* Consumers of handle_options assume the unstable options are defined in
the getopts::Matches value if -Z unstable-options is set, but that's not
the case if there weren't any actual unstable options. Fix this by
always reparsing options when -Z unstable-options is set.
* If both argument parsing attempts fail, print the error from the second
attempt rather than the first. The error from the first is very poor
whenever unstable options are present. e.g.:
$ rustc hello.rs -Z unstable-options --show-span
error: Unrecognized option: 'show-span'.
$ rustc hello.rs -Z unstable-options --pretty --pretty
error: Unrecognized option: 'pretty'.
$ rustc hello.rs -Z unstable-options --pretty --bad-option
error: Unrecognized option: 'pretty'.
* On the second parse, add a separate pass to reject unstable options if
-Z unstable-options wasn't specified.
Fixes#21715.
r? @pnkfelix
Display::fmt for str calls into Formatter::pad, which is modest in size
and also pulls in string-related functions for its truncation and padding
abilities. For size-critical programs (e.g. embedded), this call site
may be the only reason Formatter::pad is linked into the output.
This commit performs another pass over the `std::char` module for stabilization.
Some minor cleanup is performed such as migrating documentation from libcore to
libunicode (where the `std`-facing trait resides) as well as a slight
reorganiation in libunicode itself. Otherwise, the stability modifications made
are:
* `char::from_digit` is now stable
* `CharExt::is_digit` is now stable
* `CharExt::to_digit` is now stable
* `CharExt::to_{lower,upper}case` are now stable after being modified to return
an iterator over characters. While the implementation today has not changed
this should allow us to implement the full set of case conversions in unicode
where some characters can map to multiple when doing an upper or lower case
mapping.
* `StrExt::to_{lower,upper}case` was added as unstable for a convenience of not
having to worry about characters expanding to more characters when you just
want the whole string to get into upper or lower case.
This is a breaking change due to the change in the signatures of the
`CharExt::to_{upper,lower}case` methods. Code can be updated to use functions
like `flat_map` or `collect` to handle the difference.
[breaking-change]
Closes#20333
This commit performs another pass over the `std::char` module for stabilization.
Some minor cleanup is performed such as migrating documentation from libcore to
libunicode (where the `std`-facing trait resides) as well as a slight
reorganiation in libunicode itself. Otherwise, the stability modifications made
are:
* `char::from_digit` is now stable
* `CharExt::is_digit` is now stable
* `CharExt::to_digit` is now stable
* `CharExt::to_{lower,upper}case` are now stable after being modified to return
an iterator over characters. While the implementation today has not changed
this should allow us to implement the full set of case conversions in unicode
where some characters can map to multiple when doing an upper or lower case
mapping.
* `StrExt::to_{lower,upper}case` was added as unstable for a convenience of not
having to worry about characters expanding to more characters when you just
want the whole string to get into upper or lower case.
This is a breaking change due to the change in the signatures of the
`CharExt::to_{upper,lower}case` methods. Code can be updated to use functions
like `flat_map` or `collect` to handle the difference.
[breaking-change]
These new APIs have had some time to bake now, and no pressing issues have come
up so they should be ok for stabilizing. Specifically, these two APIs were
stabilized:
* `slice::from_raw_parts`
* `slice::from_raw_parts_mut`
Add tests checking that a number of feature gates are gating their features
Namely:
* `quote`
* `link_args`
* `link_llvm_intrinsics`
* `thread_local`
* `unsafe_destructor`
Also updates test for `plugin_registrar` to make it clear that
it is only testing the `plugin_registrar` feature gate.
Cc #22820. (Latter is not fixed, since there are still a bunch more feature-gates that need tests. But I wanted to stop here and move on to something else.)
There was a buildbot failure recently of an arithmetic overflow in the isaac
module of the rand crate, so I've imported the isaac implementation from
out-of-tree which makes somewhat more liberal usage of the wrapping primitives.
Hopefull this quelches any future overflow!
Switching from generic bounds to trait objects and having un-inlined
inner methods should cut down on the size of Debug impls, since we care
about the speed of a Debug implementation way less than binary bloat.
There was a buildbot failure recently of an arithmetic overflow in the isaac
module of the rand crate, so I've imported the isaac implementation from
out-of-tree which makes somewhat more liberal usage of the wrapping primitives.
Hopefull this quelches any future overflow!
This may not be quite ready to go out, I fixed some docs but suspect I missed a bunch.
I also wound up fixing a bunch of redundant `[]` suffixes, but on closer inspection I don't believe that can land until after a snapshot.
Namely:
* `quote`
* `link_args`
* `link_llvm_intrinsics`
* `thread_local`
* `unsafe_destructor`
Also updates test for `plugin_registrar` to make it clear that
it is only testing the `plugin_registrar` feature gate.
Cc #22820.
Motivated by the test output not lining up when it could, I normalized all of the issue-* tests.
While doing it, I found some lexer tests that could be unignored and fixed an int -> isize.
If we end the `scoped` call with a semicolon, the `JoinGuard` will be
dropped and not returned from the `map`. The thread will start up and
we immediately block, making for a very expensive sequential loop.
rustc will ICE if you specify an outfile path that is bare without a
directory. As a workaround, before this -o ./foo will work
It wasn't clear to me where I could put a test that actually invokes rustc from a shell, although I think I can add doctests to that machinery in librustc_driver that will arrange for this to be called with arguments that would trigger the ICE
Motivated by the test output not lining up when it could, I normalized all of the issue-* tests.
While doing it, I found some lexer tests that could be unignored and fixed an int -> isize.
If we end the `scoped` call with a semicolon, the `JoinGuard` will be
dropped and not returned from the `map`. The thread will start up and
we immediately block, making for a very expensive sequential loop.
The docs currently define `array_expr`s as:
array_expr : '[' \"mut\" ? vec_elems? ']' ;
array_elems : [expr [',' expr]*] | [expr ',' \"..\" expr] ;
`vec_elems` is not defined anywhere else so it is probably a typo for `array_elems`.
Building on #22076, I've added some tests for stable methods in f32 and f64 that didn't have any before.
Please let me know if there are any improvements I can make, and I am happy to make them! 📬
Building on #22076, I've added some tests for stable methods in f32 and f64 that didn't have any before.
Please let me know if there are any improvements I can make, and I am happy to make them! 📬
* "let mut text" was previously of &String type. Now it is of &str type.
* Update the slicing syntax. Both &text[] and text.slice_from() evaluate
to a &str.
* We were passing a u32 to expr_usize. Call expr_u32 instead.
r? @steveklabnik
Fixes#23166
Right now the rust upgrade in cargo is blocked on fixing this overflow. If a
this benchmark is run it will trigger an overflow error today:
#[bench]
fn foo(b: &mut test::Bencher) {}
This commit adds a check on each iteration of the loop that the maximum
multiplier (10) doesn't overflow, and if it does just return the results so far.
* "let met text" was previously of &String type. Now it is of &str type.
* Update the slicing syntax. Both &text[] and text.slice_from() evaluate
to a &str.
* We were passing a u32 to expr_usize. Call expr_u32 instead.
This is a hack, but I don't think we can do much better as long as `derive` is running at the syntax expansion phase.
If the `custom_derive` feature gate is enabled, this works with user-defined traits and syntax extensions. Without the gate, you can't use e.g. `#[derive_Clone]` directly, so this does not change the stable language.
To make this effective, we now check gated attributes both before and after macro expansion. This uncovered a number of tests that were missing feature gates.
This PR also cleans up the deriving code somewhat, and forbids some previously-meaningless attribute syntax. For this reason it's technically a
[breaking-change]
r? @sfackler
We require the *deferred* loading, not just an opportunistic
asynchronous loading. I think `<script defer>` is safe to use,
according to <http://caniuse.com/#feat=script-defer>.
This is a hack, but I don't think we can do much better as long as `derive` is
running at the syntax expansion phase.
If the custom_derive feature gate is enabled, this works with user-defined
traits and syntax extensions. Without the gate, you can't use e.g. #[derive_Clone]
directly, so this does not change the stable language.
This commit also cleans up the deriving code somewhat, and forbids some
previously-meaningless attribute syntax. For this reason it's technically a
[breaking-change]
Being a person who somehow has taken a liking to premature optimisation, my knee-jerk reaction to
locking in std handles was preamble resembling following snippet:
let stdout = stdout();
let lstdout = stdout.lock();
let stdin = stdin();
let lstdin = stdin.lock();
and then reading from the locked handle like this:
let mut letter = [0; 1];
lstdin.read(&mut letter).unwrap();
As it is now this code will deadlock because the `read` method attempts to lock stdout as well!
r? @alexcrichton
---
Either way, I find flushing stdout when stdin is used debatable. I believe people who write prompts should take care to flush stdout when necessary themselves.
Another idea: Would be cool if locks on std handles would be taken for a thread, rather than a handle, so given preamble (first code snippet)
stdin.lock()
or more generally
stdin.read(…)
worked fine. I.e. if more than a single lock are all taken inside the same thread, it would work, though not sure if our synchronisation primitives are expressive enough to make it possible.
The `std::net` primitives should be ready for use now and as a result the old
ones are now deprecated and slated for removal. Most TCP/UDP functionality is
now available through `std::net` but the `std::old_io::net::pipe` module is
removed entirely from the standard library.
Unix socket funtionality can be found in sfackler's [`unix_socket`][unix] crate
and there is currently no replacement for named pipes on Windows.
[unix]: https://crates.io/crates/unix_socket
[breaking-change]
since there are separate checks that apply to Copy (and Send uses the
generic defaulted trait rules). Also prohibit `Sized` from being
manually implemented for now.
now have a simple set of trait def-ids. During coherence, we use a
separate table to track the default impls for any given trait so that we
can report a nice error. This fixes various bugs in the metadata
encoding that led to `ty::trait_has_default_impl` yielding the wrong
values in the cross-crate case. (In particular, default impl def-ids
were not included in the list of all impl def-ids; I debated fixing just
that, but this approach seemed cleaner overall, since we usually treat
the "defaulted" bit on traits as being a property of the trait, and now
iterating over a list of impls doesn't intermingle default impls with
normal impls.)
Right now the rust upgrade in cargo is blocked on fixing this overflow. If a
this benchmark is run it will trigger an overflow error today:
#[bench]
fn foo(b: &mut test::Bencher) {}
This commit adds a check on each iteration of the loop that the maximum
multiplier (10) doesn't overflow, and if it does just return the results so far.
The `std::net` primitives should be ready for use now and as a result the old
ones are now deprecated and slated for removal. Most TCP/UDP functionality is
now available through `std::net` but the `std::old_io::net::pipe` module is
removed entirely from the standard library.
Unix socket funtionality can be found in sfackler's [`unix_socket`][unix] crate
and there is currently no replacement for named pipes on Windows.
[unix]: https://crates.io/crates/unix_socket
[breaking-change]
This should fix#22615. Previously, the playpen links grabbed the content of all `.rusttest` containers on the same level to build the URL. Now they just select the one before the `pre` they are shown in.
I have only tested this by changing the file in my local build of the docs (not by running rustdoc itself).
This concretely improves type inference of some cases (see included
test). I assume the compiler struggles to reason about multiple layers
of generic type parameters (even with associated-type equalities) but
*can* understand pure associated types, since they are always directly
computable from the input types.
Thanks to @shepmaster for noticing the issue with `Cloned` (I took that example as a test case).
Being a person who somehow has taken a liking to premature optimisation, my knee-jerk reaction to
locking in std handles was preamble resembling following snippet:
let stdout = stdout();
let lstdout = stdout.lock();
let stdin = stdin();
let lstdin = stdin.lock();
and then reading from the locked handle like this:
let mut letter = [0; 1];
lstdin.read(&mut letter).unwrap();
As it is now this code will deadlock because the `read` method attempts to lock stdout as well!
Unstable items used in a macro expansion will now always trigger
stability warnings, *unless* the unstable items are directly inside a
macro marked with `#[allow_internal_unstable]`. IOW, the compiler warns
unless the span of the unstable item is a subspan of the definition of a
macro marked with that attribute.
E.g.
#[allow_internal_unstable]
macro_rules! foo {
($e: expr) => {{
$e;
unstable(); // no warning
only_called_by_foo!();
}}
}
macro_rules! only_called_by_foo {
() => { unstable() } // warning
}
foo!(unstable()) // warning
The unstable inside `foo` is fine, due to the attribute. But the
`unstable` inside `only_called_by_foo` is not, since that macro doesn't
have the attribute, and the `unstable` passed into `foo` is also not
fine since it isn't contained in the macro itself (that is, even though
it is only used directly in the macro).
In the process this makes the stability tracking much more precise,
e.g. previously `println!("{}", unstable())` got no warning, but now it
does. As such, this is a bug fix that may cause [breaking-change]s.
The attribute is definitely feature gated, since it explicitly allows
side-stepping the feature gating system.
---
This updates `thread_local!` macro to use the attribute, since it uses
unstable features internally (initialising a struct with unstable
fields).
This commit performs a stabilization pass over the `std::fs` module now that
it's had some time to bake. The change was largely just adding `#[stable]` tags,
but there are a few APIs that remain `#[unstable]`.
The following apis are now marked `#[stable]`:
* `std::fs` (the name)
* `File`
* `Metadata`
* `ReadDir`
* `DirEntry`
* `OpenOptions`
* `Permissions`
* `File::{open, create}`
* `File::{sync_all, sync_data}`
* `File::set_len`
* `File::metadata`
* Trait implementations for `File` and `&File`
* `OpenOptions::new`
* `OpenOptions::{read, write, append, truncate, create}`
* `OpenOptions::open` - this function was modified, however, to not attempt to
reject cross-platform openings of directories. This means that some platforms
will succeed in opening a directory and others will fail.
* `Metadata::{is_dir, is_file, len, permissions}`
* `Permissions::{readonly, set_readonly}`
* `Iterator for ReadDir`
* `DirEntry::path`
* `remove_file` - like with `OpenOptions::open`, the extra windows code to
remove a readonly file has been removed. This means that removing a readonly
file will succeed on some platforms but fail on others.
* `metadata`
* `rename`
* `copy`
* `hard_link`
* `soft_link`
* `read_link`
* `create_dir`
* `create_dir_all`
* `remove_dir`
* `remove_dir_all`
* `read_dir`
The following apis remain `#[unstable]`.
* `WalkDir` and `walk` - there are many methods by which a directory walk can be
constructed, and it's unclear whether the current semantics are the right
ones. For example symlinks are not handled super well currently. This is now
behind a new `fs_walk` feature.
* `File::path` - this is an extra abstraction which the standard library
provides on top of what the system offers and it's unclear whether we should
be doing so. This is now behind a new `file_path` feature.
* `Metadata::{accessed, modified}` - we do not currently have a good
abstraction for a moment in time which is what these APIs should likely be
returning, so these remain `#[unstable]` for now. These are now behind a new
`fs_time` feature
* `set_file_times` - like with `Metadata::accessed`, we do not currently have
the appropriate abstraction for the arguments here so this API remains
unstable behind the `fs_time` feature gate.
* `PathExt` - the precise set of methods on this trait may change over time and
some methods may be removed. This API remains unstable behind the `path_ext`
feature gate.
* `set_permissions` - we may wish to expose a more granular ability to set the
permissions on a file instead of just a blanket \"set all permissions\" method.
This function remains behind the `fs` feature.
The following apis are now `#[deprecated]`
* The `TempDir` type is now entirely deprecated and is [located on
crates.io][tempdir] as the `tempdir` crate with [its source][github] at
rust-lang/tempdir.
[tempdir]: https://crates.io/crates/tempdir
[github]: https://github.com/rust-lang/tempdir
The stability of some of these APIs has been questioned over the past few weeks
in using these APIs, and it is intentional that the majority of APIs here are
marked `#[stable]`. The `std::fs` module has a lot of room to grow and the
material is [being tracked in a RFC issue][rfc-issue].
[rfc-issue]: rust-lang/rfcs#939
Closes#22879
[breaking-change]
This module is now superseded by the `std::process` module. This module still
has some room to expand to get quite back up to parity with the `old_io`
version, and there is a [tracking issue][issue] for feature requests as well as
known room for expansion.
[issue]: https://github.com/rust-lang/rfcs/issues/941
[breaking-change]
With this PR in-place constants are handled correctly with respect to debug location assignment.
The PR also adds an (unrelated) test case for debug locations in `extern \"C\"` functions.
Fixes#22432
This separates definitions of struct stat and other typedefs between Android and Linux on ARM (Android has a non-standard one). This makes functions such as `File::metadata()` work correctly and makes one able to check file's size. All tests from std (and also run-pass: stat.rs) now pass on ARM Linux. Fixes#20007.
It had been a source of huge bloat in rustdoc outputs. Of course, we can simply disable compiler docs (as `rustc` generates over 90M of HTML) but this approach fares better even after such decision.
Each directory now has `sidebar-items.js`, which immediately calls `initSidebarItems` with a JSON sidebar data. This file is shared throughout every item in the sidebar. The current item is highlighted via a separate JS snippet (`window.sidebarCurrent`). The JS file is designed to be loaded asynchronously, as the sidebar is rendered before the content and slow sidebar loading blocks the entire rendering. For the minimal accessibility without JS, links to the parent items are left in HTML.
In the future, it might also be possible to integrate crates data with the same fashion: `sidebar-items.js` at the root path will do that. (Currently rustdoc skips writing JS in that case.)
This has a huge impact on the size of rustdoc outputs. Originally it was 326MB uncompressed (37.7MB gzipped, 6.1MB xz compressed); it is 169MB uncompressed (11.9MB gzipped, 5.9MB xz compressed) now. The sidebar JS only takes 10MB uncompressed & 0.3MB gzipped.
The new `io` module has had some time to bake and this commit stabilizes some of
the utilities associated with it. This commit also deprecates a number of
`std::old_io::util` functions and structures.
These items are now `#[stable]`
* `Cursor`
* `Cursor::{new, into_inner, get_ref, get_mut, position, set_position}`
* Implementations of I/O traits for `Cursor<T>`
* Delegating implementations of I/O traits for references and `Box` pointers
* Implementations of I/O traits for primitives like slices and `Vec<T>`
* `ReadExt::bytes`
* `Bytes` (and impls)
* `ReadExt::chain`
* `Chain` (and impls)
* `ReadExt::take` (and impls)
* `BufReadExt::lines`
* `Lines` (and impls)
* `io::copy`
* `io::{empty, Empty}` (and impls)
* `io::{sink, Sink}` (and impls)
* `io::{repeat, Repeat}` (and impls)
These items remain `#[unstable]`
* Core I/O traits. These may want a little bit more time to bake along with the
commonly used methods like `read_to_end`.
* `BufReadExt::split` - this function may be renamed to not conflict with
`SliceExt::split`.
* `Error` - there are a number of questions about its representation,
`ErrorKind`, and usability.
These items are now `#[deprecated]` in `old_io`
* `LimitReader` - use `take` instead
* `NullWriter` - use `io::sink` instead
* `ZeroReader` - use `io::repeat` instead
* `NullReader` - use `io::empty` instead
* `MultiWriter` - use `broadcast` instead
* `ChainedReader` - use `chain` instead
* `TeeReader` - use `tee` instead
* `copy` - use `io::copy` instead
[breaking-change]
This commit is an implementation of [RFC 563][rfc] which adds a new
`cfg(debug_assertions)` directive which is specially recognized and calculated
by the compiler. The flag is turned off at any optimization level greater than 1
and may also be explicitly controlled through the `-C debug-assertions`
flag.
[rfc]: https://github.com/rust-lang/rfcs/pull/563
The `debug_assert!` and `debug_assert_eq!` macros now respect this instead of
the `ndebug` variable and `ndebug` no longer holds any meaning to the standard
library.
Code which was previously relying on `not(ndebug)` to gate expensive code should
be updated to rely on `debug_assertions` instead.
Closes#22492
[breaking-change]
Unstable items used in a macro expansion will now always trigger
stability warnings, *unless* the unstable items are directly inside a
macro marked with `#[allow_internal_unstable]`. IOW, the compiler warns
unless the span of the unstable item is a subspan of the definition of a
macro marked with that attribute.
E.g.
#[allow_internal_unstable]
macro_rules! foo {
($e: expr) => {{
$e;
unstable(); // no warning
only_called_by_foo!();
}}
}
macro_rules! only_called_by_foo {
() => { unstable() } // warning
}
foo!(unstable()) // warning
The unstable inside `foo` is fine, due to the attribute. But the
`unstable` inside `only_called_by_foo` is not, since that macro doesn't
have the attribute, and the `unstable` passed into `foo` is also not
fine since it isn't contained in the macro itself (that is, even though
it is only used directly in the macro).
In the process this makes the stability tracking much more precise,
e.g. previously `println!(\"{}\", unstable())` got no warning, but now it
does. As such, this is a bug fix that may cause [breaking-change]s.
The attribute is definitely feature gated, since it explicitly allows
side-stepping the feature gating system.
---
This updates `thread_local!` macro to use the attribute, since it uses
unstable features internally (initialising a struct with unstable
fields).
According to Apple's [arm64 calling convention](https://developer.apple.com/library/ios/documentation/Xcode/Conceptual/iPhoneOSABIReference/Articles/ARM64FunctionCallingConventions.html#//apple_ref/doc/uid/TP40013702-SW1) varargs always are passed
through stack. Since `open` is actually a vararg function on Darwin,
it means that older declaration caused permissions to be taken from
stack, while passed through register => it set file permissions
to garbage and it was simply impossible to read/delete files after they
were created.
They way this commit handles it is to preserve compatibility with
existing code - it simply creates a shim unsafe function so all existing
callers continue work as nothing happened.
This commit performs a stabilization pass over the `std::fs` module now that
it's had some time to bake. The change was largely just adding `#[stable]` tags,
but there are a few APIs that remain `#[unstable]`.
The following apis are now marked `#[stable]`:
* `std::fs` (the name)
* `File`
* `Metadata`
* `ReadDir`
* `DirEntry`
* `OpenOptions`
* `Permissions`
* `File::{open, create}`
* `File::{sync_all, sync_data}`
* `File::set_len`
* `File::metadata`
* Trait implementations for `File` and `&File`
* `OpenOptions::new`
* `OpenOptions::{read, write, append, truncate, create}`
* `OpenOptions::open` - this function was modified, however, to not attempt to
reject cross-platform openings of directories. This means that some platforms
will succeed in opening a directory and others will fail.
* `Metadata::{is_dir, is_file, len, permissions}`
* `Permissions::{readonly, set_readonly}`
* `Iterator for ReadDir`
* `DirEntry::path`
* `remove_file` - like with `OpenOptions::open`, the extra windows code to
remove a readonly file has been removed. This means that removing a readonly
file will succeed on some platforms but fail on others.
* `metadata`
* `rename`
* `copy`
* `hard_link`
* `soft_link`
* `read_link`
* `create_dir`
* `create_dir_all`
* `remove_dir`
* `remove_dir_all`
* `read_dir`
The following apis remain `#[unstable]`.
* `WalkDir` and `walk` - there are many methods by which a directory walk can be
constructed, and it's unclear whether the current semantics are the right
ones. For example symlinks are not handled super well currently. This is now
behind a new `fs_walk` feature.
* `File::path` - this is an extra abstraction which the standard library
provides on top of what the system offers and it's unclear whether we should
be doing so. This is now behind a new `file_path` feature.
* `Metadata::{accessed, modified}` - we do not currently have a good
abstraction for a moment in time which is what these APIs should likely be
returning, so these remain `#[unstable]` for now. These are now behind a new
`fs_time` feature
* `set_file_times` - like with `Metadata::accessed`, we do not currently have
the appropriate abstraction for the arguments here so this API remains
unstable behind the `fs_time` feature gate.
* `PathExt` - the precise set of methods on this trait may change over time and
some methods may be removed. This API remains unstable behind the `path_ext`
feature gate.
* `set_permissions` - we may wish to expose a more granular ability to set the
permissions on a file instead of just a blanket "set all permissions" method.
This function remains behind the `fs` feature.
The following apis are now `#[deprecated]`
* The `TempDir` type is now entirely deprecated and is [located on
crates.io][tempdir] as the `tempdir` crate with [its source][github] at
rust-lang/tempdir.
[tempdir]: https://crates.io/crates/tempdir
[github]: https://github.com/rust-lang/tempdir
The stability of some of these APIs has been questioned over the past few weeks
in using these APIs, and it is intentional that the majority of APIs here are
marked `#[stable]`. The `std::fs` module has a lot of room to grow and the
material is [being tracked in a RFC issue][rfc-issue].
[rfc-issue]: https://github.com/rust-lang/rfcs/issues/939
[breaking-change]
The two main sub-modules, `c_str` and `os_str`, have now had some time to bake
in the standard library. This commits performs a sweep over the modules adding
various stability tags.
The following APIs are now marked `#[stable]`
* `OsString`
* `OsStr`
* `OsString::from_string`
* `OsString::from_str`
* `OsString::new`
* `OsString::into_string`
* `OsString::push` (renamed from `push_os_str`, added an `AsOsStr` bound)
* various trait implementations for `OsString`
* `OsStr::from_str`
* `OsStr::to_str`
* `OsStr::to_string_lossy`
* `OsStr::to_os_string`
* various trait implementations for `OsStr`
* `CString`
* `CStr`
* `NulError`
* `CString::new` - this API's implementation may change as a result of
rust-lang/rfcs#912 but the usage of `CString::new(thing)` looks like it is
unlikely to change. Additionally, the `IntoBytes` bound is also likely to
change but the set of implementors for the trait will not change (despite the
trait perhaps being renamed).
* `CString::from_vec_unchecked`
* `CString::as_bytes`
* `CString::as_bytes_with_nul`
* `NulError::nul_position`
* `NulError::into_vec`
* `CStr::from_ptr`
* `CStr::as_ptr`
* `CStr::to_bytes`
* `CStr::to_bytes_with_nul`
* various trait implementations for `CStr`
The following APIs remain `#[unstable]`
* `OsStr*Ext` traits remain unstable as the organization of `os::platform` is
uncertain still and the traits may change location.
* `AsOsStr` remains unstable as generic conversion traits are likely to be
rethought soon.
The following APIs were deprecated
* `OsString::push_os_str` is now called `push` and takes `T: AsOsStr` instead (a
superset of the previous functionality).
This commit is an implementation of [RFC 563][rfc] which adds a new
`cfg(debug_assertions)` directive which is specially recognized and calculated
by the compiler. The flag is turned off at any optimization level greater than 1
and may also be explicitly controlled through the `-C debug-assertions`
flag.
[rfc]: https://github.com/rust-lang/rfcs/pull/563
The `debug_assert!` and `debug_assert_eq!` macros now respect this instead of
the `ndebug` variable and `ndebug` no longer holds any meaning to the standard
library.
Code which was previously relying on `not(ndebug)` to gate expensive code should
be updated to rely on `debug_assertions` instead.
Closes#22492
[breaking-change]
This module is now superseded by the `std::process` module. This module still
has some room to expand to get quite back up to parity with the `old_io`
version, and there is a [tracking issue][issue] for feature requests as well as
known room for expansion.
[issue]: https://github.com/rust-lang/rfcs/issues/941
[breaking-change]
The main gist of this PR is commit 1077efb which removes the list of supertraits from the `TraitDef` and pulls them into a separate table, which is accessed via `lookup_super_predicates`. This is analogous to `lookup_predicates`, which gets the complete where clause. This allows us to create the `TraitDef`, which contains the list generics and so forth, without fully knowing the list of supertraits. This in turn allows the *supertrait listing* to contain references to associated types like `<Self as Foo>::Item`, which were previously impossible because conversion required having the `TraitDef` for `Foo`.
We do not yet support `Self::Item` in a supertrait listing. This doesn't work because to convert that, it attempts to expand out the full set of supertraits, which are in the process of being created. This could potentially be worked out by having the expansion of supertraits proceed in a lazy fashion, but we'd have to define shadowing rules for associated types which we don't currently have.
Along the way (in 9de9ec5) I also removed the restriction against duplicate bounds and generalized the code so that it can handle having the same supertrait multiple times with different arguments, e.g. `Foo : Bar<i32> + Bar<u32>`. This restriction was serving no particular purpose, since the same trait could be extended multiple times indirectly, and in the era of multidispatch it is actively harmful.
This is technically a [breaking-change] because it affects the definition of a super-trait. Anything in a where clause that looks like `where Self : Foo` is now considered a supertrait. Because cycles are disallowed in supertraits, that could lead to some errors. This has not been observed in any existing code.
r? @nrc
Unstable items used in a macro expansion will now always trigger
stability warnings, *unless* the unstable items are directly inside a
macro marked with `#[allow_internal_unstable]`. IOW, the compiler warns
unless the span of the unstable item is a subspan of the definition of a
macro marked with that attribute.
E.g.
#[allow_internal_unstable]
macro_rules! foo {
($e: expr) => {{
$e;
unstable(); // no warning
only_called_by_foo!();
}}
}
macro_rules! only_called_by_foo {
() => { unstable() } // warning
}
foo!(unstable()) // warning
The unstable inside `foo` is fine, due to the attribute. But the
`unstable` inside `only_called_by_foo` is not, since that macro doesn't
have the attribute, and the `unstable` passed into `foo` is also not
fine since it isn't contained in the macro itself (that is, even though
it is only used directly in the macro).
In the process this makes the stability tracking much more precise,
e.g. previously `println!("{}", unstable())` got no warning, but now it
does. As such, this is a bug fix that may cause [breaking-change]s.
The attribute is definitely feature gated, since it explicitly allows
side-stepping the feature gating system.
This concretely improves type inference of some cases (see included
test). I assume the compiler struggles to reason about multiple layers
of generic type parameters (even with associated-type equalities) but
*can* understand pure associated types, since they are always directly
computable from the input types.
According to Apple arm64 calling convention varargs always are passed
through stack. Since `open` is actually a vararg function on Darwin's,
it means that older declaration caused permissions to be taken from
stack, while passed through register => it set file permissions
to garbage and it was simply impossible to read/delete files after they
were created.
They way this commit handles it is to preserve compatibility with
existing code - it simply creates a shim unsafe function so all existing
callers continue work as nothing happened.
It had been a source of huge bloat in rustdoc outputs. Of course,
we can simply disable compiler docs (as `rustc` generates over 90M
of HTML) but this approach fares better even after such decision.
Each directory now has `sidebar-items.js`, which immediately calls
`initSidebarItems` with a JSON sidebar data. This file is shared
throughout every item in the sidebar. The current item is
highlighted via a separate JS snippet (`window.sidebarCurrent`).
The JS file is designed to be loaded asynchronously, as the sidebar
is rendered before the content and slow sidebar loading blocks
the entire rendering. For the minimal accessibility without JS,
links to the parent items are left in HTML.
In the future, it might also be possible to integrate crates data
with the same fashion: `sidebar-items.js` at the root path will do
that. (Currently rustdoc skips writing JS in that case.)
This has a huge impact on the size of rustdoc outputs. Originally
it was 326MB uncompressed (37.7MB gzipped, 6.1MB xz compressed);
it is 169MB uncompressed (11.9MB gzipped, 5.9MB xz compressed) now.
The sidebar JS only takes 10MB uncompressed & 0.3MB gzipped.
Automatic has-same-types testing methodology can be found in #22501.
Because most of them don't work with `--pretty=typed`, compile-fail tests were manually audited.
r? @aturon
This same source is being built in the Cargo ecosystem and hence needs to build
on stable Rust as well. This commit places the `no_std` attribute along with the
`no_std` feature behind a `cfg_attr` flag so they are not processed when
compiled on crates.io
This stability attribute was left out by accident and the stability pass has
since picked up the ability to check for this. As a result, crates are currently
getting warnings for implementations of `Index`.
Updates to the bison grammar to account for recent grammar additions and new tests. In particular:
* Support parsing `impl MyTrait for .. { }`
* Support parsing ExprQualifiedPaths without \"as TRAIT_REF\" such as `<Foo>::bar(&Foo)`
* Support parsing \"for\" clauses at the beginning of where clauses such as `where for<'a, 'b> &'a T: Bar<'b>`
Currently, the list of files linted in `tidy.py` is unordered. It seems more appropriate for more frequently appearing files (like `.rs`) to appear at the top of the list and for \"other files\" to appear at the very end. This PR also changes the wildcard import of `check_license()` into an explicit one.
```
Before: After:
* linted 4 .sh files * linted 5034 .rs files
* linted 4 .h files * linted 29 .c files
* linted 29 .c files * linted 28 .py files
* linted 2 .js files * linted 4 .sh files
* linted 0 other files * linted 4 .h files
* linted 28 .py files * linted 2 .js files
* linted 5034 .rs files * linted 0 other files
```
r? @brson
The new `io` module has had some time to bake and this commit stabilizes some of
the utilities associated with it. This commit also deprecates a number of
`std::old_io::util` functions and structures.
These items are now `#[stable]`
* `Cursor`
* `Cursor::{new, into_inner, get_ref, get_mut, position, set_position}`
* Implementations of I/O traits for `Cursor<T>`
* Delegating implementations of I/O traits for references and `Box` pointers
* Implementations of I/O traits for primitives like slices and `Vec<T>`
* `ReadExt::bytes`
* `Bytes` (and impls)
* `ReadExt::chain`
* `Chain` (and impls)
* `ReadExt::take` (and impls)
* `BufReadExt::lines`
* `Lines` (and impls)
* `io::copy`
* `io::{empty, Empty}` (and impls)
* `io::{sink, Sink}` (and impls)
* `io::{repeat, Repeat}` (and impls)
These items remain `#[unstable]`
* Core I/O traits. These may want a little bit more time to bake along with the
commonly used methods like `read_to_end`.
* `BufReadExt::split` - this function may be renamed to not conflict with
`SliceExt::split`.
* `Error` - there are a number of questions about its representation,
`ErrorKind`, and usability.
These items are now `#[deprecated]` in `old_io`
* `LimitReader` - use `take` instead
* `NullWriter` - use `io::sink` instead
* `ZeroReader` - use `io::repeat` instead
* `NullReader` - use `io::empty` instead
* `MultiWriter` - use `broadcast` instead
* `ChainedReader` - use `chain` instead
* `TeeReader` - use `tee` instead
* `copy` - use `io::copy` instead
[breaking-change]
This commit deprecates the majority of std::old_io::fs in favor of std::fs and
its new functionality. Some functions remain non-deprecated but are now behind a
feature gate called `old_fs`. These functions will be deprecated once
suitable replacements have been implemented.
The compiler has been migrated to new `std::fs` and `std::path` APIs where
appropriate as part of this change.
us to construct trait-references and do other things without forcing a
full evaluation of the supertraits. One downside of this scheme is that
we must invoke `ensure_super_predicates` before using any construct that
might require knowing about the super-predicates.
This allows to create proper debuginfo line information for items inlined from other crates (e.g. instantiations of generics). Only the codemap's 'metadata' is stored in a crate's metadata. That is, just filename, positions of line-beginnings, etc. but not the actual source code itself.
Crate metadata size is increased by this change because spans in the encoded ASTs take up space now:
```
BEFORE AFTER
libcore 36 MiB 39.6 MiB +10%
libsyntax 51.1 MiB 60.5 MiB +18.4%
libcollections 11.2 MiB 12.8 MiB +14.3%
```
This only affects binaries containing metadata (rlibs and dylibs), executables should not be affected in size.
Fixes#19228 and probably #22226.
This allows to create proper debuginfo line information for items inlined from other crates (e.g. instantiations of generics).
Only the codemap's 'metadata' is stored in a crate's metadata. That is, just filename, line-beginnings, etc. but not the actual source code itself. We are thus missing the opportunity of making Rust the first "open-source-only" programming language out there. Pity.
The failing concurrency example was doing something different from the
working example. This commit changes just enough of the failing example
to (1) still fail with the same error, (2) tries to do the same as the
working example (increment a vector value and print it).
r? @steveklabnik
This same source is being built in the Cargo ecosystem and hence needs to build
on stable Rust as well. This commit places the `no_std` attribute along with the
`no_std` feature behind a `cfg_attr` flag so they are not processed when
compiled on crates.io
Many of the modifications putting in `Box::new` calls also include a
pointer to Issue 22405, which tracks going back to `box <expr>` if
possible in the future.
(Still tried to use `Box<_>` where it sufficed; thus some tests still
have `box_syntax` enabled, as they use a mix of `box` and `Box::new`.)
Precursor for overloaded-`box` and placement-`in`; see Issue 22181.
This is the kind of change that one is expected to need to make to
accommodate overloaded-`box`.
----
Note that this is not *all* of the changes necessary to accommodate
Issue 22181. It is merely the subset of those cases where there was
already a let-binding in place that made it easy to add the necesasry
type ascription.
(For unnamed intermediate `Box` values, one must go down a different
route; `Box::new` is the option that maximizes portability, but has
potential inefficiency depending on whether the call is inlined.)
----
There is one place worth note, `run-pass/coerce-match.rs`, where I
used an ugly form of `Box<_>` type ascription where I would have
preferred to use `Box::new` to accommodate overloaded-`box`. I
deliberately did not use `Box::new` here, because that is already done
in coerce-match-calls.rs.
----
Precursor for overloaded-`box` and placement-`in`; see Issue 22181.
This stability attribute was left out by accident and the stability pass has
since picked up the ability to check for this. As a result, crates are currently
getting warnings for implementations of `Index`.
Rebase and follow-through on work done by @cmr and @aatch.
Implements most of rust-lang/rfcs#560. Errors encountered from the checks during building were fixed.
The checks for division, remainder and bit-shifting have not been implemented yet.
See also PR #20795
cc @Aatch ; cc @nikomatsakis
Make `test/run-pass/backtrace.rs` more robust about own host environment
Namely, I have been annoyed in the past when I have done `RUST_BACKTRACE=1 make check` only to discover (again) that such a trick causes this test to fail, because it assumes that the `RUST_BACKTRACE` environment variable is not set.
Fix#22870
aarch64-linux-android build has been broken since #22839.
Aarch64 android has _Unwind_GetIPInfo, so re-define this only for arm32 android.
r? @alexcrichton
This changes the type of some public constants/statics in libunicode.
Notably some `&'static &'static [(char, char)]` have changed
to `&'static [(char, char)]`. The regexp crate seems to be the
sole user of these, yet this is technically a [breaking-change]
Associated types are now treated as part of the public API by the privacy checker.
If you were exposing a private type in your public API via an associated type, make that type public:
``` diff
pub struct PublicType { .. }
- struct Struct { .. }
+ pub struct Struct { .. }
pub trait PublicTrait {
type Output;
fn foo(&self) -> Self::Output;
}
impl PublicTrait for PublicType {
type Output = Struct;
fn foo(&self) -> Struct { // `Struct` is part of the public API, it must be marked as `pub`lic
..
}
}
```
[breaking-change]
---
r? @nikomatsakis
closes#22912
Fix the return type in the comments.
An old commit 082bfde412 (\"Fallout of std::str stabilization\") removed
the example of FromStr::from_str(), this commit adds it back. But
the example of StrExt::parse() is still kept with an additinal note.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
The API this exposes is a little strange (being attached to `static`s),
so it makes sense to conservatively feature gate it. If it is highly
popular, it is possible to reverse this gating.
* The lint visitor's visit_ty method did not recurse, and had a
reference to the now closed#10894
* The newly enabled recursion has only affected the `deprectated` lint
which now detects uses of deprecated items in trait impls and
function return types
* Renamed some references to `CowString` and `CowVec` to `Cow<str>` and
`Cow<[T]>`, respectively, which appear outside of the crate which
defines them
* Replaced a few instances of `InvariantType<T>` with
`PhantomData<Cell<T>>`
* Disabled the `deprecated` lint in several places that
reference/implement traits on deprecated items which will get cleaned
up in the future
* Unfortunately, this means that if a library declares
`#![deny(deprecated)]` and marks anything as deprecated, it will have
to disable the lint for any uses of said item, e.g. any impl the now
deprecated item
For any library that denies deprecated items but has deprecated items
of its own, this is a [breaking-change]
I had originally intended for the lint to ignore uses of deprecated items that are declared in the same crate, but this goes against some previous test cases that expect the lint to capture *all* uses of deprecated items, so I maintained the previous approach to avoid changing the expected behavior of the lint.
Tested locally on OS X, so hopefully there aren't any deprecated item uses behind a `cfg` that I may have missed.
* Make num::UpperHex private. I was unable to determine why this struct
is public. The num module itself is not public, and the UpperHex struct
is not referenced anywhere in the core::fmt module. (Only the UpperHex
trait is reference.) num::LowerHex is not public.
* Remove the suffix parameters from the macros that generate integral
display traits.
The code to print the Debug::fmt suffixes was removed when Show was
renamed to Debug. It was an intentional change. From RFC 0565:
* Focus on the *runtime* aspects of a type; repeating information such
as suffixes for integer literals is not generally useful since that
data is readily available from the type definition.
* Because Show was renamed to Debug, rename show! to debug!.
However, I left the ICEs rather than switching to nicer error
reporting there; these cases *should* be detected prior to hitting
trans, and thus ICE'ing here is appropriate.
(So why switch to `eval_const_expr_partial`? Because I want to try to
eliminate all uses of `eval_const_expr` entirely.)
1. Detect and report arithmetic overflow during const-expr eval.
2. Instead `eval_const_expr_partial` returning `Err(String)`, it now
has a dedicated enum of different cases. The main benefit of this
is the ability to pass along an interpretable payload, namely the
two inputs that caused an overlfow.
I attempted to minimize fallout to error output in tests, but some was
unavoidable. Those changes are in a follow-on commit.
This allows computation to proceed and find further errors.
(However, this is also annoying at times when the subsequent errors
are just reporting that a ty_err occurred. I have thoughts on ways to
fix this that I will experiment with separately.)
This only replaces the conditional arith-overflow asserts with
unconditional errors from the guts of const-eval; it does *not*
attempt to sanely handle such errors e.g. with a nice error message
from `rustc`. So the same test that led me to add this commit are
still failing, and must be addressed.
(The bug was in `impl RandomAccessIterator for Rev`; it may or may not
have been innocuous, depending on what guarantees one has about the
behavior of `idx` for the underlying iterator.)
Regarding the `rand` changes: It is unfortunate that Wrapping(T) does
not support the `+=` operator. We may want to try to fix that before
1.0 to make porting code like this palatable.
Regarding `std::rand`, just arith-overflow in first example from
`std::rand::random()` doc.
* `core::num`: adjust `UnsignedInt::is_power_of_two`,
`UnsignedInt::next_power_of_two`, `Int::pow`.
In particular for `Int::pow`: (1.) do not panic when `base`
overflows if `acc` never observes the overflowed `base`, and (2.)
if `acc` does observe the overflowed `base`, make sure we only
panic if we would have otherwise (e.g. during a computation of
`base * base`).
* also in `core::num`: avoid underflow during computation of `uint::MAX`.
* `std::num`: adjust tests `uint::test_uint_from_str_overflow`,
`uint::test_uint_to_str_overflow`, `strconv`
* `coretest::num`: adjust `test::test_int_from_str_overflow`.
* The error patterns had a typo.
* Our current constant evaluation would silently allow the overflow
(filed as Issue 22531).
* The overflowing-mul test was accidentally doing addition instead of
multiplication.
* `collections::btree::node`: accommodate (transient) underflow.
* `collections::btree::map`: avoid underflow during `fn next`
for `BTreeMap::range` methods.
* `collections::slice`: note that pnkfelix deliberately used
`new_pos_wrapping` only once; the other cases of arithmetic do not
over- nor underflow, which is a useful property to leave implicitly
checked/documented via the remaining calls to `fn new_pos(..)`.
* `collections::vec_deque` applied wrapping ops (somewhat blindly)
to two implementation methods, and many tests.
* `std::collections:#️⃣:table` : Use `OverflowingOps` trait to
track overflow during `calculate_offsets` and `calculate_allocation`
functions.
These return the result of the operation *plus* an overflow/underflow bit.
This can make it easier to write operations where you want to chain
some arithmetic together, but also want to return a flag signalling if
overflow every occurred.
During my clean-up of rebase errors, I took the opportunity to implement
parse_opt_bool so that it isn't identical to parse_bool wrapped in
`Some`.
parse_opt_bool considers no value to be true, a value of 'y', 'yes' or
'on' to be true and 'n', 'no' or 'off' to be false. All other values are
an error.
Many of the core rust libraries have places that rely on integer
wrapping behaviour. These places have been altered to use the wrapping_*
methods:
* core:#️⃣:sip - A number of macros
* core::str - The `maximal_suffix` method in `TwoWaySearcher`
* rustc::util::nodemap - Implementation of FnvHash
* rustc_back::sha2 - A number of macros and other places
* rand::isaac - Isaac64Rng, changed to use the Wrapping helper type
Some places had "benign" underflow. This is when underflow or overflow
occurs, but the unspecified value is not used due to other conditions.
* collections::bit::Bitv - underflow when `self.nbits` is zero.
* collections:#️⃣:{map,table} - Underflow when searching an empty
table. Did cause undefined behaviour in this case due to an
out-of-bounds ptr::offset based on the underflowed index. However the
resulting pointers would never be read from.
* syntax::ext::deriving::encodable - Underflow when calculating the
index of the last field in a variant with no fields.
These cases were altered to avoid the underflow, often by moving the
underflowing operation to a place where underflow could not happen.
There was one case that relied on the fact that unsigned arithmetic and
two's complement arithmetic are identical with wrapping semantics. This
was changed to use the wrapping_* methods.
Finally, the calculation of variant discriminants could overflow if the
preceeding discriminant was `U64_MAX`. The logic in `rustc::middle::ty`
for this was altered to avoid the overflow completely, while the
remaining places were changed to use wrapping methods. This is because
`rustc::middle::ty::enum_variants` now throws an error when the
calculated discriminant value overflows a `u64`.
This behaviour can be triggered by the following code:
```
enum Foo {
A = U64_MAX,
B
}
```
This commit also implements the remaining integer operators for
Wrapped<T>.
Adds overflow checking to integer addition, multiplication, and subtraction
when `-Z force-overflow-checks` is true, or if `--cfg ndebug` is not passed to
the compiler. On overflow, it panics with `arithmetic operation overflowed`.
Also adds `overflowing_add`, `overflowing_sub`, and `overflowing_mul`
intrinsics for doing unchecked arithmetic.
[breaking-change]
Namely, I have been annoyed in the past when I have done
`RUST_BACKTRACE=1 make check` only to discover (again) that such a
trick causes this test to fail, because it assumes that the
`RUST_BACKTRACE` environment variable is not set.
This is a series of individual but correlated changes to the metadata format. The changes are significant enough that it (finally) bumps the metadata encoding version. In brief, they altogether reduce the total size of stage1 binaries by 27% (!!!!). Almost every low-hanging fruit has been considered and fixed; see the individual commits for details.
Detailed library (not just metadata) size changes for x86_64-unknown-linux-gnu stage1 binaries (baseline being 3a96d6a981):
````
before after delta path
--------- --------- ------ --------------------------------
1706146 1050412 38.4% liballoc-4e7c5e5c.rlib
398576 152454 61.8% libarena-4e7c5e5c.rlib
71441 56892 20.4% libarena-4e7c5e5c.so
14424754 5084102 64.8% libcollections-4e7c5e5c.rlib
39143186 14743118 62.3% libcore-4e7c5e5c.rlib
195574 188150 3.8% libflate-4e7c5e5c.rlib
153123 152603 0.3% libflate-4e7c5e5c.so
477152 215262 54.9% libfmt_macros-4e7c5e5c.rlib
77728 66601 14.3% libfmt_macros-4e7c5e5c.so
1216936 684104 43.8% libgetopts-4e7c5e5c.rlib
207846 181116 12.9% libgetopts-4e7c5e5c.so
349722 147530 57.8% libgraphviz-4e7c5e5c.rlib
60196 49197 18.3% libgraphviz-4e7c5e5c.so
729842 259906 64.4% liblibc-4e7c5e5c.rlib
349358 247014 29.3% liblog-4e7c5e5c.rlib
88878 83163 6.4% liblog-4e7c5e5c.so
1968508 732840 62.8% librand-4e7c5e5c.rlib
1968204 696326 64.6% librbml-4e7c5e5c.rlib
283207 206589 27.1% librbml-4e7c5e5c.so
72369394 46401230 35.9% librustc-4e7c5e5c.rlib
11941372 10498483 12.1% librustc-4e7c5e5c.so
2717894 1983272 27.0% librustc_back-4e7c5e5c.rlib
501900 464176 7.5% librustc_back-4e7c5e5c.so
15058 12588 16.4% librustc_bitflags-4e7c5e5c.rlib
4008268 2961912 26.1% librustc_borrowck-4e7c5e5c.rlib
837550 785633 6.2% librustc_borrowck-4e7c5e5c.so
6473348 6095470 5.8% librustc_driver-4e7c5e5c.rlib
1448785 1433945 1.0% librustc_driver-4e7c5e5c.so
95483688 94779704 0.7% librustc_llvm-4e7c5e5c.rlib
43516815 43487809 0.1% librustc_llvm-4e7c5e5c.so
938140 817236 12.9% librustc_privacy-4e7c5e5c.rlib
182653 176563 3.3% librustc_privacy-4e7c5e5c.so
4390288 3543284 19.3% librustc_resolve-4e7c5e5c.rlib
872981 831824 4.7% librustc_resolve-4e7c5e5c.so
18176426 14795426 18.6% librustc_trans-4e7c5e5c.rlib
3657354 3480026 4.8% librustc_trans-4e7c5e5c.so
16815076 13868862 17.5% librustc_typeck-4e7c5e5c.rlib
3274439 3123898 4.6% librustc_typeck-4e7c5e5c.so
21372308 14890582 30.3% librustdoc-4e7c5e5c.rlib
4501971 4172202 7.3% librustdoc-4e7c5e5c.so
8055028 2951044 63.4% libserialize-4e7c5e5c.rlib
958101 710016 25.9% libserialize-4e7c5e5c.so
30810208 15160648 50.8% libstd-4e7c5e5c.rlib
6819003 5967485 12.5% libstd-4e7c5e5c.so
58850950 31949594 45.7% libsyntax-4e7c5e5c.rlib
9060154 7882423 13.0% libsyntax-4e7c5e5c.so
1474310 1062102 28.0% libterm-4e7c5e5c.rlib
345577 323952 6.3% libterm-4e7c5e5c.so
2827854 1643056 41.9% libtest-4e7c5e5c.rlib
517811 452519 12.6% libtest-4e7c5e5c.so
2274106 1761240 22.6% libunicode-4e7c5e5c.rlib
--------- --------- ------ --------------------------------
499359187 363465583 27.2% total
````
Some notes:
* Uncompressed metadata compacts very well. It is less visible for compressed metadata but still it achieves about 5~10% reduction.
* *Every* commit is designed to reduce the metadata in one way. There is absolutely no negative impact associated to changes (that's why the table above doesn't contain a minus delta).
* I've confirmed that this compiles through `make all`, making it almost correct. Other platforms have to be tested though.
* Oh, I'll rebase this as soon as I have spare time, but I guess this needs an extensive review anyway.
* I haven't rigorously checked the encoder and decoder performance. I tried to minimize the impact (some encodings are actually simpler than the original), but I'm not sure.
Fixes#2743, #9303 (partially) and #21482.
Fix the return type in the comments.
An old commit 082bfde412 ("Fallout of std::str stabilization") removed
the example of FromStr::from_str(), this commit adds it back. But
the example of StrExt::parse() is still kept with an additinal note.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Previously every auto-serialized tags are strongly typed. However
this is not strictly required, and instead it can be exploited
to provide the optimal encoding for smaller integers. This commit
repurposes `EsI8`/`EsU8` through `EsI64`/`EsU64` tags to represent
*any* integers with given ranges: It is now possible to encode
`42u64` as two bytes `EsU8 0x2a`, for example.
There are some limitations:
* It does not apply to non-auto-serialized tags for obvious reasons.
Fortunately, we have already eliminated the biggest source of
such tag in favor of auto-serialized tags: `tag_table_id`.
* Bigger tags cannot be used to represent smaller types.
* Signed tags and unsigned tags do not mix.
This avoids a biggish eight-byte `tag_table_id` tag in favor of
autoserialized integer tags, which are smaller and can be later
used to encode them in the optimal number of bytes. `NodeId` was
u32 after all.
Previously:
<------------- len1 -------------->
tag_table_* <len1> tag_table_id 88 <nodeid in 8 bytes>
tag_table_val <len2> <actual data>
<-- len2 --->
Now:
<--------------- len --------------->
tag_table_* <len> U32 <nodeid in 4 bytes> <actual data>
We try to move the data when the length can be encoded in
the much smaller number of bytes. This interferes with indices and
type abbreviations however, so this commit introduces a public
interface to get and mark a "stable" (i.e. not affected by
relaxation) position of the current pointer.
The relaxation logic only moves a small data, currently at most
256 bytes, as moving the data can be costly. There might be
further opportunities to allow more relaxation by moving fields
around, which I didn't seriously try.
They replace the existing `EsEnumVid`, `EsVecLen` and `EsMapLen`
tags altogether; the meaning of them can be easily inferred
from the enclosing tag. It also has an added benefit of
encodings for smaller variant ids or lengths being more compact
(5 bytes to 2 bytes).
For the reference, while it is designed to be selectively enabled,
it was essentially enabled throughout every snapshot and nightly
as far as I can tell. This makes the usefulness of `EsLabel` itself
questionable, as it was quite rare that `EsLabel` broke the build.
It had consumed about 20~30% of metadata (!) and so this should be
a huge win.
It doesn't serve any useful purpose. It *might* be useful when
there are some tags that are generated by `Encodable` and
not delimited by any tags, but IIUC it's not the case.
Previous:
<-------------------- len1 ------------------->
EsEnum <len1> EsEnumVid <vid> EsEnumBody <len2> <arg1> <arg2>
<--- len2 -->
Now:
<----------- len1 ---------->
EsEnum <len1> EsEnumVid <vid> <arg1> <arg2>
Many auto-serialization tags are fixed-size (note: many ordinary
tags are also fixed-size but for now this commit ignores them),
so having an explicit length is a waste. This moves any
auto-serialization tags with an implicit length before other tags,
so a test for them is easy. A preliminary experiment shows this
has at least 1% gain over the status quo.
EBML tags are encoded in a variable-length unsigned int (vuint),
which is clever but causes some tags to be encoded in two bytes
while there are really about 180 tags or so. Assuming that there
wouldn't be, say, over 1,000 tags in the future, we can use much
more efficient encoding scheme. The new scheme should support
at most 4,096 tags anyway.
This also flattens a scattered tag namespace (did you know that
0xa9 is followed by 0xb0?) and makes a room for autoserialized tags
in 0x00 through 0x1f.
They are, with a conjunction of `start_tag` and `end_tag`, commonly
used to write a document with a binary data of known size. However
the use of `start_tag` makes the length always 4 bytes long, which
is almost not optimal (requiring the relaxation step to remedy).
Directly using `wr_tagged_*` methods is better for both readability
and resulting metadata size.
* count_ones/zeros, trailing_ones/zeros return u32, not usize
* rotate_left/right take u32, not usize
* RADIX, MANTISSA_DIGITS, DIGITS, BITS, BYTES are u32, not usize
Doesn't touch pow because there's another PR for it.
cc https://github.com/rust-lang/rust/issues/22240
r? @Gankro
The API this exposes is a little strange (being attached to `static`s),
so it makes sense to conservatively feature gate it. If it is highly
popular, it is possible to reverse this gating.
* count_ones/zeros, trailing_ones/zeros return u32, not usize
* rotate_left/right take u32, not usize
* RADIX, MANTISSA_DIGITS, DIGITS, BITS, BYTES are u32, not usize
Doesn't touch pow because there's another PR for it.
[breaking-change]
* The lint visitor's visit_ty method did not recurse, and had a
reference to the now closed#10894
* The newly enabled recursion has only affected the `deprectated` lint
which now detects uses of deprecated items in trait impls and
function return types
* Renamed some references to `CowString` and `CowVec` to `Cow<str>` and
`Cow<[T]>`, respectively, which appear outside of the crate which
defines them
* Replaced a few instances of `InvariantType<T>` with
`PhantomData<Cell<T>>`
* Disabled the `deprecated` lint in several places that
reference/implement traits on deprecated items which will get cleaned
up in the future
* Disabled the `exceeding_bitshifts` lint for
compile-fail/huge-array-simple test so it doesn't shadow the expected
error on 32bit systems
* Unfortunately, this means that if a library declares
`#![deny(deprecated)]` and marks anything as deprecated, it will have
to disable the lint for any uses of said item, e.g. any impl the now
deprecated item
For any library that denies deprecated items but has deprecated items
of its own, this is a [breaking-change]
This commits blanket marks the API of the `std::process` module as `#[stable]`.
The module's API is very similar to the old `std::old_io::process` API and has
generally had quite a bit of time to bake both before and after the new module
landed.
Remove the synthetic \"region bound\" from closures and instead update how
type-outlives works for closure types so that it ensures that all upvars
outlive the region in question. This gives the same guarantees but
without introducing artificial regions (and gives better error messages
to boot). This is refactoring towards #3696.
r? @pnkfelix
Namely, I have been annoyed in the past when I have done
`RUST_BACKTRACE=1 make check` only to discover (again) that such a
trick causes this test to fail, because it assumes that the
`RUST_BACKTRACE` environment variable is not set.
Fix#22870
This changes the type of some public constants/statics in libunicode.
Notably some `&'static &'static [(char, char)]` have changed
to `&'static [(char, char)]`. The regexp crate seems to be the
sole user of these, yet this is technically a [breaking-change]
type-outlives works for closure types so that it ensures that all upvars
outlive the region in question. This gives the same guarantees but
without introducing artificial regions (and gives better error messages
to boot).
This is an implementation of RFC 899 and adds stdio functionality to the new
`std::io` module. Details of the API can be found on the RFC, but from a high
level:
* `io::{stdin, stdout, stderr}` constructors are now available. There are also
`*_raw` variants for unbuffered and unlocked access.
* All handles are globally shared (excluding raw variants).
* The stderr handle is no longer buffered.
* All handles can be explicitly locked (excluding the raw variants).
The `print!` and `println!` machinery has not yet been hooked up to these
streams just yet. The `std::fmt::output` module has also not yet been
implemented as part of this commit.
This commits blanket marks the API of the `std::process` module as `#[stable]`.
The module's API is very similar to the old `std::old_io::process` API and has
generally had quite a bit of time to bake both before and after the new module
landed.
The one modification made to the API is that `Stdio::capture` is now named
`stdio::piped`.
[breaking-change]
... objects
For method calls through trait objects, we currently generate the llvm
function argument attributes using the non-opaque method signature that
still has the trait object fat pointer for the self pointer. This leads
to attributes that are plain wrong, e.g. noalias. As we don't know
anything about the concrete type of the underlying object, we must
replace the self argument with an opaque i8 pointer before applying the
attributes.
this is the same problem as openbsd (#22792).
without the patch, liblibc don't build.
@mneumann please comment.
I have encountered this problem while building some rust libs with `target=x86_64-unknown-dragonfly` (while working on #22794)
Check for unbounded recursion during dropck.
Such recursion can be introduced by the erroneous use of non-regular types (aka types employing polymorphic recursion), which Rust does not support.
Fix#22443
