There are now 3 queries:
- inherent_impls(def-id): for a given type, get a `Rc<Vec<DefId>>` with
all its inherent impls. This internally uses `crate_inherent_impls`,
doing some hacks to keep the current deps (which, btw, are not clearly
correct).
- crate_inherent_impls(crate): gathers up a map from types
to `Rc<Vec<DefId>>`, touching the entire krate, possibly generating
errors.
- crate_inherent_impls_overlap_check(crate): performs overlap checks
between the inherent impls for a given type, generating errors.
This "on-demand" task both checks for errors and computes the custom
unsized kind, if any. This task is only defined on impls of
`CoerceUnsized`; invoking it on any other kind of impl results in a bug.
This is just to avoid having an `Option`, could easily be changed.
Remove unused adt-def insertion by constructor DefIndex
It looks to me like ADT definitions weren't being looked up by constructor id, and a test run supports my theory.
In any case, I'm not sure it would have worked in its current configuration. If I understand correctly, the `adt_def` map entry from constructor id -> adt def would only be present after a successful call to `queries::adt_def::get` with the proper ADT `DefIndex`. Trying to look up an adt_def by the constructor index prior to a successful lookup by ADT index would fail since `item.kind` would be `EntryKind::Fn` (for the constructor function) and so would trigger the `bug!`.
r? @nikomatsakis
Correctly get source for metatdata-only crate type
Closes#40535
However, I'm not sure how to approach writing a regression test since I'm still working on a reduced test case from the code that caused the ICE in the first place. It's not enough to have an unknown `extern crate` in a metadata crate, it depends on a few extra arguments but I'm not sure which yet.
Also replaced the `unwrap()` with a more informative `expect()`.
r? @jseyfried
Introduce HirId, a replacement for ast::NodeId after lowering to HIR
This is the first step towards implementing #40303. This PR introduces the `HirId` type and generates a `HirId` for everything that would be assigned one (i.e. stuff in the HIR), but the HIR data types still use `NodeId` for now. Changing that is a big refactoring that I want to do in a separate PR.
A `HirId` uniquely identifies a node in the HIR of the current crate. It is composed of the `owner`, which is the `DefIndex` of the directly enclosing `hir::Item`, `hir::TraitItem`, or `hir::ImplItem` (i.e. the closest "item-like"), and the `local_id` which is unique within the given owner.
This PR is also running a number of consistency checks for the generated `HirId`s:
- Does `NodeId` in the HIR have a corresponding `HirId`?
- Is the `owner` part of each `HirId` consistent with its position in the HIR?
- Do the numerical values of the `local_id` part all lie within a dense range of integers?
cc @rust-lang/compiler
r? @eddyb or @nikomatsakis
This commit is a random stab in the dark to fix the spurious failures on #39518.
The leading theory of the spurious failures on Windows is that the compiler is
loading a path in the `deps` folder, passing it to `link.exe`, and then this is
racing with Cargo itself updating those paths.
This race, however, has a few unique properties:
* It's isolated to just libstd. Most crates are never passed to the linker and
simultaneously being worked on by Cargo. Cargo's typical execution of the
dependency graph never hits this problem.
* The crates are already all located in the sysroot in addition to the `deps`
folder. This means that the compiler actually has two candidates of crates to
load, and it's just arbitrarily rejecting one.
Together this means that we shouldn't need to fix this problem "in the large"
and we can instead just fix it in this isolated situation (hopefully). To solve
this the compiler's been updated to prefer crates from the sysroot to leave
Cargo's structure to itself.
We'll see if this actually allows the PR to land...
convert AdtDef::destructor to on-demand
This removes the `Cell` from `AdtDef`. Also, moving destructor validity
checking to on-demand (forced during item-type checking) ensures that
invalid destructors can't cause ICEs.
Fixes#38868.
Fixes#40132.
r? @eddyb
Implement function-like procedural macros ( `#[proc_macro]`)
Adds the `#[proc_macro]` attribute, which expects bare functions of the kind `fn(TokenStream) -> TokenStream`, which can be invoked like `my_macro!()`.
cc rust-lang/rfcs#1913, #38356
r? @jseyfried
cc @nrc
This removes the Cell from AdtDef. Also, moving destructor validity
checking to on-demand (forced during item-type checking) ensures that
invalid destructors can't cause ICEs.
Fixes#38868.
Fixes#40132.
[MIR] SwitchInt Everywhere
Something I've been meaning to do for a very long while. This PR essentially gets rid of 3 kinds of conditional branching and only keeps the most general one - `SwitchInt`. Primary benefits are such that dealing with MIR now does not involve dealing with 3 different ways to do conditional control flow. On the other hand, constructing a `SwitchInt` currently requires more code than what previously was necessary to build an equivalent `If` terminator. Something trivially "fixable" with some constructor methods somewhere (MIR needs stuff like that badly in general).
Some timings (tl;dr: slightly faster^1 (unexpected), but also uses slightly more memory at peak (expected)):
^1: Not sure if the speed benefits are because of LLVM liking the generated code better or the compiler itself getting compiled better. Either way, its a net benefit. The CORE and SYNTAX timings done for compilation without optimisation.
```
AFTER:
Building stage1 std artifacts (x86_64-unknown-linux-gnu -> x86_64-unknown-linux-gnu)
Finished release [optimized] target(s) in 31.50 secs
Finished release [optimized] target(s) in 31.42 secs
Building stage1 compiler artifacts (x86_64-unknown-linux-gnu -> x86_64-unknown-linux-gnu)
Finished release [optimized] target(s) in 439.56 secs
Finished release [optimized] target(s) in 435.15 secs
CORE: 99% (24.81 real, 0.13 kernel, 24.57 user); 358536k resident
CORE: 99% (24.56 real, 0.15 kernel, 24.36 user); 359168k resident
SYNTAX: 99% (49.98 real, 0.48 kernel, 49.42 user); 653416k resident
SYNTAX: 99% (50.07 real, 0.58 kernel, 49.43 user); 653604k resident
BEFORE:
Building stage1 std artifacts (x86_64-unknown-linux-gnu -> x86_64-unknown-linux-gnu)
Finished release [optimized] target(s) in 31.84 secs
Building stage1 compiler artifacts (x86_64-unknown-linux-gnu -> x86_64-unknown-linux-gnu)
Finished release [optimized] target(s) in 451.17 secs
CORE: 99% (24.66 real, 0.20 kernel, 24.38 user); 351096k resident
CORE: 99% (24.36 real, 0.17 kernel, 24.18 user); 352284k resident
SYNTAX: 99% (52.24 real, 0.56 kernel, 51.66 user); 645544k resident
SYNTAX: 99% (51.55 real, 0.48 kernel, 50.99 user); 646428k resident
```
cc @nikomatsakis @eddyb
Previously AdtDef variants contained ConstInt for each discriminant, which did not really reflect
the actual type of the discriminants. Moving the type into AdtDef allows to easily put the type
into metadata and also saves bytes from ConstVal overhead for each discriminant.
Also arguably the code is cleaner now :)
Make reprs use a structured representation instead of a slice
This is needed for `-z reorder-fields`. The old design uses a slice taken from HIR, plus a cache that lazily parses. The new design stores it directly in the `AdtDef` as a `ReprOptions`. We're doing this now because we need to be able to add reprs that don't necessarily exist in HIR for `-z reorder-fields`, but it needs to happen anyway.
`lookup_repr_hints` should be mostly deprecated. I want to remove it from `layout` before closing this, unless people think that should be a separate PR. The `[WIP]` is because of this. The problem with closing this as-is is that the code here isn't actually testable until some parts of the compiler start using it.
r? @eddyb
Expand derive macros in the MacroExpander
This removes the expand_derives function, and sprinkles the functionality throughout the Invocation Collector, Expander and Resolver.
Fixes https://github.com/rust-lang/rust/issues/39326
r? @jseyfried
Implement kind="static-nobundle" (RFC 1717)
This implements the "static-nobundle" library kind (last item from #37403).
Rustc handles "static-nobundle" libs very similarly to dylibs, except that on Windows, uses of their symbols do not get marked with "dllimport". Which is the whole point of this feature.
