This lets the C++ code in the rt handle the (slightly) tricky parts of
random number generation: e.g. error detection/handling, and using the
values of the `#define`d options to the various functions.
It now:
- can be explicitly seeded from user code (`seed_task_rng`) or from the
environment (`RUST_SEED`, a positive integer)
- automatically reseeds itself from the OS *unless* it was seeded by
either method above
- has more documentation
This provides 2 methods: .reseed() and ::from_seed that modify and
create respecitively.
Implement this trait for the RNGs in the stdlib for which this makes
sense.
This is implemented as a wrapper around another RNG. It is designed
to allow the actual implementation to be changed without changing
the external API (e.g. it currently uses a 64-bit generator on 64-
bit platforms, and a 32-bit one on 32-bit platforms; but one could
imagine that the IsaacRng may be deprecated later, and having this
ability to switch algorithms without having to update the points of
use is convenient.)
This is the recommended general use RNG.
The former reads from e.g. /dev/urandom, the latter just wraps any
std::rt::io::Reader into an interface that implements Rng.
This also adds Rng.fill_bytes for efficient implementations of the above
(reading 8 bytes at a time is inefficient when you can read 1000), and
removes the dependence on src/rt (i.e. rand_gen_seed) although this last
one requires implementing hand-seeding of the XorShiftRng used in the
scheduler on Linux/unixes, since OSRng relies on a scheduler existing to
be able to read from /dev/urandom.
This is 2x faster on 64-bit computers at generating anything larger
than 32-bits.
It has been verified against the canonical C implementation from the
website of the creator of ISAAC64.
Also, move `Rng.next` to `Rng.next_u32` and add `Rng.next_u64` to
take full advantage of the wider word width; otherwise Isaac64 will
always be squeezed down into a u32 wasting half the entropy and
offering no advantage over the 32-bit variant.
This commit fixes all of the fallout of the previous commit which is an attempt
to refine privacy. There were a few unfortunate leaks which now must be plugged,
and the most horrible one is the current `shouldnt_be_public` module now inside
`std::rt`. I think that this either needs a slight reorganization of the
runtime, or otherwise it needs to just wait for the external users of these
modules to get replaced with their `rt` implementations.
Other fixes involve making things pub which should be pub, and otherwise
updating error messages that now reference privacy instead of referencing an
"unresolved name" (yay!).
This implements the necessary logic for gating particular features off by default in the compiler. There are a number of issues which have been wanting this form of mechanism, and this initially gates features which we have open issues for.
Additionally, this should unblock #9255
This pull request changes to memory layout of the `CrateMap` struct to use static slices instead of raw pointers. Most of the discussion took place [here](63b5975efa (L1R92)) .
The memory layout of CrateMap changed, without bumping the version number in the struct. Another, more backward compatible, solution would be to keep the old code and increase the version number in the new struct. On the other hand, the `annihilate_fn` pointer was removed without bumping the version number recently.
At the moment, the stage0 compiler does not use the new memory layout, which would lead the segfaults during stage0 compilation, so I've added a dummy `iter_crate_map` function for stage0, which does nothing. Again, this could be avoided if we'd bump the version number in the struct and keep the old code.
I'd like to use a normal `for` loop [here](https://github.com/fhahn/rust/compare/logging-unsafe-removal?expand=1#L1R109),
for child in children.iter() {
do_iter_crate_map(child, |x| f(x), visited);
}
but for some reason this only yields `error: unresolved enum variant, struct or const 'Some'` and I have no idea why.
A few features are now hidden behind various #[feature(...)] directives. These
include struct-like enum variants, glob imports, and macro_rules! invocations.
Closes#9304Closes#9305Closes#9306Closes#9331
The root issue is that dlerror isn't reentrant or even thread safe.
The solution implemented here is to make a yielding spin lock over an
AtomicFlag. This is pretty hacky, but the best we can do at this point.
As far as I can tell, it isn't possible to create a global mutex without
having to initialize it in a single threaded context.
The Windows code isn't affected since errno is thread-local on Windows
and it's running in an atomically block to ensure there isn't a green
thread context switch.
Closes#8156
This PR solves one of the pain points with c-style enums. Simplifies writing a fn to convert from an int/uint to an enum. It does this through a `#[deriving(FromPrimitive)]` syntax extension.
Before this is committed though, we need to discuss if `ToPrimitive`/`FromPrimitive` has the right design (cc #4819). I've changed all the `.to_int()` and `from_int()` style functions to return `Option<int>` so we can handle partial functions. For this PR though only enums and `extra::num::bigint::*` take advantage of returning None for unrepresentable values. In the long run it'd be better if `i64.to_i8()` returned `None` if the value was too large, but I'll save this for a future PR.
Closes#3868.
The root issue is that dlerror isn't reentrant or even thread safe.
The Windows code isn't affected since errno is thread-local on Windows
and it's running in an atomically block to ensure there isn't a green
thread context switch.
Closes#8156
This adds a large doc-block to the top of the std::logging module explaining how
to use it. This is mostly just making sure that all the information in the
manual's section about logging is also here (in case someone decides to look
into this module first).
This also removes the old console_{on,off} methods. As far as I can tell, the
functions were only used by the compiler, and there's no reason for them to be
used because they're all turned off by default anyway (maybe they were turned on
by default at some point...)
I believe that this is the final nail in the coffin and closes#5021
The general idea of hyperlinking between crates is that it should require as
little configuration as possible, if any at all. In this vein, there are two
separate ways to generate hyperlinks between crates:
1. When you're generating documentation for a crate 'foo' into folder 'doc',
then if foo's external crate dependencies already have documented in the
folder 'doc', then hyperlinks will be generated. This will work because all
documentation is in the same folder, allowing links to work seamlessly both
on the web and on the local filesystem browser.
The rationale for this use case is a package with multiple libraries/crates
that all want to link to one another, and you don't want to have to deal with
going to the web. In theory this could be extended to have a RUST_PATH-style
searching situtation, but I'm not sure that it would work seamlessly on the
web as it does on the local filesystem, so I'm not attempting to explore this
case in this pull request. I believe to fully realize this potential rustdoc
would have to be acting as a server instead of a static site generator.
2. One of foo's external dependencies has a #[doc(html_root_url = "...")]
attribute. This means that all hyperlinks to the dependency will be rooted at
this url.
This use case encompasses all packages using libstd/libextra. These two
crates now have this attribute encoded (currently at the /doc/master url) and
will be read by anything which has a dependency on libstd/libextra. This
should also work for arbitrary crates in the wild that have online
documentation. I don't like how the version is hard-wired into the url, but I
think that this may be a case-by-case thing which doesn't end up being too
bad in the long run.
Closes#9539
One downside with this current implementation is that since BigInt's
default is now 64 bit, we can convert larger BigInt's to a primitive,
however the current implementation on 32 bit architectures does not
take advantage of this fact.
It is simply defined as `f64` across every platform right now.
A use case hasn't been presented for a `float` type defined as the
highest precision floating point type implemented in hardware on the
platform. Performance-wise, using the smallest precision correct for the
use case greatly saves on cache space and allows for fitting more
numbers into SSE/AVX registers.
If there was a use case, this could be implemented as simply a type
alias or a struct thanks to `#[cfg(...)]`.
Closes#6592
The mailing list thread, for reference:
https://mail.mozilla.org/pipermail/rust-dev/2013-July/004632.html
It was a little ambiguous before how explicitl positional parameters and
implicit positional parameters intermingled, and this clarifies how the two
intermingle. This also updates a little bit of documentation/code examples
elsewhere as well.
It was a little ambiguous before how explicitl positional parameters and
implicit positional parameters intermingled, and this clarifies how the two
intermingle. This also updates a little bit of documentation/code examples
elsewhere as well.
Creates a wrapper around a mutable reference to the iterator.
This is useful to allow applying iterator adaptors while still
retaining ownership of the original iterator value.
Example::
let mut xs = range(0, 10);
// sum the first five values
let partial_sum = xs.by_ref().take(5).fold(0, |a, b| a + b);
assert!(partial_sum == 10);
// xs.next() is now `5`
assert!(xs.next() == Some(5));
std::vec: Sane implementations for connect_vec and concat_vec
Avoid unnecessary copying of subvectors, and calculate the needed space
beforehand. These implementations are simple but better than the
previous.
Also only implement it once, for all `Vector<T>` using:
impl<'self, T: Clone, V: Vector<T>> VectorVector<T> for &'self [V]
Closes#9581
Avoid unnecessary copying of subvectors, and calculate the needed space
beforehand. These implementations are simple but better than the
previous.
Also only implement it once, for all `Vector<T>` using:
impl<'self, T: Clone, V: Vector<T>> VectorVector<T> for &'self [V]
performance improved according to the bench test:
before
test vec::bench::concat ... bench: 74818 ns/iter (+/- 408)
test vec::bench::connect ... bench: 87066 ns/iter (+/- 376)
after
test vec::bench::concat ... bench: 17724 ns/iter (+/- 126)
test vec::bench::connect ... bench: 18353 ns/iter (+/- 691)
Closes#9581
std::vec: Use a valid value as lifetime dummy in iterator
The current implementation uses `&v[0]` for the lifetime struct field,
but that is a dangling pointer for iterators derived from zero-length
slices.
Example:
let v: [int, ..0] = []; println!("{:?}", v.iter())
std::vec::VecIterator<,int>{ptr: (0x7f3768626100 as *()), end: (0x7f3768626100 as *()), lifetime: &139875951207128}
To replace this parameter, use a field of type `Option<&'self ()>`
that is simply initialized with `None`, but still allows the iterator to
have a lifetime parameter.
This now makes it unsafe to save the pointer returned by .with_c_str
as that pointer now may be pointing at a stack allocated array.
I arbitrarily chose 32 bytes as the length of the stack vector, and
so it might not be the most optimal size.
before:
test c_str::bench::bench_with_c_str_long ... bench: 539 ns/iter (+/- 91)
test c_str::bench::bench_with_c_str_medium ... bench: 97 ns/iter (+/- 2)
test c_str::bench::bench_with_c_str_short ... bench: 70 ns/iter (+/- 5)
after:
test c_str::bench::bench_with_c_str_long ... bench: 542 ns/iter (+/- 13)
test c_str::bench::bench_with_c_str_medium ... bench: 53 ns/iter (+/- 6)
test c_str::bench::bench_with_c_str_short ... bench: 19 ns/iter (+/- 0)
The current implementation uses `&v[0]` for the lifetime struct field,
but that is a dangling pointer for iterators derived from zero-length
slices.
Example:
let v: [int, ..0] = []; println!("{:?}", v.iter())
std::vec::VecIterator<,int>{ptr: (0x7f3768626100 as *()), end: (0x7f3768626100 as *()), lifetime: &139875951207128}
To replace this parameter, use a field of type `Option<&'self ()>`
that is simply initialized with `None`, but still allows the iterator to
have a lifetime parameter.
This lifts various restrictions on the runtime, for example the character limit
when logging a message. Right now the old debug!-style macros still involve
allocating (because they use fmt! syntax), but the new debug2! macros don't
involve allocating at all (unless the formatter for a type requires allocation.
This also includes a fix for yielding from single-threaded schedulers where the scheduler would stop working before its work queue was empty. Fixes the deadlocks that this patch had previously.
Fix#7752.
~~(The glob API is a little funky; I tried to make a small test for it, which I'll add to the end of this description, and its not clear whether globfree is supposed to free solely the structure allocated by glob itself, or if it is going to try to free more than that.)~~ (The previous note was a user-error: I was misusing the CString API.)
Anyway, this seems to work in terms of calling errfunc where expected.)
```rust
#[allow(unused_imports)];
use std::libc::types::os::arch::c95::{c_char, c_int, size_t};
use std::libc::funcs::posix01::glob;
use std::libc::types::os::common::posix01::glob_t;
use std::libc::consts::os::posix01::{GLOB_APPEND, GLOB_DOOFFS, GLOB_ERR,
GLOB_MARK, GLOB_NOCHECK, GLOB_NOSORT,
GLOB_NOESCAPE, GLOB_NOSPACE,
GLOB_ABORTED, GLOB_NOMATCH};
use std::ptr;
use std::c_str;
#[fixed_stack_segment]
fn main() {
let mut g = glob_t {
gl_pathc: 0, // size_t,
__unused1: 0, // c_int,
gl_offs: 2, // size_t,
__unused2: 0, // c_int,
gl_pathv: ptr::null(), // **c_char,
__unused3: ptr::null(), // *c_void,
__unused4: ptr::null(), // *c_void,
__unused5: ptr::null(), // *c_void,
__unused6: ptr::null(), // *c_void,
__unused7: ptr::null(), // *c_void,
__unused8: ptr::null(), // *c_void,
};
extern "C" fn errfunc(_epath: *c_char, _errno: int) -> int {
println!("errfunc called");
return 0;
}
struct Reduced { pathc: size_t, offs: size_t, pathv: **c_char, }
impl Reduced {
fn from(g: &glob_t) -> Reduced {
Reduced {pathc: g.gl_pathc, offs: g.gl_offs, pathv: g.gl_pathv}
}
}
do ("*.rs/*").with_c_str |pat| {
println!("calling glob");
unsafe { glob::glob(pat, GLOB_DOOFFS, errfunc, &mut g); }
println!("After glob call");
println!("g: {:?}", Reduced::from(&g));
for i in range(0, g.gl_pathc as int) {
unsafe {
let p : **c_char = ptr::offset(g.gl_pathv, g.gl_offs as int + i);
let x = c_str::CString::new(*p, false);
match x.as_str() {
Some(s) => {
println!("gl_pathc[{:d}]: {:?}", i, s);
}
None => {
println!("gl_pathc[{:d}]: unvalid", i);
}
}
}
}
}
println!("calling globfree on g: {:?}", g);
unsafe { glob::globfree(&mut g); }
println!("after globfree call");
}
```
If there's no TLS key just yet, then there's nothing to unsafely borrow, so
continue returning None. This prevents causing the runtime to abort itself when
logging before the runtime is fully initialized.
Closes#9487
r? @brson
Moved OwnedStr doc comments from impl to trait.
Added a few #[inline] hints.
The doc comment changes make the source a bit harder to read, as
documentation and implementation no longer live right next to each
other. But this way they at least appear in the docs.
This lifts various restrictions on the runtime, for example the character limit
when logging a message. Right now the old debug!-style macros still involve
allocating (because they use fmt! syntax), but the new debug2! macros don't
involve allocating at all (unless the formatter for a type requires allocation.
If there's no TLS key just yet, then there's nothing to unsafely borrow, so
continue returning None. This prevents causing the runtime to abort itself when
logging before the runtime is fully initialized.
Closes#9487
Previously, any package would match any other package ID when searching
using the rust_path_hack, so long as the directory had one or more crate
files in it. Now, rustpkg checks that the parent directory matches the
package ID.
Closes#9273
In "/src/libstd/char.rs", there are function and method definitions for `is_lowercase()`, `is_uppercase()`, `is_whitespace()`, etc. However, there was no function or method for control characters, so I added the `is_control()` function and method definitions along with documentation and tests. Running `./configure && make check` shows that all tests for `is_control()` pass.
Progress on #7981
This doesn't completely close the issue because `struct A;` is still allowed, and it's a much larger change to disallow that. I'm also not entirely sure that we want to disallow that. Regardless, punting that discussion to the issue instead.
