2113 Commits

Author SHA1 Message Date
Jonathan Reams
e9a5bc8d1e Fix units and default values for heartbeats options
Set the ZMQ_HEARTBEAT_TIMEOUT to default to the value of
ZMQ_HEARTBEAT_IVL if it's not explicitly set.
Change the units of ZMQ_HEARTBEAT_TTL to milliseconds in the API
and round down to the nearest decisecond so that all the options
are using the same units.
Make the maximum heartbeat TTL match the spec (6553 seconds)
2015-06-26 14:25:58 -04:00
Jonathan Reams
cbb3b176a6 Add ZMTP heartbeats
This commit adds ZMTP connection heartbeats described in
http://rfc.zeromq.org/spec:37/ZMTP.
2015-06-24 13:24:53 -04:00
Marin Atanasov Nikolov
fa3b5d8b03 Typo fix 2015-06-23 14:45:51 +03:00
Constantin Rack
15f9ee19cb Solution: allow brackets in tcp ipv6 address 2015-06-23 08:54:49 +02:00
Jens Auer
e70b5efa0b Reuse buffer when no messages depend on it. 2015-06-14 14:01:04 +02:00
Jens Auer
51cb57e2c9 Fixed wrong handling of shared messages
The shared reference count was not shared but copied. msg_t cannot
store the refcnt itsef but has to store a pointer to an externally
allocated (shared) refcnter. The changes to lmsg are reverted to
use content_t again. Howver, this introduces an allocation in v2_decoder
when creating the message which can be avoided. When allocating the reception
buffer, space is allocated for the maximum number of reference counts
(8192 / max_vsm_size = 8192/64 = 128 zmq:atomic_counter objects). This
increases the buffer by 128*sizeof(atomic_counter) = 128*4 = 512 bytes only.
When creating a message, the refcnt member is set to the address of one of the
pre-allocated atomic_counter_t objects. To do so, a new msg_t type zcmsg
is introduced because msg::copy must discriminate between the message types
when releasing memory.
2015-06-14 13:35:33 +02:00
Jens Auer
dfe1908008 Fixed wrong buffer end detection in v2_decoder.
zero-copy msg_t::init cannot be used when the message exceeds either
the buffer end or the last received byte. To detect this, the buffer
is now resized to the numnber of received bytes.
2015-06-14 13:35:23 +02:00
Kapp Arnaud
8ffd723271 Probably fix build issue on Solaris.
See #1220 recent comments. This change is untested yet seems
safe.
2015-06-13 22:08:14 +02:00
Min RK
158d77d63e don't check POLLOUT for single-socket proxies 2015-06-06 21:05:32 -07:00
Kapp Arnaud
c35971a603 Fix initialization order.
Fixes #1431.
2015-06-07 02:21:32 +02:00
Richard Newton
dc949624e1 Handle infinite hwms properly 2015-06-05 19:06:36 +01:00
Richard Newton
15eecf4cf4 Update high water marking to allow changing high water mark after connection established. 2015-06-05 17:14:55 +01:00
Richard Newton
9f8b8af926 Fail ZMQ_SNDHWM and ZMQ_RCVHWM setsockopt if already connected. 2015-06-05 10:43:40 +01:00
Jens Auer
e9b403a7b1 zero-copy message receive
Construct messages from a reference-counted buffer allocated once
per receive instead of copying the data.
2015-06-03 23:03:14 +02:00
Jens Auer
611e96c701 Allocation-free msg::init_data
With a msg_t size of 64 bytes, it becomes possible to embedd the content_t's members
struct for large messages directly in the msg_t. This saves the dynamic allocation
of content_t obejcts when using msg_t::init_data.

content_t contains a zmq::atomic_counter_t object which is not a POD in C++98
and thus it cannot be used as a member of the union u. To bypass this, C++11
is used which has relaxed rules for POD and atomic_counter is a C++11-POD. An
alternative would have been to make atomic_counter a classical POD by removing
constructors and all private member functions, i.e. have a struct and free functions
to manipulate it.

A new msg_t::init function is added which decides to either to copy the data for size<64 bytes
or use msg_t::init_data to do zero-copy initialization.
2015-06-03 22:44:33 +02:00
Pieter Hintjens
a07244cda2 Merge pull request #1414 from hintjens/master
Problem: source file headers are somewhat confusing about LGPLv3
2015-06-02 23:40:52 +02:00
Pieter Hintjens
8620c3e032 Problem: source file headers are somewhat confusing about LGPLv3
Of course people still "can" distributed the sources under the
LGPLv3. However we provide COPYING.LESSER with additional grants.

Solution: specify these grants in the header of each source file.
2015-06-02 22:33:55 +02:00
ALEKSANDR MELNIKOV
1eb8e2a69f HPUX build and gmake check issues solution (#1412) 2015-06-02 20:17:15 +06:00
Rik van der Heijden
aa2e74c2a2 Fix another degradation, CPU maxes out when POLLOUT is set because poll exits on POLLOUT and doesn't wait for POLLIN. 2015-05-27 14:27:53 +02:00
Rik van der Heijden
ae630cf3eb Fix degradation from #1382, POLLOUT was tested but not requested 2015-05-24 22:39:36 +02:00
Constantin Rack
8a3296c1aa Solution: add missing brackets 2015-05-23 23:18:10 +02:00
Brian Knox
537626258f remove temp printf from stream_engine 2015-05-19 09:38:53 -04:00
Peter Korsgaard
e89577d30c libzmq: Fix pkg-config files for static linking
Libzmq uses C++ standard library features, so users of it should link
against that as well when statically linking.

Add it to Libs.private so users using pkg-config automatically gets the
correct linker flags.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
2015-05-13 08:01:34 +02:00
Pieter Hintjens
eead3bc661 Merge pull request #1388 from rikvdh/issue-1382
Fix for issue #1382
2015-05-01 11:18:56 +02:00
Rik van der Heijden
7b9e9b838d Issue #1382: Do not send data to backend when there are no listeners (+ tests) 2015-05-01 10:39:29 +02:00
Martin Hurton
2e06737bf4 pub: Don't delay pipe termination 2015-04-28 07:48:54 +02:00
Martin Hurton
09a65c5598 push: Don't delay pipe termination 2015-04-28 07:48:04 +02:00
somdoron
c2dcc80602 change minimum version to windows vista and implement dummy condition variable for lower versions 2015-04-26 09:16:01 +03:00
evoskuil
30bd7c481e Fix integer narrowing issues. 2015-04-21 22:26:32 -07:00
Pieter Hintjens
594e3dcceb Problem: shutdown asserts if WSASTARUP wasn't done previously
This is a silly assertion that causes problems if libzmq.dll is
called in some esoteric ways.

Solution: if the shutdown code detects WSANOTINITIALISED, then
exit silently.

Fixes #1377
Fixes #1144
2015-04-20 12:53:37 +02:00
Pieter Hintjens
f053854e51 Merge pull request #1360 from hurtonm/master
Adjust number of sent messages on hiccups
2015-02-19 22:09:58 +01:00
Martin Hurton
5ed6ac60e2 Adjust number of sent messages on hiccups
Not adjusting the sent message count may lead to situation when SUB
socket does not forward its subscriptions.
2015-02-19 21:47:26 +01:00
Thomas Rodgers
700f7bfede resolve #1347 addresses issue of no metadata on identity frame 2015-02-19 13:22:32 -06:00
Thomas Rodgers
4b948b1f3b Code cleanup 2015-02-18 21:24:57 -06:00
Thomas Rodgers
cf0804bb7c Remove unused include 2015-02-18 21:10:50 -06:00
Thomas Rodgers
638ddeb404 resolve #1347 Support limited metadata for STREAM sockets 2015-02-18 12:28:58 -06:00
evoskuil
bc53d710ff Fix error return for clock_gettime. 2015-02-13 23:10:39 -08:00
evoskuil
a8f11b3c3d Add clock_gettime implementation for OSX. 2015-02-13 22:50:28 -08:00
somdoron
deaa89656f thread safety bug - lock when sending 2015-02-13 09:30:29 +02:00
somdoron
5a897f7509 allocate reaper_signal only when needed 2015-02-12 20:42:31 +02:00
somdoron
6749c9b3eb thread safety 2015-02-12 18:54:22 +02:00
evoskuil
2fd2fca531 Make implicit cast explicit. 2015-02-11 13:10:55 -08:00
evoskuil
6724f41427 Fix forcing to bool. 2015-02-11 13:09:25 -08:00
evoskuil
1a0d134666 Add missing EOL at EOF. 2015-02-04 16:27:57 -08:00
somdoron
862cd41c65 implement client socket type and drop messages when more flag is set on client and server 2015-02-02 11:59:13 +02:00
somdoron
5632b57b4a adds server socket type and routing id to msg 2015-02-02 01:17:37 +02:00
Thomas Rodgers
937d485e74 Add cast to make Clang happy 2015-01-28 16:38:14 -06:00
Thomas Rodgers
e6c45f5376 Support using compielr intrinsics for atomic operations
Works with GCC and most likely clang
2015-01-28 10:27:15 -06:00
Thomas Rodgers
d8c15cf43b Code cleanup - removing unused local 2015-01-27 12:40:35 -06:00
Thomas Rodgers
1c72bf4e55 Clean up of metadata_t
* 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.
2015-01-27 09:33:47 -06:00