This was intended as part of the I/O timeouts commit, but it was mistakenly
forgotten. The type of the timeout argument is not guaranteed to remain constant
into the future.
With `~[T]` no longer growable, the `FromIterator` impl for `~[T]` doesn't make
much sense. Not only that, but nearly everywhere it is used is to convert from
a `Vec<T>` into a `~[T]`, for the sake of maintaining existing APIs. This turns
out to be a performance loss, as it means every API that returns `~[T]`, even a
supposedly non-copying one, is in fact doing extra allocations and memcpy's.
Even `&[T].to_owned()` is going through `Vec<T>` first.
Remove the `FromIterator` impl for `~[T]`, and adjust all the APIs that relied
on it to start using `Vec<T>` instead. This includes rewriting
`&[T].to_owned()` to be more efficient, among other performance wins.
Also add a new mechanism to go from `Vec<T>` -> `~[T]`, just in case anyone
truly needs that, using the new trait `FromVec`.
[breaking-change]
The code in resolve erroneously assumed that private enums weren't visited, so
the logic was adjusted to check to see if the enum definition itself was public.
Closes#11680
As part of #5527 I had to make some changes here and I just couldn't take it anymore. Refactor the writeback code. Should be functionally equivalent to the old stuff.
r? @pcwalton
This code does not belong in libstd, and rather belongs in a dedicated crate. In
the future, the syntax::ext::format module should move to the fmt_macros crate
(hence the name of the crate), but for now the fmt_macros crate will only
contain the format string parser.
The entire fmt_macros crate is marked #[experimental] because it is not meant
for general consumption, only the format!() interface is officially supported,
not the internals.
This is a breaking change for anyone using the internals of std::fmt::parse.
Some of the flags have moved to std::fmt::rt, while the actual parsing support
has all moved to the fmt_macros library.
[breaking-change]
There was no reason to remove them from slice. They're testing methods
defined in slice, so that's where they belong.
Leave vec with copies of the partition/partitioned tests because it has
its own implementation of those methods.
Bring back the Decodable impl for ~[T], this time using FromVec. It's
still not recommended that anyone use this, but at least it's available
if necessary.
Add a new trait FromVec with one self-less method from_vec(). This is
kind of like FromIterator, but it consumes a Vec<T>. It's only
implemented for ~[T], but the idea is post-DST it can be implemented for
any Boxed<[T]>.
API Changes:
- from_base64() returns Result<Vec<u8>, FromBase64Error>
- from_hex() returns Result<Vec<u8>, FromHexError>
- json::List is a Vec<Json>
- Decodable is no longer implemented on ~[T] (but Encodable still is)
- DecoderHelpers::read_to_vec() returns a Result<Vec<T>, E>
A few methods in slice that used to return ~[T] now return Vec<T>:
- VectorVector.concat/connect_vec() returns Vec<T>
- slice::unzip() returns (Vec<T>, Vec<U>)
- ImmutableCloneableVector.partitioned() returns (Vec<T>, Vec<T>)
- OwnedVector.partition() returns (Vec<T>, Vec<T>)
This used to create a Vec<T> and then call .move_iter().collect() to
convert to a ~[T]. We can't do that anymore, so construct the ~[T] in
place instead. This has the added benefit of avoiding an unnecessary
memory copy (from the Vec<T> to the ~[T]).
As part of the shift from ~[T] to Vec<T>, recently ~[T] was made
non-growable. However, the FromIterator implementation for ~[T] was left
intact (albeit implemented inefficiently), which basically provided a
loophole to grow a ~[T] despite its non-growable nature. This is a
problem, both for performance reasons and because it encourages APIs to
continue returning ~[T] when they should return Vec<T>. Removing
FromIterator forces these APIs to adopt the correct type.
Furthermore, during today's weekly meeting it was decided that we should
remove all instances of ~[T] from the standard libraries in favor of
Vec<T>. Removing the FromIterator impl makes sense to do as a result.
This commit only includes the removal of the FromIterator impl. The
subsequent commits involve handling all of the breakage that results,
including changing APIs to use Vec<T> instead of ~[T]. The precise API
changes are documented in the subsequent commit messages, but each
commit is not individually marked as a breaking change.
Finally, a new trait FromVec is introduced that provides a mechanism to
convert Vec<T> back into ~[T] if truly necessary. It is a bit awkward to
use by design, and is anticipated that it will be more useful in a
post-DST world to convert to an arbitrary Foo<[T]> smart pointer.
[breaking-change]
This code does not belong in libstd, and rather belongs in a dedicated crate. In
the future, the syntax::ext::format module should move to the fmt_macros crate
(hence the name of the crate), but for now the fmt_macros crate will only
contain the format string parser.
The entire fmt_macros crate is marked #[experimental] because it is not meant
for general consumption, only the format!() interface is officially supported,
not the internals.
This is a breaking change for anyone using the internals of std::fmt::parse.
Some of the flags have moved to std::fmt::rt, while the actual parsing support
has all moved to the fmt_macros library.
[breaking-change]
This was intended as part of the I/O timeouts commit, but it was mistakenly
forgotten. The type of the timeout argument is not guaranteed to remain constant
into the future.
This commit brings the local_data api up to modern rust standards with a few key
improvements:
* All functionality is now exposed as a method on the keys themselves. Instead
of importing std::local_data, you now use "key.set()" and "key.get()".
* All closures have been removed in favor of RAII functionality. This means that
get() and get_mut() no long require closures, but rather return
Option<SmartPointer> where the smart pointer takes care of relinquishing the
borrow and also implements the necessary Deref traits
* The modify() function was removed to cut the local_data interface down to its
bare essentials (similarly to how RefCell removed set/get).
[breaking-change]
This PR is an implementation of `set_timeout`, `set_read_timeout`, and `set_write_timeout` for TCP, UDP, and Unix streams (named pipes on windows).
The implementation was much more difficult than I imagined it was going to be throughout the 9 categories ({tcp, udp, unix} x {windows, unix, green}). The major snag is that libuv doesn't support canceling writes, so I chose to word the `set_write_timeout` documentation in such a way that it accomadates the behavior seen when running around with libgreen.
The first commit is from #13751, and I just included it to pre-emptively avoid rebase conflicts. The following commits are relevant to this PR. The tests aren't quite passing on windows just yet, but I should have those working by tomorrow once my VM is back up and running. For now, I wanted to see what others' thoughts were on this strategy.
This commit brings the local_data api up to modern rust standards with a few key
improvements:
* The `pop` and `set` methods have been combined into one method, `replace`
* The `get_mut` method has been removed. All interior mutability should be done
through `RefCell`.
* All functionality is now exposed as a method on the keys themselves. Instead
of importing std::local_data, you now use "key.replace()" and "key.get()".
* All closures have been removed in favor of RAII functionality. This means that
get() and get_mut() no long require closures, but rather return
Option<SmartPointer> where the smart pointer takes care of relinquishing the
borrow and also implements the necessary Deref traits
* The modify() function was removed to cut the local_data interface down to its
bare essentials (similarly to how RefCell removed set/get).
[breaking-change]
This is the last remaining networkig object to implement timeouts for. This
takes advantage of the CancelIo function and the already existing asynchronous
I/O functionality of pipes.
This commit implements the set{,_read,_write}_timeout() methods for the
libuv-based networking I/O objects. The implementation details are commented
thoroughly throughout the implementation.
This commit has an implementation of the previous commit's timeout interface for
I/O objects on unix platforms. For implementation details, see the large comment
at the end of libnative/io/net.rs which talks about the general strategy taken.
Thankfully, all of these implementations can share code because they're
performing all the same operations.
This commit does not implement timeouts for named pipes on windows, only tcp/udp
objects on windows (which are quite similar to their unix equivalents).
These timeouts all follow the same pattern as established by the timeouts on
acceptors. There are three methods: set_timeout, set_read_timeout, and
set_write_timeout. Each of these sets a point in the future after which
operations will time out.
Timeouts with cloned objects are a little trickier. Each object is viewed as
having its own timeout, unaffected by other objects' timeouts. Additionally,
timeouts do not propagate when a stream is cloned or when a cloned stream has
its timeouts modified.
This commit is just the public interface which will be exposed for timeouts, the
implementation will come in later commits.
Fix#13965.
This commit adopts the second strategy I outlined in #13965, where the bulk of the code is still "smoke tested" (in the sense that rustdoc attempts to run it, sending all of the generated output into a locally allocated `MemWriter`). The part of the code that is ignored (but included in the presentation) is isolated to a three-line `main` function that invokes the core rendering routine.
In the generated rustdoc output, this leads to a small break between the two code blocks, but I do not think this is a large issue.
bors
Kevin Ballard
Alex Crichton
Patrick Walton