Static analysis says:
src\zmq.cpp(489): error V220: Suspicious sequence of types castings: memsize -> 32-bit integer -> memsize. The value being casted: '* count_'.
src\zmq.cpp(510): error V127: An overflow of the 32-bit 'nread' variable is possible inside a long cycle which utilizes a memsize-type loop counter.

I've silenced the warning on line 489 and ignored the other.
But also, it looks to me like there's a serious bug here: The
out-parameter "count_" is never set to zero before we start
incrementing it. So its final value will always be between
1 and 2 times its initial value. The fix seems obvious.

[LIBZMQ-414] Fix an error in the inline assembly for Thumb2.

Fix a bug reported on the mailing list.

Both memcmp and strcmp return zero on equal, nonzero on nonequal;
so all of these tests were backwards.

The original committer fixed the failure by comparing 22 bytes instead
of the correct 21, so that the assertions would trigger only if the
22nd byte happened to match exactly --- which was rare.

The correct fix is to compare the right number of bytes with the
right sense.  (I think all of the ".addr" fields are null-terminated,
in which case it's more appropriate to use strcmp throughout.)

Notice that ZeroMQ has never been compiled for Thumb2 before,
and I personally don't make any guarantees that it will actually
behave correctly once compiled. But after this patch, it is at
least *possible* to compile it for Thumb2.

(Thumb2 is the target for most iOS devices.)

Exchange greeting messages for all socket types

Fixes #415.

Revert zmq_poll NULL poll items check to 2.2 behavior - let the poll items count filter out empty poll sets and not return a sometimes unexpected EFAULT error status

Fix reordering compile errors

Fix invalid address metadata for ZMQ_EVENT_DISCONNECTED

Fix issue #406

When a peer reconnects, the router socket receives an identity
message containing this peer id. When this happens, the current
implementation crashes.

This patch makes a router socket to silently ignore all identity
messages coming from reconnected peers.

Fix LIBZMQ-404: zmq_term not truly re-entrant

zmq_term can not safely be re-entered with pgm transport.
Fix proposed by Steven McCoy.

Correction for LIBZMQ-335

That way it can be used more appropriately between processes.

By assigning a SECURITY_DESCRIPTOR to the event we gain the ability to
share it between service and console programs. We also added
EVENT_MODIFY_STATE as a requirement to OpenEvent so we can SetEvent later
in the method.

Revert zmq_poll NULL poll items check to 2.2 behavior - let the poll items count filter out empty poll sets and not return a sometimes unexpected EFAULT error status

Small code cleanup

Fix issue #397

When closing an ipc listener, the library may try to unlink
the associated file. When this fails, the underlying
socket is not marked as retired and this triggers
assertion failure.

Fixes issue #397

Be more conservative about when to generate ZMQ_EVENT_CLOSE_FAILED

This also fixes a bug in tcp_connecter and tcp_listener, which
generated the event not when they failed to close the socket but
when the succeed to close it.

Small cleanup

Rewrite event processing in io_thread

Code cleanup

Reverting fix for #393

Fixed issue #393 - Android build error

Fixed HWM doc on ZMQ_PULL

Making Android&pyzmq friendly version with -avoid-version