Friday April 9, 2010

[Time] NameMessage
[13:52] mikko sustrik: thanks for r716f4ac!
[13:59] sustrik np, i wanted to implement it anyway :)
[13:59] mikko i'll add it to the php extension soon
[13:59] mikko is next release likely to be 2.0.7 ?
[14:00] mikko i guess i'll just add check for zeromq > 2.0.6
[14:00] sustrik well, if you want to add support for getsockopt you should chack for 2.0.7
[14:01] sustrik check*
[14:01] mikko i'm gonna check for that in any case because most of the tests fail with previous versions
[14:01] mikko due to the 'closing unused socket'
[14:03] sustrik ack
[14:12] mikko sustrik: can you up the version number in github?
[14:12] mikko it shows 2.0.6 still (naturally pkg-config check fails)
[14:20] sustrik mato: you there?
[14:20] sustrik where's the version number defined?
[14:23] sustrik there's one in; i wonder whether that's the only instance...
[14:23] mato sustrik: in, right at the top, there's a nice big comment that tells you all about it :-)
[14:23] mato yes, it is
[14:23] sustrik great
[14:24] CIA-5 zeromq2: 03Martin Sustrik 07master * r6ea76e9 10/ : version bumped to 2.0.7 -
[14:39] mikko thanks!
[14:40] mikko sorry, missed the follow-up question
[14:42] mikko checking that libzmq version is at least 2.0.7... yes
[14:42] mikko works !
[14:42] mikko awesome
[14:47] sustrik :)
[16:25] sustrik mato: can you have a look at this:
[16:25] sustrik
[16:28] mato sustrik: AGAIN?!
[16:29] sustrik it only happens if you download the github generated tar
[16:29] mato aw crap
[16:29] sustrik no idea what's the difference
[16:29] mato you're not supposed to download the github generated tar :)
[16:29] sustrik i haven't
[16:29] sustrik someone else did
[16:32] rossij If one needs to secure send messages does one have to implement outside of zeromq or is their some way to emable secure network traffic. From looking at the docs i am guessing I would have to do everything.
[16:32] mato rossij: see
[16:33] rossij mato: thank you
[16:33] mato rossij: no problem
[16:36] mato sustrik: the problem is i need some way to differentiate a "release" tarball from anything else
[16:37] sustrik what's release tarbal?
[16:37] mato one created with make dist
[16:37] mato which includes the generated documentation
[16:37] mato and thus does not depend on asciidoc being present
[16:38] sustrik one thing puzzles me: what's the difference between git clone and downloading and unpacking the tarbal?
[16:38] sustrik i would assume i get exactly the same result
[16:38] sophacles Has anyone ever used 0m one way sockets (UPSTREAM/DOWNSTEAM) in conjunction with "digital diodes"?
[16:38] sustrik 8o
[16:38] mato sustrik: the difference is that the git clone contains a .git directory so configure knows it's a git clone
[16:39] mato sustrik: if there's no git directory it assumes it's building a release tarball
[16:39] sustrik a-ha
[16:40] sustrik sophacles: what's "digital diodes"? sounds intriguing.
[16:40] mato the only way to work around it would be to have make dist include some special "stamp" file in the distribution tarball and do the detection on that
[16:40] sophacles lemme find the paper, but short version is: hardware enforced one-way packet flow
[16:41] sophacles secure network concept power grid folks (and other critical infrastructure) are looking at
[16:41] sustrik mato: it's your call
[16:41] sustrik but depending on .git sounds strange
[16:43] mato well i didn't think anyone would be grabbing tarballs that were not made with make dist
[16:45] kristianpaul hola :)
[16:45] sustrik hola
[16:48] kristianpaul what is luuid ?
[16:48] kristianpaul i cant compile zeromq because of that
[16:48] sustrik what's your system?
[16:49] kristianpaul debian squeeze
[16:49] mato kristianpaul: apt-get install uuid-dev
[16:50] mato sustrik: can't we just disable the 'download source' link on github, or (better) redirect it to the download page?
[16:50] mato sustrik: that would make the problem go away
[16:50] sustrik mato: thx
[16:50] mato sustrik: if people want to use the git version then they should just clone it
[16:50] sustrik can we?
[16:50] mato i don't know
[16:50] sustrik never seen such a button
[16:54] mato looks like it can't be done :-(
[16:57] kristianpaul compiled :)
[16:59] kristianpaul i cloned this git:// and was okay
[16:59] sophacles sustrik: for the digital diode paper we are looking at
[17:01] sustrik sophacles: i give it a look later on
[17:01] sustrik however, with one-way communication TCP is not possible, right?
[17:04] sophacles sustrik correct, hence our looking at middleware that has one-way semantics.
[17:05] sophacles so then we can have a box that on one interface has a sock_downstream listening, and pushes the data over another interface which is a hardware diode
[17:05] sustrik sophacles: 0MQ is potentially capable of that but you must write true udp transport for that
[17:06] sustrik true udp = unrealiable, no acks no nacks
[17:06] sustrik couple of people already announced that they want to implement it
[17:06] sustrik but obviously nobody have even started with it
[17:07] kristianpaul can i install zeromq i a 8 bit microcontroller?
[17:07] sustrik kristianpaul: no idea
[17:07] kristianpaul it uses sdcc
[17:07] kristianpaul ok
[17:08] sustrik you need at least c++ runtime, pthreads and libuuid
[17:08] kristianpaul }:/
[17:08] kristianpaul then not :)
[17:08] sophacles this is why im looking at the gateway concept. it provides realiability to the place where the diode is implemented, after the diode there is no reliablility of course, but hte assumption is the hard part is reliability to the diode and ont he other side of the diode
[17:09] sustrik then just implement udp transport in 0mq and you can place a forwarder device on each end of the diode
[17:10] sustrik tcp->forwarder->udp->diode->forwarder->tcp
[17:10] sophacles right, thats why we like 0mq for this :)
[17:10] sustrik implementing udp shouldn't be a big deal
[17:11] sustrik if you decide to go that way let me know and i'll point you in the right direction
[17:11] sophacles ok, thanks, for now we have enough for our proposal tho :)
[17:11] sustrik goodo
[17:16] CIA-5 zeromq2: 03Martin Lucina 07master * r1d28dc9 10/ :
[17:16] CIA-5 zeromq2: Fix for Issue #14
[17:16] CIA-5 zeromq2: Don't fail hard if an unreleased tarball is being built and asciidoc is
[17:16] CIA-5 zeromq2: not installed; instead just print a big fat warning -
[17:25] mato sustrik: ok, fixed, i decided not to care and just print a big fat warning if someone has a tarball from git/git clone and no asciidoc
[17:26] sustrik sure, np
[17:26] mato the github issue tracker is a bit sucky
[17:26] mato i seem to get random emails
[17:26] mato ah
[17:26] sustrik yes, people do
[17:26] sustrik get random emials
[17:27] mato yes, but the reason the emails look random is because you're entering issues with no structure
[17:27] mato e.g. GH-13
[17:27] sustrik eee
[17:27] mato I get an email with a looooong subject, and a body that says "Here's the test program"
[17:27] mato which is totally unparseable at 1st glance :-)
[17:28] sustrik what's the magic then?
[17:28] mato we should have a template for issues
[17:28] mato so that the body of the email actually makes sense
[17:28] sustrik ah
[17:28] sustrik lemme see
[17:28] mato but i'm not sure if github can do that, i know that google code can
[17:29] sustrik still i don't get why emails are send _randomly_
[17:29] mato me neither
[17:29] sustrik can we turn it off?
[17:29] mato don't know
[17:29] mato i'll look into it and try and file bugs/contact the github people
[17:29] mato it's very hard to follow the way it works now
[17:30] sustrik hm, i wanted to ask you to kill the jira
[17:31] sustrik but it seems it's maybe to early for that
[17:31] mato maybe
[17:31] sustrik but you can definitely kill the old svn
[17:31] sustrik nobody using it anymore
[17:31] mato not my job, email ewen
[17:31] sustrik ok
[17:31] mato but i suggest you wait until we can kill the jira
[17:32] mato then you can get ewen to just kill the entire svfastmq vm
[17:32] mato in one go
[17:32] mato i will try and talk to the github people to see if we can get some improvements made to the issue tracker
[17:32] sustrik ack
[17:32] mato templates that users must fill in when submitting would be especially good
[17:33] mato "Please describe your Operating System, version," and so on
[17:33] mato otherwise it'll be a total mess
[17:33] sustrik yup
[17:34] dirtmcgirt sustrik: about the java problem
[17:35] dirtmcgirt sustrik: OS is OS X
[17:35] dirtmcgirt getting a "Exception in thread "main" java.lang.UnsatisfiedLinkError: /usr/local/lib/libjzmq.0.dylib: no suitable image found." with the latest checkout of jzmq
[17:35] dirtmcgirt trying the test program: "java -Djava.library.path=/usr/local/lib -classpath ../src/Zmq.jar:. local_lat tcp:// 1 100"
[17:35] sustrik is there the library on the path?
[17:35] dirtmcgirt yep
[17:36] sustrik /usr/local/lib/libjzmq.0.dylib
[17:36] sustrik what does "no suitable image found" mean then?
[17:37] dirtmcgirt going to try an rebuild from scratch
[17:38] mato sustrik: dinnertime, cyl in kut?
[17:38] sustrik yes
[17:41] sustrik dirtmcgirt: i cannot help you much as i have no experience with osx, however, try asking at the mailing list, there are definitely OSX people there
[17:58] dirtmcgirt dirtmcgirt: no known problems on linux (CentOS 5)?
[17:58] dirtmcgirt oops, sustrik
[17:58] sustrik i am not aware of any
[20:36] sophacles interesting: on host A - an ubuntu maching, the toy client and server from the webpage compile just fine, on a debian machine tho, the client breaks with errors about context_t and socket_t not being members of zmq
[20:36] sophacles anyone seen this before?
[20:45] sophacles hahaha nevermind -- the deb box had an ancient copy of zeromq1 sitting on it
[22:19] dirtmcgirt any binary builds for java?
[22:19] dirtmcgirt jar files?
[22:24] dirtmcgirt getting build problems on CentOS 5
[22:24] dirtmcgirt make[2]: util.hpp: Command not found