rustc: Implement #[link(cfg(..))] and crt-static
This commit is an implementation of [RFC 1721] which adds a new target feature
to the compiler, `crt-static`, which can be used to select how the C runtime for
a target is linked. Most targets dynamically linke the C runtime by default with
the notable exception of some of the musl targets.
[RFC 1721]: https://github.com/rust-lang/rfcs/blob/master/text/1721-crt-static.md
This commit first adds the new target-feature, `crt-static`. If enabled, then
the `cfg(target_feature = "crt-static")` will be available. Targets like musl
will have this enabled by default. This feature can be controlled through the
standard target-feature interface, `-C target-feature=+crt-static` or
`-C target-feature=-crt-static`.
Next this adds an gated and unstable `#[link(cfg(..))]` feature to enable the
`crt-static` semantics we want with libc. The exact behavior of this attribute
is a little squishy, but it's intended to be a forever-unstable
implementation detail of the liblibc crate.
Specifically the `#[link(cfg(..))]` annotation means that the `#[link]`
directive is only active in a compilation unit if that `cfg` value is satisfied.
For example when compiling an rlib, these directives are just encoded and
ignored for dylibs, and all staticlibs are continued to be put into the rlib as
usual. When placing that rlib into a staticlib, executable, or dylib, however,
the `cfg` is evaluated *as if it were defined in the final artifact* and the
library is decided to be linked or not.
Essentially, what'll happen is:
* On MSVC with `-C target-feature=-crt-static`, the `msvcrt.lib` library will be
linked to.
* On MSVC with `-C target-feature=+crt-static`, the `libcmt.lib` library will be
linked to.
* On musl with `-C target-feature=-crt-static`, the object files in liblibc.rlib
are removed and `-lc` is passed instead.
* On musl with `-C target-feature=+crt-static`, the object files in liblibc.rlib
are used and `-lc` is not passed.
This commit does **not** include an update to the liblibc module to implement
these changes. I plan to do that just after the 1.14.0 beta release is cut to
ensure we get ample time to test this feature.
cc #37406
This commit is an implementation of [RFC 1721] which adds a new target feature
to the compiler, `crt-static`, which can be used to select how the C runtime for
a target is linked. Most targets dynamically linke the C runtime by default with
the notable exception of some of the musl targets.
[RFC 1721]: https://github.com/rust-lang/rfcs/blob/master/text/1721-crt-static.md
This commit first adds the new target-feature, `crt-static`. If enabled, then
the `cfg(target_feature = "crt-static")` will be available. Targets like musl
will have this enabled by default. This feature can be controlled through the
standard target-feature interface, `-C target-feature=+crt-static` or
`-C target-feature=-crt-static`.
Next this adds an gated and unstable `#[link(cfg(..))]` feature to enable the
`crt-static` semantics we want with libc. The exact behavior of this attribute
is a little squishy, but it's intended to be a forever-unstable
implementation detail of the liblibc crate.
Specifically the `#[link(cfg(..))]` annotation means that the `#[link]`
directive is only active in a compilation unit if that `cfg` value is satisfied.
For example when compiling an rlib, these directives are just encoded and
ignored for dylibs, and all staticlibs are continued to be put into the rlib as
usual. When placing that rlib into a staticlib, executable, or dylib, however,
the `cfg` is evaluated *as if it were defined in the final artifact* and the
library is decided to be linked or not.
Essentially, what'll happen is:
* On MSVC with `-C target-feature=-crt-static`, the `msvcrt.lib` library will be
linked to.
* On MSVC with `-C target-feature=+crt-static`, the `libcmt.lib` library will be
linked to.
* On musl with `-C target-feature=-crt-static`, the object files in liblibc.rlib
are removed and `-lc` is passed instead.
* On musl with `-C target-feature=+crt-static`, the object files in liblibc.rlib
are used and `-lc` is not passed.
This commit does **not** include an update to the liblibc module to implement
these changes. I plan to do that just after the 1.14.0 beta release is cut to
ensure we get ample time to test this feature.
cc #37406
rustdoc: Fix some local inlining issues
* Only inline public items when inlining glob imports.
* Never inline while in a private module or a child of a private module.
* Never inline impls. This allowed the removal of a workaround in the
rendering code.
test: Move missing-items to a ui test
This test is failing on nightly for unknown reasons, and my best guess is a
difference in grep versions which is interpreting symbols differently. For now
let's just move this to a ui test and hope it fixes nightlies.
* Only inline public items when inlining glob imports.
* Never inline while in a private module or a child of a private module.
* Never inline impls. This allowed the removal of a workaround in the
rendering code.
This test is failing on nightly for unknown reasons, and my best guess is a
difference in grep versions which is interpreting symbols differently. For now
let's just move this to a ui test and hope it fixes nightlies.
Fix syntax error in the compiler
Currently `rustc` accepts the following code: `fn f<'a>() where 'a {}`. This should be a syntax error, shouldn't it?
Not sure if my changes actually compile, waiting for the LLVM to build.
On fmt string with unescaped `{` note how to escape
On cases of malformed format strings where a `{` hasn't been properly escaped, like `println!("{");`, present a NOTE explaining how to escape the `{` char.
Fix#34300.
Add foreign formatting directive detection.
This teaches `format_args!` how to interpret format printf- and
shell-style format directives. This is used in cases where there are
unused formatting arguments, and the reason for that *might* be because
the programmer is trying to use the wrong kind of formatting string.
This was prompted by an issue encountered by simulacrum on the #rust IRC
channel. In short: although `println!` told them that they weren't using
all of the conversion arguments, the problem was in using printf-syle
directives rather than ones `println!` would undertand.
Where possible, `format_args!` will tell the programmer what they should
use instead. For example, it will suggest replacing `%05d` with `{:0>5}`,
or `%2$.*3$s` with `{1:.3$}`. Even if it cannot suggest a replacement,
it will explicitly note that Rust does not support that style of directive,
and direct the user to the `std::fmt` documentation.
-----
**Example**: given:
```rust
fn main() {
println!("%.*3$s %s!\n", "Hello,", "World", 4);
println!("%1$*2$.*3$f", 123.456);
}
```
The compiler outputs the following:
```text
error: multiple unused formatting arguments
--> local/fmt.rs:2:5
|
2 | println!("%.*3$s %s!\n", "Hello,", "World", 4);
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
note: argument never used
--> local/fmt.rs:2:30
|
2 | println!("%.*3$s %s!\n", "Hello,", "World", 4);
| ^^^^^^^^
note: argument never used
--> local/fmt.rs:2:40
|
2 | println!("%.*3$s %s!\n", "Hello,", "World", 4);
| ^^^^^^^
note: argument never used
--> local/fmt.rs:2:49
|
2 | println!("%.*3$s %s!\n", "Hello,", "World", 4);
| ^
= help: `%.*3$s` should be written as `{:.2$}`
= help: `%s` should be written as `{}`
= note: printf formatting not supported; see the documentation for `std::fmt`
= note: this error originates in a macro outside of the current crate
error: argument never used
--> local/fmt.rs:6:29
|
6 | println!("%1$*2$.*3$f", 123.456);
| ^^^^^^^
|
= help: `%1$*2$.*3$f` should be written as `{0:1$.2$}`
= note: printf formatting not supported; see the documentation for `std::fmt`
```
Move all Linux/OSX CI infastructure to Travis
This commit configures our `.travis.yml` to test the full suite of tests we have
on Buildbot right now. A whole mess of docker images are added to the `src/ci`
directory which represent all the build environments for each configuration.
Each of these environments is then configured in `.travis.yml` to run on the
auto branch.
Note that the full matrix of tests aren't intended to be run on all PRs.
Instead, we continue to run only one entry in the matrix, forcing all others to
finish quickly. Only the `auto` branch should run the full matrix of builds.
Also note that the infrastructure hasn't quite been allocated yet to the
rust-lang/rust repository, so everything is disabled for now except for the one
build that happens on PRs. Once that infrastructure is allocated though we can
enable this and let it fly!
Notable modifications from the current test suite today:
* Android tests are run in rustbuild instead of the makefiles, for whatever
reason I couldn't get the makefiles to work on Travis.
* A debuginfo test was updated to work with the current version of the Android
NDK.
* Some dependencies in `mk/tests.mk` were fixed to allow running tests in
parallel.
Don't provide hint to add lifetime on impl items
``` rust
use std::str::FromStr;
pub struct Foo<'a> {
field: &'a str,
}
impl<'a> FromStr for Foo<'a> {
type Err = ();
fn from_str(path: &str) -> Result<Self, ()> {
Ok(Foo { field: path })
}
}
```
would give the following hint:
``` nocode
help: consider using an explicit lifetime parameter as shown: fn from_str(path: &'a str) -> Result<Self, ()>
--> <anon>:9:5
|
9 | fn from_str(path: &str) -> Result<Self, ()> {
| ^
```
which is never correct, since then there will be a lifetime mismatch between the `impl` and the trait.
Remove this hint for all `impl` items.
Re: #37363.
rustdoc: add line breaks to where clauses a la rustfmt
Much like my last PR for rustdoc (#36679), this adds line breaks to certain statements based on their line length. Here the focus was on where clauses.
Some examples:
- [Where clause in a trait function](https://shiva.icesoldier.me/custom-std/std/iter/trait.Iterator.html?search=#method.unzip) (also in the trait header block at the top of the page)
- [Where clause on a bare function](https://shiva.icesoldier.me/doc-custom2/petgraph/visit/fn.depth_first_search.html)
- [Where clauses in trait impls on a struct](https://shiva.icesoldier.me/custom-std/std/collections/struct.HashMap.html) (scroll to the bottom) These are regularly not on their own line, but will be given their own line now if their "prefix text" doesn't give them enough room to sensibly print their constraints. HashMap's trait impls provide some examples of both behaviors.
The libstd links above are the whole docs rendered with this, and the "bare function" link above is in another set that pulls some notable crates together. `petgraph` was the one that brought this request up, and that collection also includes [itertools](https://shiva.icesoldier.me/doc-custom2/itertools/trait.Itertools.html) which provided an easy sample to test with.
r? @GuillaumeGomez
On cases of malformed format strings where a `{` hasn't been properly
escaped, like `println!("{");`, present a note explaining how to escape
the `{` char.
Group unused import warnings per import list
Given a file
``` rust
use std::collections::{BinaryHeap, BTreeMap, BTreeSet};
fn main() {}
```
Show a single warning, instead of three for each unused import:
``` nocode
warning: unused imports, #[warn(unused_imports)] on by default
--> file2.rs:1:24
|
1 | use std::collections::{BinaryHeap, BTreeMap, BTreeSet};
| ^^^^^^^^^^ ^^^^^^^^ ^^^^^^^^
```
Include support for lints pointing at `MultilineSpan`s, instead of just
`Span`s.
Fixes#16132.
This commit configures our `.travis.yml` to test the full suite of tests we have
on Buildbot right now. A whole mess of docker images are added to the `src/ci`
directory which represent all the build environments for each configuration.
Each of these environments is then configured in `.travis.yml` to run on the
auto branch.
Note that the full matrix of tests aren't intended to be run on all PRs.
Instead, we continue to run only one entry in the matrix, forcing all others to
finish quickly. Only the `auto` branch should run the full matrix of builds.
Also note that the infrastructure hasn't quite been allocated yet to the
rust-lang/rust repository, so everything is disabled for now except for the one
build that happens on PRs. Once that infrastructure is allocated though we can
enable this and let it fly!
Notable modifications from the current test suite today:
* Android tests are run in rustbuild instead of the makefiles, for whatever
reason I couldn't get the makefiles to work on Travis.
* A debuginfo test was updated to work with the current version of the Android
NDK.
* Some dependencies in `mk/tests.mk` were fixed to allow running tests in
parallel.
Show one error for duplicated type definitions
For the following code:
``` rustc
struct Bar;
struct Bar;
fn main () {
}
```
show
``` nocode
error[E0428]: a type named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error: aborting due to previous error
```
instead of
``` nocode
error[E0428]: a type named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error[E0428]: a value named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error: aborting due to 2 previous errors
```
Fixes#35767.
Don't spin expanding stmt macros.
If we can't make progress when parsing a macro expansion as a statement then we should just bail.
This alleviates the symptoms shown in e.g. #37113 and #37234 but it doesn't fix the problem that parsing invalid enum bodies (and others) leaves the parser in a crappy state.
I'm not sold on this strategy (checking `tokens_consumed`), so if anyone has a better idea, I'm all ears!
Make E0243/E0244 message consistent with E0107
E0243/E0233 prints `expected {}, found {}` on the span note, while E0107 prints it on the first line. This is confusing when both error occur simultaneously.
This PR makes E0243/E0233 print `expected {}, found {}` on the first line.
Code:
``` rust
struct Foo<'a, 'b> {
s: &'a str,
t: &'b str,
}
type Bar<T, U> = Foo<T, U>;
```
rustc output (before):
```
error[E0107]: wrong number of lifetime parameters: expected 2, found 0
--> test.rs:6:18
|
6 | type Bar<T, U> = Foo<T, U>;
| ^^^^^^^^^ expected 2 lifetime parameters
error[E0244]: wrong number of type arguments
--> test.rs:6:18
|
6 | type Bar<T, U> = Foo<T, U>;
| ^^^^^^^^^ expected no type arguments, found 2
```
rustc output (after):
```
error[E0107]: wrong number of lifetime parameters: expected 2, found 0
--> /tmp/test.rs:6:18
|
6 | type Bar<T, U> = Foo<T, U>;
| ^^^^^^^^^ expected 2 lifetime parameters
error[E0244]: wrong number of type arguments: expected 0, found 2
--> /tmp/test.rs:6:18
|
6 | type Bar<T, U> = Foo<T, U>;
| ^^^^^^^^^ expected no type arguments
```
This teaches `format_args!` how to interpret format printf- and
shell-style format directives. This is used in cases where there are
unused formatting arguments, and the reason for that *might* be because
the programmer is trying to use the wrong kind of formatting string.
This was prompted by an issue encountered by simulacrum on the #rust IRC
channel. In short: although `println!` told them that they weren't using
all of the conversion arguments, the problem was in using printf-syle
directives rather than ones `println!` would undertand.
Where possible, `format_args!` will tell the programmer what they should
use instead. For example, it will suggest replacing `%05d` with `{:0>5}`,
or `%2$.*3$s` with `{1:.3$}`. Even if it cannot suggest a replacement,
it will explicitly note that Rust does not support that style of directive,
and direct the user to the `std::fmt` documentation.
Don't provide hint to add lifetime on impl items that implement a trait.
```rust
use std::str::FromStr;
pub struct Foo<'a> {
field: &'a str,
}
impl<'a> FromStr for Foo<'a> {
type Err = ();
fn from_str(path: &str) -> Result<Self, ()> {
Ok(Foo { field: path })
}
}
```
would give the following hint:
```nocode
help: consider using an explicit lifetime parameter as shown: fn from_str(path: &'a str) -> Result<Self, ()>
--> <anon>:9:5
|
9 | fn from_str(path: &str) -> Result<Self, ()> {
| ^
```
which is never correct, since then there will be a lifetime mismatch
between the impl and the trait.
Remove this hint for impl items that implement a trait.
Fix regression involving custom derives on items with `$crate`
The regression was introduced in #37213.
I believe we cannot make the improvements from #37213 work with the current custom derive setup (c.f. https://github.com/rust-lang/rust/issues/37637#issuecomment-258959145) -- we'll have to wait for `TokenStream`'s API to improve.
Fixes#37637.
r? @nrc
For the following code:
```rustc
struct Bar;
struct Bar;
fn main () {
}
```
show
```nocode
error[E0428]: a type named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error: aborting due to previous error
```
instead of
```nocode
error[E0428]: a type named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error[E0428]: a value named `Bar` has already been defined in this module
--> src/test/compile-fail/E0428.rs:12:1
|
11 | struct Bar;
| ----------- previous definition of `Bar` here
12 | struct Bar;
| ^^^^^^^^^^^
error: aborting due to 2 previous errors
```
_match: correct max_slice_length logic
The logic used to be wildly wrong, but before the HAIR patch its wrongness was in most cases hidden by another bug.
Fixes#37598.
r? @nikomatsakis
