* There is no clear reason why the map should hold const std::strings
* This class is never derived, there doesn't seem to be a compelling
reason to ever do so, so no need to make virtual members
* In general const member data is an anti-pattern, the *only* reason
is to prevent assignability, and the accepted idiom for that is to
to declare the assigment operator private. This change does so, and
also prevents copy construction.
ZMQ_INVERT_MATCHING reverses the PUB/SUB prefix matching. The subscription
list becomes a rejection list. The PUB socket sends messages to all
connected (X)SUB sockets that do not have any matching subscription.
Whenever the option is used on a PUB/XPUB socket, any connecting SUB
sockets must also set it or they will reject everything the publisher
sends them. XSUB sockets are unaffected because they do not filter out
incoming messages.
Symptom is that ZMQ_STREAM sockets in 4.1.0 and 4.1.1 generate zero
sized messages on each new connection, unlike 4.0.x which did not do
this.
Person who made this commit also changed test cases so that contract
breakage did not show. Same person was later banned for persistently
poor form in CZMQ contributions.
Solution: enable connect notifications on ZMQ_STREAM sockets using a
new ZMQ_STREAM_NOTIFY setting. By default, socket does not deliver
notifications, and behaves as in 4.0.x.
Fixes#1316
When pipe::write succeeds, it takes control of the message's data buffer.
When it fails, it has not taken control. The caller should clean up the
message appropriately (msg::close).
pipe_t.write only takes control of the underlying message memory when it
succeeds. When it returns failure, we must close the message ourselves to
clean up that memory.
This patch is sponsored by FarSounder, Inc (farsounder.com)
Allows non-C/C++ based clients easy access to the peer's IP address via
zmq_msg_gets(&msg, "Peer-Address") instead of zmq_msg_get(&msg, ZMQ_SRCFD)
followed by calls to getpeername and getnameinfo
Increasing it would have at least two benefits -
* More messages would be 'VSM' messages, so it would reduce allocation
overhead a bit.
* Remove any chance of false sharing of things that are, by design,
pushed by value onto a ypipe_t<msg_t> which is shared between two threads.
The only downside I see is slightly increased memory consumption on memory
constrained applications.
- Full discussion of this rationale is part of issue #1295
zmq_atomic_counter_dec returned a 'bool' value, yet this isn't
defined by standard, so causes compile errors in upstream code.
Solution: return an int that can be safely converted to bool if
needed by bindings.
Solution: as libzmq already provides this across all platforms,
expose an atomic counter API. I've not wrapped atomic pointers,
though someone who needs this may want to do so.
E.g. when server is not configured, and client tries PLAIN security,
there is no hint of why this does not work.
Solution: add debugging output for this case. Note that the various
debugging outputs for security failures should probably be sent to
an inproc monitor of some kind.
auth mechanisms were only enabled when ZMTP handshake
is latest version, meaning that connections from old sockets
would skip authentication altogether
Solution: set defaults back to infinity, and add new context
option, ZMQ_BLOCKY that the user can set to false to get a
less surprising behavior on context termination. Eg.
zmq_ctx_set (ctx, ZMQ_BLOCKY, false);
There are two todo comments in curve_client.cpp and curve_server.cpp that suggest
checking the return code of sodium_init() call. sodium_init() returns -1 on error,
0 on success and 1 if it has been called before and is already initalized:
https://github.com/jedisct1/libsodium/blob/master/src/libsodium/sodium/core.c
This commit adds a ZMQ_POLLPRI flag that maps to poll()'s POLLPRI
flag.
This flags does nothing for OMQ sockets. It's only useful for raw
file descriptor (be it socket or file).
This flag does nothing if poll() is not the underlying polling
function. So it is Linux only.
For OS X, the microseconds field is implemented as an int type. The implicit narrowing in the initializer list throws a compiler error for some compilers with C++11 support turned on. The specific error message is: "error: non-constant-expression cannot be narrowed from type 'long' to '__darwin_suseconds_t' (aka 'int') in initializer list [-Wc++11-narrowing]".
Tested on Clang 5.1.0 and Mac OS X 10.9.4.
When Curve authentication is used, libsodium opens a file
descriptor to /dev/urandom to generate random bytes. When
the ZMQ context terminates, it should ensure that file gets
closed.
It's bad practice to start by testing all exceptional conditions
and then dropping through to the 'normal' condition. Apart from
being inefficient, it's deceptive to the user. Conditional code
should always try to show the natural expectation of the code,
with exceptional cases coming last.
Solution: clean up this code.
Solution: change setsockopts on printable keys to expect 41, nor 40
bytes. Code still accepts 40 bytes for compatibility, and copies the
key to a well-terminated string before using it.
Fixes#1148
Updated:
src/thread.cpp: On older z/OS UNIX System Services,
pthread_{get,set}schedparam is not present (searching the
Internet suggests it may be present in later version than
the one being used for z/OS UNIX System Services porting).
Make zmq::thread_t::setSchedulingParameters() a no-op on
z/OS UNIX System Services.
NOTE: pthread_{get,set}schedparam appear to have been introduced
by POSIX.1-2001 or IEEE 1003.1-2004 so may not be universally
available, and thus more platforms may need this "no-op" treatment.
Updated:
src/metdata.hpp: Remove explicit "const" from key of std::map<>
because the key is implicitly const already (see
http://en.cppreference.com/w/cpp/container/map and
http://www.cplusplus.com/reference/map/map/).
On some platforms (such as z/OS UNIX System Services) explicitly
declaring the map key as "const" causes template expansion errors
as it tries to create separate allocators for "const const std::string"
and "const std::string" only to find that they clash. (Presumably
some compilers collapse these into one earlier.)
There are no template expansion errors if the map key is left to be
implicitly const.
Updated:
src/signaler.cpp: Add close_wait_ms() static function to loop
when receiving EAGAIN in response to close(), with ms long
sleeps, up to a maximum limit (default 2000ms == 2 seconds);
used in signaler_t::~signaler_t() destructor.
Rationale: In a real-time environment it is sometime mandatory to tune
threads priority and scheduling policy. This is required by our users
who mixes real-time and server threads within
the same process. It's not planned to support this on non-pthread
platforms (e.g. Windows).
Since https://github.com/zeromq/libzmq/commit/350a1a, TCP addresses
get resolved asynchronously, so zmq_connect no longer returned an
error on incorrect addresses.
This is troublesome since we rely on some error checking to catch
blatant errors.
Solution add some upfront syntax checking that catches at least the
obvious kinds of errors (invalid characters, wrong or missing port
number).
