| Age | Commit message (Collapse) | Author | Lines |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
typo
|
|
|
|
|
|
|
|
|
|
Reported be @ElBaha
|
|
Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>
|
|
This bug showed up because the visitor only visited the path of the implemented
trait via walk_path (with no corresponding visit_path function). I have modified
the visitor to use visit_path (which is now overridable), and the privacy
visitor overrides this function and now properly checks for the privacy of all
paths.
Closes #10857
|
|
This bug showed up because the visitor only visited the path of the implemented
trait via walk_path (with no corresponding visit_path function). I have modified
the visitor to use visit_path (which is now overridable), and the privacy
visitor overrides this function and now properly checks for the privacy of all
paths.
Closes #10857
|
|
Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>
|
|
Specifically, we can now use:
+ beginning-of-defun
+ end-of-defun
+ mark-defun
where "defun" means a Rust item.
+ Add tests in rust-mode-tests.el
+ Fix indentation in rust-mode-tests.el
+ Add support for trait to Imenu
|
|
- Removed the `log` keyword;
- Removed keyword duplicates;
- Highlighted `const` as `Error` rather than `StorageClass`; and
- Highlighted all the reserved keywords as `Error` rather than as
`Keyword`.
(As usual, these highlightings can be overridden if desired.)
|
|
This fixes a regression introduced in #10793.
Having a colorscheme which highlights Float the same as Number (I
believe most do), I hadn't noticed that having the special case of "5."
floats (which was one of the added features in #10793) last made it take
precedence, and so it was left to @thestinger to notice it.
The regression meant that in `5.0`, the `5.` was a `rustFloat` (linked
by default to `Float`) and the `0` was a `rustDecNumber` (linked by
default to `Number`), and for `5.0f32` the `5.` was a `rustFloat` and
the `0f32` was a second `rustFloat` (and thus appeared correctly, though
for the wrong reason).
|
|
The first commit was approved from another pull request, but I wanted to rebase LTO on top of it.
LTO is not turned on by default at all, and it's hidden behind a `-Z` flag. I have added a few small tests for it, however.
|
|
Makes the `dead_code` lint a citizen of stage0 to silence the "unknown lint"
warnings.
|
|
This commit implements LTO for rust leveraging LLVM's passes. What this means
is:
* When compiling an rlib, in addition to insdering foo.o into the archive, also
insert foo.bc (the LLVM bytecode) of the optimized module.
* When the compiler detects the -Z lto option, it will attempt to perform LTO on
a staticlib or binary output. The compiler will emit an error if a dylib or
rlib output is being generated.
* The actual act of performing LTO is as follows:
1. Force all upstream libraries to have an rlib version available.
2. Load the bytecode of each upstream library from the rlib.
3. Link all this bytecode into the current LLVM module (just using llvm
apis)
4. Run an internalization pass which internalizes all symbols except those
found reachable for the local crate of compilation.
5. Run the LLVM LTO pass manager over this entire module
6a. If assembling an archive, then add all upstream rlibs into the output
archive. This ignores all of the object/bitcode/metadata files rust
generated and placed inside the rlibs.
6b. If linking a binary, create copies of all upstream rlibs, remove the
rust-generated object-file, and then link everything as usual.
As I have explained in #10741, this process is excruciatingly slow, so this is
*not* turned on by default, and it is also why I have decided to hide it behind
a -Z flag for now. The good news is that the binary sizes are about as small as
they can be as a result of LTO, so it's definitely working.
Closes #10741
Closes #10740
|
|
Hi, first pull request here so let me know if I've missed any of the procedure.
|
|
|
|
|
|
|
|
|
|
Right now whenever an rlib file is linked against, all of the metadata from the
rlib is pulled in to the final staticlib or binary. The reason for this is that
the metadata is currently stored in a section of the object file. Note that this
is intentional for dynamic libraries in order to distribute metadata bundled
with static libraries.
This commit alters the situation for rlib libraries to instead store the
metadata in a separate file in the archive. In doing so, when the archive is
passed to the linker, none of the metadata will get pulled into the result
executable. Furthermore, the metadata file is skipped when assembling rlibs into
an archive.
The snag in this implementation comes with multiple output formats. When
generating a dylib, the metadata needs to be in the object file, but when
generating an rlib this needs to be separate. In order to accomplish this, the
metadata variable is inserted into an entirely separate LLVM Module which is
then codegen'd into a different location (foo.metadata.o). This is then linked
into dynamic libraries and silently ignored for rlib files.
While changing how metadata is inserted into archives, I have also stopped
compressing metadata when inserted into rlib files. We have wanted to stop
compressing metadata, but the sections it creates in object file sections are
apparently too large. Thankfully if it's just an arbitrary file it doesn't
matter how large it is.
I have seen massive reductions in executable sizes, as well as staticlib output
sizes (to confirm that this is all working).
|
|
|
|
|
|
|
|
|
|
This moves `std::rand::distribitions::{Normal, StandardNormal}` to `...::distributions::normal`, reexporting `Normal` from `distributions` (and similarly for `Exp` and Exp1`), and adds:
- Log-normal
- Chi-squared
- F
- Student T
all of which are implemented in C++11's random library. Tests in https://github.com/huonw/random-tests/commit/0424b8aded5e608ae386c1f917934a726d9cac6a. Note that these are approximately half documentation & half implementation (of which a significant portion is boilerplate `}`'s and so on).
|
|
Last LLVM update seems to have fixed whatever prevented LLVM integrated assembler from generating correct unwind tables on Windows. This PR switches Windows builds to use internal assembler by default.
Compilation via external assembler can still be requested via the newly added `-Z no-integrated-as` option.
Closes #8809
|
|
Closes #10853
|
|
Specifically, we can now use:
+ beginning-of-defun
+ end-of-defun
+ mark-defun
where "defun" means a Rust item.
+ Add tests in rust-mode-tests.el
+ Fix indentation in rust-mode-tests.el
+ Add support for trait to Imenu
|
|
|
|
|
|
Closes #10853
|
|
A comment that I previously added to ast::DefStruct is incorrect. Here's the modification.
|
|
|
|
Before applying this patch, the included testcase fails with:
```
src/test/run-pass/xcrate-trait-lifetime-param.rs:20:10: 20:28 error: wrong number of lifetime parameters: expected 0 but found 1
src/test/run-pass/xcrate-trait-lifetime-param.rs:20 impl <'a> other::FromBuf<'a> for Reader<'a> {
^~~~~~~~~~~~~~~~~~
```
There's another example in my comments to #10506.
|
|
|
|
PR for issue #1749 mainly to get some feedback and suggestion. This adds a pass that warns if a function, struct, enum, or static item is never used. For the following code,
```rust
pub static pub_static: int = 0;
static priv_static: int = 0;
static used_static: int = 0;
pub fn pub_fn() { used_fn(); }
fn priv_fn() { let unused_struct = PrivStruct; }
fn used_fn() {}
pub struct PubStruct();
struct PrivStruct();
struct UsedStruct1 { x: int }
struct UsedStruct2(int);
struct UsedStruct3();
pub enum pub_enum { foo1, bar1 }
enum priv_enum { foo2, bar2 }
enum used_enum { foo3, bar3 }
fn foo() {
bar();
let unused_enum = foo2;
}
fn bar() {
foo();
}
fn main() {
let used_struct1 = UsedStruct1 { x: 1 };
let used_struct2 = UsedStruct2(1);
let used_struct3 = UsedStruct3;
let t = used_static;
let e = foo3;
}
```
it would add the following warnings:
```rust
/home/ktt3ja/test.rs:2:0: 2:28 warning: code is never used: `priv_static`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:2 static priv_static: int = 0;
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/ktt3ja/test.rs:6:0: 6:48 warning: code is never used: `priv_fn`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:6 fn priv_fn() { let unused_struct = PrivStruct; }
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/ktt3ja/test.rs:10:0: 10:20 warning: code is never used: `PrivStruct`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:10 struct PrivStruct();
^~~~~~~~~~~~~~~~~~~~
/home/ktt3ja/test.rs:16:0: 16:29 warning: code is never used: `priv_enum`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:16 enum priv_enum { foo2, bar2 }
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/ktt3ja/test.rs:19:0: 22:1 warning: code is never used: `foo`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:19 fn foo() {
/home/ktt3ja/test.rs:20 bar();
/home/ktt3ja/test.rs:21 let unused_enum = foo2;
/home/ktt3ja/test.rs:22 }
/home/ktt3ja/test.rs:24:0: 26:1 warning: code is never used: `bar`, #[warn(dead_code)] on by default
/home/ktt3ja/test.rs:24 fn bar() {
/home/ktt3ja/test.rs:25 foo();
/home/ktt3ja/test.rs:26 }
```
Furthermore, I would like to solicit some test cases since I haven't tested extensively and I'm still unclear about some of the things in here. For example, I'm not sure how reexports would affect this and just assumed that LiveContext (which is a copy of reachable::ReachableContext) does enough work to handle it. Also, the test case above doesn't include any impl or methods, etc.
|
|
This transitions the snapshot dependency process to understand that our
snapshots are now a single static binary rather than an array of files.
|
|
|
|
|
|
|
|
|
|
|
