* Allow setting a seed in websocket prng to workaround a valgrind bug
in Xenial.
* Coverage collection in Azp.
* Fixup blacklists to avoid zlib bugs.
* Use native b2 features for sanitizers and valgrind.
* Expanded Windows build matrix.
* Add additional clang (with libc++) builds.
Signed-off-by: Damian Jarek <damian.jarek93@gmail.com>
- Don't set the global link settings (`<link>`) to avoid dependency
ordering issues in the future.
- Allow the user to decide on the link settings for dependencies
(filesystem/coroutine).
- Make dependencies in examples explicit.
- Remove dependency of examples on Boost.Filesystem (was not used there).
- Deduplicate build settings.
- Speed up test compilation by avoiding rebuilding of main test file.
Signed-off-by: Damian Jarek <damian.jarek93@gmail.com>
fix#1279, close#1319
This enables users to improve compilation performance by explicitly
instantiating the stream template in another TU.
Signed-off-by: Damian Jarek <damian.jarek93@gmail.com>
This resolves a medium vulnerability described in the
Beast Hybrid Assessment Report by Bishop Fox, where masks generated
for use with outgoing WebSocket client frames use an insufficient
source of entropy and a non-cryptographically secure pseudo-random
number generator.
By default, all newly constructed WebSocket streams will use a
uniquely seeded secure PRNG (ChaCha20 in counter mode). As this may
result in increased CPU resource consumption, the function
websocket::stream::secure_prng() may be used to select a faster but
less secure PRNG, for the case where the caller knows that the secure
generator is not necessary.
On some systems, std::random_device may produce insufficient entropy
to securely seed the PRNG. As this condition cannot be detected by
Beast, callers may use the function websocket::seed_prng() called
once at startup to provide at least 256 bits of entropy which will
be used to uniquely seed all subsequent PRNGs.
Split the websocket read test to reduce compile time memory usage
by a small amount, which ought to be enough to let it compile in
a constrained environment like Travis.
Signed-off-by: Damian Jarek <damian.jarek93@gmail.com>
* teardown_tag is replaced with teardown_role, a client/server
flag used to determine whether the shutdown is performed
before or after reading the EOF. This is in accordance with
RFC6455 7.1.1:
https://tools.ietf.org/html/rfc6455#section-7.1.1
Actions Required:
* Modify signatures of teardown and async_teardown to use
teardown_role instead of teardown_tag
* Change calls to teardown and async_teardown to pass the
correct role: client or server depending on context.