[zeromq-dev] zmq_msg_move

Andrew Hume andrew at research.att.com
Wed Aug 1 16:03:38 CEST 2012

this just needs to be decided.

i propose that you should need to zmq_close, and furthermore,
that you need to zmq_close a msg you have zmq_msg_init()ed
in the case you do not do a zmq_recv on it.

my reasoning is thus:

1) despite your claim below, the manual actually says "initialise ... msg to represent an empty massage".
2) no where is there any defn of an empty message.
3) however, intuitively, one feels that an empty message is pretty much teh same as one of size 0,
	so that these should be equivalent:

		zmq_msg_init_size(&msg, 0);

	(this is reminiscent of the debates about whether malloc(0) allocates memory.
	it certainly might; the point is that malloc(n) returns enough space to store n bytes
	and not necessarily exactly n bytes.) similarly, zmq_msg_init_size says it allocates
	enough resources to store a message of n bytes; that implies to me that there
	may be allocation involved even if n is 0.

4) it certainly is plausible (even likely), that zmq_msg_init_size(&msg, 0) will
	allocate storage, and so you need the sequence

		zmq_msg_init_size(&msg, 0);
	in order to prevent memory leaks.
5) by combining 3) and 4), it seems you need to do

		zmq_msg_init(&msg, 0);
	in order to prevent memory leaks.
6) and by extension, you would need to do

		zmq_msg_move(&dmsg, &msg);
	in order to prevent memory leaks.

On Jul 31, 2012, at 10:40 PM, Pieter Hintjens wrote:

> On Wed, Aug 1, 2012 at 10:10 AM, Andrew Hume <andrew at research.att.com> wrote:
>> what is exactly the status of src after calling this function?
>> the manual page says it becomes an empty message, whatever that is.
>> do i need to call zmq_msg_close(src)?
>> is it the same as calling zmq_msg_init(src)?
> zmq_msg_init says it creates an empty message, i.e. a message of size
> zero, so that seems consistent.
> It's unclear whether you need to call zmq_msg_close on an empty
> message. The reference manual doesn't discuss this. One would assume
> it's not needed, but that may be dependent on how messages are
> implemented.
> -Pieter
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev

Andrew Hume  (best -> Telework) +1 623-551-2845
andrew at research.att.com  (Work) +1 973-236-2014
AT&T Labs - Research; member of USENIX and LOPSA

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20120801/0204de9a/attachment.htm>

More information about the zeromq-dev mailing list