IRC Log


Monday April 25, 2011

[Time] NameMessage
[00:49] phantomcircuit hello, so reading the documentation it strikes me that the pull/push mechanism results in pull sockets having local queueing
[00:49] phantomcircuit that seems like an odd decision
[01:35] dublisk Hi, zeromq will only work with itself right? I can't use it to read from sockets that aren't zeromq on the other end?
[01:50] taotetek dublisk: correct
[01:50] dublisk D'oh
[01:50] taotetek dublisk: you can of course write something to translate from a tcp socket to zeromq
[01:51] taotetek dublisk: but yeah, you can't connect a zeromq socket and standard socket together without some glue in the middle.
[01:52] dublisk I was hoping I could just receive messages which would be of whatever length it captured
[01:56] taotetek dublisk: I do quite a bit of zeromq work with python - when I need to move data between a standard tcp connection and zeromq I usually write a little connecting piece w/ python and eventlet, works pretty well
[01:56] taotetek dublisk: don't know what languages you're working with though
[01:56] dublisk C++, but how do you listen to the socket?
[01:57] taotetek dublisk: I open a tcp port, anything that connects to that port and sends gets forwarded out a zeromq socket (breaking messages apart on \n)
[02:05] taotetek dublisk: it's worked as a decent stopgap while I wait for people above stream from me in our stack to convert over to zeromq
[04:37] Taaz hii iam new to it...i would love to contribute..how do i ??
[15:55] phantomcircuit is there a binary build available for windows
[15:55] phantomcircuit ?
[15:56] Guthur phantomcircuit, not to my knowledge, but it's pretty trivial to build
[15:56] Guthur should be buildable with the express edition
[15:57] phantomcircuit yeah i've gotten it built
[15:57] phantomcircuit i cant get this to work it'll just take some doing
[15:57] Guthur what is the issue/
[15:57] Guthur ?
[15:59] phantomcircuit setup.py is trying to build a test program and is failing because it cant find a compiler
[15:59] jhawk28 phantomcircuit: they ship a windows binard as part of the pyzmq
[15:59] jhawk28 download the msi from https://github.com/zeromq/pyzmq/downloads
[16:00] phantomcircuit im gonna need to build php bindings also
[16:00] jhawk28 If you want to compile it yourself, download the Visual Studio 2008 Express
[16:01] jhawk28 http://www.microsoft.com/express/Downloads/#2008-Visual-CPP
[16:02] phantomcircuit yeah i've got that and i've actually built libzmq.lib
[16:02] jhawk28 You can use 2010 too, but all the projects are already set up for 2008
[16:03] jhawk28 when you distribute your libraries out to the servers, you need to make sure you have the VC++ redistributables that match your compiler environment
[16:03] jhawk28 kinda like http://www.microsoft.com/downloads/en/details.aspx?FamilyID=9b2da534-3e03-4391-8a4d-074b9f2bc1bf&displaylang=en
[16:04] jhawk28 there are about 3 or 4 version out there
[16:04] phantomcircuit jhawk28, basically this is just a proof of concept, if this works i will not continue using windows servers
[16:05] phantomcircuit this one just happens to be free
[16:06] Guthur I did have a minor issue build the .lib file with 2010
[16:06] Guthur the .dll was fine though
[16:07] Guthur there is a website somewhere detailing the issue, 2 secs
[16:07] phantomcircuit well both are built
[16:07] Guthur http://www.mansysadmin.com/2011/03/using-zeromq-framework-with-visual-studio-2010-tutorial/
[16:07] phantomcircuit ah well im missing the python headers
[16:07] Guthur possibly only a 2010 issue
[16:09] phantomcircuit lol visual studio is so slow
[16:09] phantomcircuit it's like a bad joke
[16:09] Guthur it gets even funnier when you start to ask what do you get for the slowness
[16:11] phantomcircuit code completion
[16:11] phantomcircuit and uh
[16:11] phantomcircuit yeah that's it
[16:12] jhawk28 even the code completion, people get resharper...
[16:13] Guthur yep, it's code completion pretty naff
[16:18] phantomcircuit hmm ok so now how do i get php bindings
[16:30] phantomcircuit hmm getting a php extension built is proving to be a challenge
[16:34] phantomcircuit i take it that nobody cares
[16:35] Guthur well now that you ask... I've never cared much for PHP
[16:37] phantomcircuit i've got a giant application written in php, currently a long running computationally expensive part of it is run in the context of the web server with a bunch of hilarious hacks
[16:37] Guthur there is a few PHPers about here on occasion
[16:38] phantomcircuit i cant really rewrite the entire platform but i can use zmq to send the work requests to python daemons
[16:39] Guthur phantomcircuit, you could try the mailing list
[16:39] Guthur i know Ian Barber is a heavy PHP user but isn't online at the moment
[16:39] Guthur you might catch him on the ML
[16:41] phantomcircuit k thanks
[17:31] mikko phantomcircuit: what is the problem with php extension?
[17:32] phantomcircuit i have no idea how to build it on windows
[17:32] phantomcircuit and i dont see a prebuilt version
[17:33] mikko have you built any other php extensions on windows?
[17:33] phantomcircuit nope
[17:34] mikko so i take it that you don't have php development env setup?
[17:34] mikko https://wiki.php.net/internals/windows/stepbystepbuild
[17:36] phantomcircuit mikko, even if i did i dont see any build instructions for zmq specifically
[17:37] mikko you would build it just like any other pecl extension
[17:42] cremes anyone working on getting 0mq to build on windows using mingw?
[17:45] mikko cremes: it should build
[17:45] cremes mikko: i'll give it a try today and report any errors i see
[17:46] mikko cremes: there is daily build on win7 using mingw32
[17:47] cremes oh, excellent
[17:47] mikko the windows build machine is a bit out of order atm though
[17:47] mikko but overall mingw32 should build without problems
[17:47] mikko pgm wont build
[17:47] mikko openpgm*
[17:48] cremes ok; i don't need pgm so that isn't a show stopper for me
[17:49] mikko phantomcircuit: seems like my build environment still works for win
[17:50] mikko might be able to get a build done
[17:52] mikko phantomcircuit: http://valokuva.org/builds/ext/vc9/nts/zmq/2011-04-25_1849/php_zmq_nts.dll / http://valokuva.org/builds/ext/vc9/ts/zmq/2011-04-25_1849/php_zmq_ts.dll
[19:43] jhawk28 any iMatix people on?
[19:45] jhawk28 pieterh
[19:45] jhawk28 pieterh: are you there?
[19:50] jhawk28 mato: you there?
[19:53] cremes jhawk28: if you have a general question about 0mq, go ahead and ask it
[19:53] cremes there are a bunch of knowledgeable folks in the channel
[19:53] jhawk28 cremes: its releated to the OpenAMQ JMS client
[19:54] cremes ask anyway; maybe a lurker has an answer for you
[19:58] jhawk28 my management has "dictated" that we must use JMS
[19:59] jhawk28 based on my research, it looks like the qpid amqp driver does not interoperate with RabbitMQ based on a "disagreement" in the spec
[19:59] jhawk28 the other option is the openamq-jms
[20:00] jhawk28 LShift looks to have got it running based on http://www.lshift.net/blog/2009/03/16/openamqs-jms-client-with-rabbitmq-server
[20:01] jhawk28 no more chatter can be seen on the subject
[20:01] jhawk28 no more commits have been made to the project: https://github.com/imatix/openamq-jms
[20:01] jhawk28 is it dead or just so stable it hasn't needed updates?
[20:03] mato jhawk28: yes?
[20:03] mato jhawk28: ah, openamq-jms
[20:04] mato jhawk28: it's basically dead
[20:05] jhawk28 I hate enterprise products
[20:06] mato :-)
[20:06] jhawk28 JMS is so verbose
[20:07] jhawk28 but yet doen't really have good error messages on what actually went wrong
[20:10] mato convince your management to use zeromq :-)
[20:11] jhawk28 we are
[20:11] jhawk28 need a durable queue for some more enterprisey things
[20:11] jhawk28 slower peformance, etc
[20:12] jhawk28 of course, I'm getting to the point where the JNA is biting us
[20:12] mato well, you could always roll your own on top of 0mq
[20:12] jhawk28 thought about it
[20:13] jhawk28 thinking I'll pursue writing a Java impl of 0mq first so that it can run in the "Enterprise Containers"
[20:15] jhawk28 I'm sure that it will be a pain since it doesn't look like the wire format is documented well
[20:15] jhawk28 but, it would be really nice to just have a Java library
[20:20] mato jhawk28: is there a particular reason why you can't just use the 0mq Java API?
[20:21] mato jhawk28: you really really don't want to go writing your own implementation, the 0mq protocols (especially the underlying semantics) are still too raw and unstable
[20:23] jhawk28 mato: in many things, the 0mq Java API is fine. Once you try using it in crappy enterprise products like JBoss, the Threading model conflicts and things like EJB's forbid using JNA
[20:23] mato jhawk28: JNA? you mean JNI?
[20:23] jhawk28 oops, yes, JNI
[20:24] mato jhawk28: is there anything that can be done on the jzmq side of things to help use in your conditions?
[20:26] jhawk28 2 approaches that I can think to fix the issue - write a remoting wrapper so the JNI can live outside the container, write a native Java impl of 0mq so that it can live inside the container
[20:26] mato i would recommend the former if it's not too hard
[20:27] jhawk28 that would be my quick and dirty first pass
[20:27] jhawk28 since it wouldnt take too long to get up and running
[20:28] jhawk28 I know there are some developments on the 0mq side that may fix the issue too like adding it to the posix spec
[20:29] mato jhawk28: well, yes, a kernel-mode implementation of 0mq would make your problem go away
[20:29] mato jhawk28: but that is still a long way away
[20:29] jhawk28 agreed, these things take time