1
0
mirror of https://github.com/fafhrd91/actix-web synced 2025-02-25 13:22:50 +01:00

207 Commits

Author SHA1 Message Date
Yuki Okushi
0d5646a8b6
Run rustfmt 2020-03-08 00:52:39 +09:00
Maxim Vorobjov
10e3e72595
Http2 client configuration to improve performance ()
* add defaults for http2 client configuration

* fix spaces

* Add changes text for extended H2 defaults buffers

* client: configurable H2 window sizes and max_http_version

* add H2 window size configuration and max_http_version to awc::ClientBuilder

* add awc::ClientBuilder H2 window sizes and max_http_version

* add test for H2 window size settings

* cleanup comment

* Apply code review fixes

* Code review fix for awc ClientBuilder

* Remove unnecessary comments on code review

* pin quote version to resolve build issue

* max_http_version to accept http::Version

* revert fix for quote broken build
2020-03-07 11:09:31 +09:00
Aaron Hill
e90950fee1
Re-apply commit 2cf7b3ad20fb823314426a5e33b0805045ec1d8a
This ended up getting reverted by , which re-introduced an unsound
use of `Pin::new_unchecked`

See my original PR  for the reasoning behind this change.
2020-03-04 11:27:58 -05:00
Yuki Okushi
b4d63667df Demote lint level to warn 2020-02-27 22:39:11 +09:00
Yuki Okushi
6cc83dbb67 Allow clippy lint for compatibility 2020-02-27 12:45:11 +09:00
Yuki Okushi
f27dd19093 Fix Clippy warnings 2020-02-27 12:39:04 +09:00
Yuki Okushi
7ba14fd113 Run rustfmt 2020-02-27 11:10:55 +09:00
Yuki Okushi
cd1765035c Avoid re-definition 2020-02-27 09:42:32 +09:00
Maksym Vorobiov
0a86907dd2 use mem::replace instead of mem::take rust 1.40+ 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
de815dd99c Fixed condition for finishing transfer of response 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
e6078bf792 Fix EncoderBody enum to align with Body::Message 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
a84b37199a Add Unpin to Body to get rid of unsafe in MessageBody 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
c05f9475c5 refactor dispatcher to avoid possible UB with DispatcherState Pin 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
69dab0063c Get rid of one more unsafe 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
ec5c779732 unlink MessageBody from Unpin 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
2e2ea7ab80 remove extra whitespaces and Unpins 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
09a391a3ca rollback changes to actix-web, awc and test-server for now 2020-02-27 09:37:05 +09:00
Maksym Vorobiov
9d04b250f9 This is a squashed commit:
- Convert MessageBody to accept Pin in poll_next

- add CHANGES and increase versions aligned to semver

- update crates to accomodate MessageBody Pin change

- fix tests and dependencies
2020-02-27 09:37:05 +09:00
Maksym Vorobiov
a4148de226 add test crashing with segfault according to 2020-02-27 09:36:30 +09:00
Maxim Vorobjov
48ef4d7a26
Add actix-http support for actix error messages ()
* Moved actix-http for actix from actix crate

* remove resolver feature

* renamed actix feature to actor

* fixed doc attr for actors, add documentation
2020-02-27 09:34:49 +09:00
Aaron Hill
71c4bd1b30
Remove uses of Pin::new_unchecked in h1 Dispatcher ()
This removes the last uses of unsafe `Pin` functions in actix-web.

This PR adds a `Pin<Box<_>>` wrapper to `DispatcherState::Upgrade`,
`State::ExpectCall`, and `State::ServiceCall`.

The previous uses of the futures `State::ExpectCall` and `State::ServiceCall`
were Undefined Behavior - a future was obtained from `self.expect.call`
or `self.service.call`, pinned on the stack, and then immediately
returned from `handle_request`. The only alternative to using `Box::pin`
would be to refactor `handle_request` to write the futures directly into
their final location, or avoid polling them before they are returned.

The previous use of `DispatcherState::Upgrade` doesn't seem to be
unsound. However, having data pinned inside an enum that we
`std::mem::replace` would require some careful `unsafe` code to ensure
that we never call `std::mem::replace` when the active variant contains
pinned data. By using `Box::pin`, we any possibility of future
refactoring accidentally introducing undefined behavior.

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-02-26 08:21:05 +09:00
Yuki Okushi
2a72e8d119
Merge branch 'master' into replace-unsafe-content-length-helper 2020-02-25 14:30:04 +09:00
Yuki Okushi
845ce3cf34 Fix doc comment 2020-02-25 07:46:03 +09:00
Rob Ede
94da08f506
increase content-length fast path to responses up to 1MB 2020-02-24 20:58:41 +00:00
Yuki Okushi
c8ccc69b93 actix-http: update time to 0.2.7 2020-02-23 07:09:00 +09:00
Daniel YU
245f96868a
impl downcast_ref for MessageBody ()
Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-02-21 13:31:51 +09:00
Aaron Hill
e6811e8818
Use #[pin_project] with ConnectorPoolSupport
This removes a use of `Pin::get_unchecked_mut`
2020-02-19 21:42:53 -05:00
Rob Ede
f266b44cb0
replace unsafe blocks in write_usize helper 2020-02-16 15:20:25 +00:00
Rob Ede
82b2786d6b
replace unsafe content length implementation 2020-02-16 14:31:05 +00:00
zero-systems
728b944360
Extensions module improvement and tests. ()
* replace get.is_some to contains_key

* Add tests

* remove unnecessary box cast

* fix missing uints

* asserts fix

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-02-07 16:08:25 +09:00
Aaron Hill
fe13789345
Use Pin<Box<S>> in BodyStream and SizedStream ()
Fixes 

A better fix would be to change `MessageBody` to take a `Pin<&mut
Self>`, rather than a `Pin<&mut Self>`. This will avoid requiring the
use of `Box` for all consumers by allowing the caller to determine how
to pin the `MessageBody` implementation (e.g. via stack pinning).

However, doing so is a breaking change that will affect every user of
`MessageBody`. By pinning the inner stream ourselves, we can fix the
undefined behavior without breaking the API.

I've included @sebzim4500's reproduction case as a new test case.
However, due to the nature of undefined behavior, this could pass (and
not segfault) even if underlying issue were to regress.

Unfortunately, until  is resolved,
it's not even possible to write a Miri test that will pass when the bug
is fixed.

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-31 09:39:34 +09:00
Aaron Hill
3033f187d2
Enforce safety of downcast_ref at compile time. ()
* Enforce safety of `downcast_ref` at compile time.

The safety of `downcast_ref` requires that `__private_get_type_id__` not
be overriden by callers, since the returned `TypeId` is used to check if
the cast is safe. However, all trait methods in Rust are public, so
users can override `__private_get_type_id__` despite it being
`#[doc(hidden)]`.

This commit makes `__private_get_type_id__` return a type with a private
constructor, ensuring that the only possible implementation is the
default implementation. A more detailed explanation is provided in the
comments added to the file.

Note that the standard library was affected by this type of issue with
the `Error::type_id` function: see https://blog.rust-lang.org/2019/05/14/Rust-1.34.2.html#whats-in-1.34.2-stable

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-30 23:43:35 +09:00
Aaron Hill
276a5a3ee4
Replace UnsafeCell with Cell in DateServiceInner ()
* Replace `UnsafeCell` with `Cell` in `DateServiceInner`

This ensures that it's impossible to cause undefined behavior by
accidentally violating Rust's aliasing rules (e.g. passing a closure to
`set_date` which ends up invoking `reset` or `update` on the inner
`DateServiceInner`).

There might be a tiny amount of overhead from copying the `Option<(Date,
Instant)>` rather than taking a reference, but it shouldn't be
measurable.

Since the wrapped type is `Copy`, a `Cell` can be used, avoiding the
runtime overhead of a `RefCell`.

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-29 21:05:08 +09:00
kevinpoitra
e634e64847 Upgrade time to 0.2.5 ()
* Use `OffsetDateTime` instead of `PrimitiveDateTime`

* Parse time strings with `PrimitiveDateTime::parse` instead of `OffsetDateTime::parse`

* Remove unused `time` dependency from actix-multipart

* Fix a few errors with time related tests from the `time` upgrade

* Implement logic to convert a RFC 850 two-digit year into a full length year, and organize time parsing related functions

* Upgrade `time` to 0.2.2

* Correctly parse C's asctime time format using time 0.2's new format patterns

* Update CHANGES.md

* Use `time` without any of its deprecated functions

* Enforce a UTC time offset when converting an `OffsetDateTime` into a Header value

* Use the more readable version of `Duration::seconds(0)`, `Duration::zero()`

* Remove unneeded conversion of time::Duration to std::time::Duration

* Use `OffsetDateTime::as_seconds_f64` instead of manually calculating the amount of seconds from nanoseconds

* Replace a few additional instances of `Duration::seconds(0)` with `Duration::zero()`

* Truncate any nanoseconds from a supplied `Duration` within `Cookie::set_max_age` to ensure two Cookies with the same amount whole seconds equate to one another

* Fix the actix-http:🍪:do_not_panic_on_large_max_ages test

* Convert `Cookie::max_age` and `Cookie::expires` examples to `time` 0.2

Mainly minor  changes. Type inference can be used alongside the new
`time::parse` method, such that the type doesn't need to be specified.
This will be useful if a refactoring takes place that changes the type.
There are also new macros, which are used where possible.

One change that is not immediately obvious, in `HttpDate`, there was an
unnecessary conditional. As the time crate allows for negative durations
(and can perform arithmetic with such), the if/else can be removed
entirely.

Time v0.2.3 also has some bug fixes, which is why I am not using a more
general v0.2 in Cargo.toml.

v0.2.3 has been yanked, as it was backwards imcompatible. This version
reverts the breaking change, while still supporting rustc back to
1.34.0.

* Add missing `time::offset` macro import

* Fix type confusion when using `time::parse` followed by `using_offset`

* Update `time` to 0.2.5

* Update CHANGES.md

Co-authored-by: Jacob Pratt <the.z.cuber@gmail.com>
2020-01-28 20:44:22 +09:00
Kai Ren
cdba30d45f Skip empty chucks for BodyStream and SizedStream ()
* Skip empty chucks for BodyStream and SizedStream when streaming response ()

* Fix tests to fail on previous implementation

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-28 18:28:09 +09:00
Aaron Hill
74dcc7366d Remove several uses of Pin::new_unchecked ()
Most of the relevant struct already had a `#[pin_project]` attribute,
but it wasn't being used.

The remaining uses of `Pin::new_unchecked` all involve going from a
`&mut T` to a `Pin<&mut T>`, without directly observing a `Pin<&mut T>`
first. As such, they cannot be replaced by `pin_project`

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-28 12:35:51 +09:00
Maxim Vorobjov
8888520d83 Add benchmark for full stack request lifecycle ()
* add benchmark for full stack request lifecycle

* add direct service benchmarks

* fix newline

* add cloneable service benchmarks

* remove cloneable bench experiments + cargo fmt

Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-25 08:05:25 +09:00
Maxim Vorobjov
58844874a0 Fixing convert UnsafeCell to RefCell in CloneableService ()
Co-authored-by: Yuki Okushi <huyuumi.dev@gmail.com>
2020-01-24 14:51:38 +09:00
Rob Ede
a3287948d1 allow explicit SameSite=None cookies ()
fixes 
2020-01-23 10:08:23 +09:00
linkmauve
abb462ef85 Replace sha1 dependency with sha-1 ()
* Replace sha1 dependency with sha-1

This other crate is being maintained, and it offers better performances
when using the `asm` feature (especially [on
AArch64](https://github.com/RustCrypto/hashes/pull/97)).

* Update CHANGES.md with the sha-1 migration

* Add a test for hash_key()
2020-01-11 02:34:31 +09:00
Nikolay Kim
e66312b664 add extra constraints 2020-01-10 11:36:59 +06:00
Nikolay Kim
1d12ba9d5f Replace brotli with brotli2 2019-12-20 13:50:07 +06:00
Nikolay Kim
8c54054844 Use .advance() intead of .split_to() 2019-12-19 09:56:14 +06:00
Nikolay Kim
1732ae8c79 fix Bodyencoding trait usage 2019-12-18 09:30:14 +06:00
Rajasekharan Vengalil
3b860ebdc7 Fix poll_ready call for WebSockets upgrade ()
* Fix poll_ready call for WebSockets upgrade

* Poll upgrade service from H1ServiceHandler too
2019-12-17 13:34:25 +06:00
Andrii Radyk
30dcaf9da0 fix deprecated Error::description () 2019-12-16 07:43:19 +06:00
Nikolay Kim
3d64d565d9 fix warnings 2019-12-13 11:46:02 +06:00
Nikolay Kim
c1deaaeb2f cleanup imports 2019-12-13 11:24:57 +06:00
Nikolay Kim
b81417c2fa fix warnings 2019-12-13 10:59:02 +06:00
Nikolay Kim
4937c9f9c2 refactor http-test server 2019-12-12 23:08:38 +06:00