}
}
}
_ if !self.warn_about_weird_lints => {}
CheckLintNameResult::Warning(msg, renamed) => {
let lint = builtin::RENAMED_AND_REMOVED_LINTS;
let (level, src) = self.sets.get_lint_level(lint,
self.cur,
Some(&specs),
&sess);
let mut err = lint::struct_lint_level(self.sess,
lint,
level,
src,
Some(li.span.into()),
&msg);
if let Some(new_name) = renamed {
err.span_suggestion_with_applicability(
li.span,
"use the new name",
new_name,
Applicability::MachineApplicable
);
}
err.emit();
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
}
CheckLintNameResult::NoLint => {
let lint = builtin::UNKNOWN_LINTS;
let (level, src) = self.sets.get_lint_level(lint,
self.cur,
Some(&specs),
self.sess);
let msg = format!("unknown lint: `{}`", name);
let mut db = lint::struct_lint_level(self.sess,
lint,
level,
src,
Some(li.span.into()),
&msg);
if name.as_str().chars().any(|c| c.is_uppercase()) {
2018-04-17 13:33:39 +00:00
let name_lower = name.as_str().to_lowercase().to_string();
if let CheckLintNameResult::NoLint =
store.check_lint_name(&name_lower, tool_name) {
db.emit();
} else {
suggestion applicabilities for libsyntax and librustc, run-rustfix tests Consider this a down payment on #50723. To recap, an `Applicability` enum was recently (#50204) added, to convey to Rustfix and other tools whether we think it's OK for them to blindly apply the suggestion, or whether to prompt a human for guidance (because the suggestion might contain placeholders that we can't infer, or because we think it has a sufficiently high probability of being wrong even though it's— presumably—right often enough to be worth emitting in the first place). When a suggestion is marked as `MaybeIncorrect`, we try to use comments to indicate precisely why (although there are a few places where we just say `// speculative` because the present author's subjective judgement balked at the idea that the suggestion has no false positives). The `run-rustfix` directive is opporunistically set on some relevant UI tests (and a couple tests that were in the `test/ui/suggestions` directory, even if the suggestions didn't originate in librustc or libsyntax). This is less trivial than it sounds, because a surprising number of test files aren't equipped to be tested as fixed even when they contain successfully fixable errors, because, e.g., there are more, not-directly-related errors after fixing. Some test files need an attribute or underscore to avoid unused warnings tripping up the "fixed code is still producing diagnostics" check despite the fixes being correct; this is an interesting contrast-to/inconsistency-with the behavior of UI tests (which secretly pass `-A unused`), a behavior which we probably ought to resolve one way or the other (filed issue #50926). A few suggestion labels are reworded (e.g., to avoid phrasing it as a question, which which is discouraged by the style guidelines listed in `.span_suggestion`'s doc-comment).
2018-05-19 21:52:24 +00:00
db.span_suggestion_with_applicability(
li.span,
"lowercase the lint name",
suggestion applicabilities for libsyntax and librustc, run-rustfix tests Consider this a down payment on #50723. To recap, an `Applicability` enum was recently (#50204) added, to convey to Rustfix and other tools whether we think it's OK for them to blindly apply the suggestion, or whether to prompt a human for guidance (because the suggestion might contain placeholders that we can't infer, or because we think it has a sufficiently high probability of being wrong even though it's— presumably—right often enough to be worth emitting in the first place). When a suggestion is marked as `MaybeIncorrect`, we try to use comments to indicate precisely why (although there are a few places where we just say `// speculative` because the present author's subjective judgement balked at the idea that the suggestion has no false positives). The `run-rustfix` directive is opporunistically set on some relevant UI tests (and a couple tests that were in the `test/ui/suggestions` directory, even if the suggestions didn't originate in librustc or libsyntax). This is less trivial than it sounds, because a surprising number of test files aren't equipped to be tested as fixed even when they contain successfully fixable errors, because, e.g., there are more, not-directly-related errors after fixing. Some test files need an attribute or underscore to avoid unused warnings tripping up the "fixed code is still producing diagnostics" check despite the fixes being correct; this is an interesting contrast-to/inconsistency-with the behavior of UI tests (which secretly pass `-A unused`), a behavior which we probably ought to resolve one way or the other (filed issue #50926). A few suggestion labels are reworded (e.g., to avoid phrasing it as a question, which which is discouraged by the style guidelines listed in `.span_suggestion`'s doc-comment).
2018-05-19 21:52:24 +00:00
name_lower,
Applicability::MachineApplicable
).emit();
}
} else {
db.emit();
}
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
}
}
}
}
for (id, &(level, ref src)) in specs.iter() {
if level == Level::Forbid {
continue
}
let forbid_src = match self.sets.get_lint_id_level(*id, self.cur, None) {
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
(Some(Level::Forbid), src) => src,
_ => continue,
};
let forbidden_lint_name = match forbid_src {
LintSource::Default => id.to_string(),
LintSource::Node(name, _) => name.to_string(),
LintSource::CommandLine(name) => name.to_string(),
};
let (lint_attr_name, lint_attr_span) = match *src {
LintSource::Node(name, span) => (name, span),
_ => continue,
};
let mut diag_builder = struct_span_err!(self.sess,
lint_attr_span,
E0453,
"{}({}) overruled by outer forbid({})",
level.as_str(),
lint_attr_name,
forbidden_lint_name);
diag_builder.span_label(lint_attr_span, "overruled by previous forbid");
match forbid_src {
LintSource::Default => &mut diag_builder,
LintSource::Node(_, forbid_source_span) => {
diag_builder.span_label(forbid_source_span,
"`forbid` level set here")
},
LintSource::CommandLine(_) => {
diag_builder.note("`forbid` lint level was set on command line")
}
}.emit();
// don't set a separate error for every lint in the group
break
}
let prev = self.cur;
if specs.len() > 0 {
self.cur = self.sets.list.len() as u32;
self.sets.list.push(LintSet::Node {
specs: specs,
parent: prev,
});
}
BuilderPush {
prev: prev,
}
}
/// Called after `push` when the scope of a set of attributes are exited.
pub fn pop(&mut self, push: BuilderPush) {
self.cur = push.prev;
}
/// Used to emit a lint-related diagnostic based on the current state of
/// this lint context.
pub fn struct_lint(&self,
lint: &'static Lint,
span: Option<MultiSpan>,
msg: &str)
-> DiagnosticBuilder<'a>
{
let (level, src) = self.sets.get_lint_level(lint, self.cur, None, self.sess);
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
lint::struct_lint_level(self.sess, lint, level, src, span, msg)
}
/// Registers the ID provided with the current set of lints stored in
/// this context.
pub fn register_id(&mut self, id: HirId) {
self.id_to_set.insert(id, self.cur);
}
pub fn build(self) -> LintLevelSets {
self.sets
}
pub fn build_map(self) -> LintLevelMap {
LintLevelMap {
sets: self.sets,
id_to_set: self.id_to_set,
}
}
}
pub struct LintLevelMap {
sets: LintLevelSets,
id_to_set: FxHashMap<HirId, u32>,
}
impl LintLevelMap {
/// If the `id` was previously registered with `register_id` when building
/// this `LintLevelMap` this returns the corresponding lint level and source
/// of the lint level for the lint provided.
///
/// If the `id` was not previously registered, returns `None`. If `None` is
/// returned then the parent of `id` should be acquired and this function
/// should be called again.
pub fn level_and_source(&self, lint: &'static Lint, id: HirId, session: &Session)
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
-> Option<(Level, LintSource)>
{
self.id_to_set.get(&id).map(|idx| {
self.sets.get_lint_level(lint, *idx, None, session)
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
})
}
2017-09-13 19:33:07 +00:00
/// Returns if this `id` has lint level information.
pub fn lint_level_set(&self, id: HirId) -> Option<u32> {
self.id_to_set.get(&id).cloned()
}
rustc: Rearchitect lints to be emitted more eagerly In preparation for incremental compilation this commit refactors the lint handling infrastructure in the compiler to be more "eager" and overall more incremental-friendly. Many passes of the compiler can emit lints at various points but before this commit all lints were buffered in a table to be emitted at the very end of compilation. This commit changes these lints to be emitted immediately during compilation using pre-calculated lint level-related data structures. Linting today is split into two phases, one set of "early" lints run on the `syntax::ast` and a "late" set of lints run on the HIR. This commit moves the "early" lints to running as late as possible in compilation, just before HIR lowering. This notably means that we're catching resolve-related lints just before HIR lowering. The early linting remains a pass very similar to how it was before, maintaining context of the current lint level as it walks the tree. Post-HIR, however, linting is structured as a method on the `TyCtxt` which transitively executes a query to calculate lint levels. Each request to lint on a `TyCtxt` will query the entire crate's 'lint level data structure' and then go from there about whether the lint should be emitted or not. The query depends on the entire HIR crate but should be very quick to calculate (just a quick walk of the HIR) and the red-green system should notice that the lint level data structure rarely changes, and should hopefully preserve incrementality. Overall this resulted in a pretty big change to the test suite now that lints are emitted much earlier in compilation (on-demand vs only at the end). This in turn necessitated the addition of many `#![allow(warnings)]` directives throughout the compile-fail test suite and a number of updates to the UI test suite.
2017-07-27 04:51:09 +00:00
}
impl<'a> HashStable<StableHashingContext<'a>> for LintLevelMap {
#[inline]
fn hash_stable<W: StableHasherResult>(&self,
hcx: &mut StableHashingContext<'a>,
hasher: &mut StableHasher<W>) {
let LintLevelMap {
ref sets,
ref id_to_set,
} = *self;
id_to_set.hash_stable(hcx, hasher);
let LintLevelSets {
ref list,
lint_cap,
} = *sets;
lint_cap.hash_stable(hcx, hasher);
hcx.while_hashing_spans(true, |hcx| {
list.len().hash_stable(hcx, hasher);
// We are working under the assumption here that the list of
// lint-sets is built in a deterministic order.
for lint_set in list {
::std::mem::discriminant(lint_set).hash_stable(hcx, hasher);
match *lint_set {
LintSet::CommandLine { ref specs } => {
specs.hash_stable(hcx, hasher);
}
LintSet::Node { ref specs, parent } => {
specs.hash_stable(hcx, hasher);
parent.hash_stable(hcx, hasher);
}
}
}
})
}
}
impl<HCX> HashStable<HCX> for LintId {
#[inline]
fn hash_stable<W: StableHasherResult>(&self,
hcx: &mut HCX,
hasher: &mut StableHasher<W>) {
self.lint_name_raw().hash_stable(hcx, hasher);
}
}
impl<HCX> ToStableHashKey<HCX> for LintId {
type KeyType = &'static str;
#[inline]
fn to_stable_hash_key(&self, _: &HCX) -> &'static str {
self.lint_name_raw()
}
}