Tuesday April 26, 2011

[Time] NameMessage
[01:49] hhummel I just wandered in. I'm having trouble installing zeromq on fedora. Is there a particular directory I should unpack into so that the linker can find the libraries?
[02:16] hhummel I'm having trouble linking a simple c++ "hello world" app. I know I'm supposed to be linking to .lib files, but I don't see them. Can anyone help? I'm using C++ on Fedora 13.
[03:41] Dantman Anyone got tips for solving an UnsatisfiedLinkError for the jzmq library... even though everything seams to be in order...
[03:42] Dantman java.library.path is /usr/local/lib, double checked that... That dir contains libjzmq.{a,la,so,so.0,so.0.0.0}, libzmq.{a,la,so,so.1,so.1.0.0}
[03:50] Dantman Anyone compiled jzmq on 64bit linux?
[04:51] phantomcircuit Dantman, i've been trying to build php-zmq pretty much all day
[07:28] djc pieterh: is there a reason isn't out yet?
[08:47] drbobbeaty pieterh: any news on the revised (2.1.6?) release? With all the nice fixes I've seen in the mailing list, it'd be nice to see them in a 2.1.x release. Along with the other fixes, of course. Just curious if you had an estimate yet.
[11:55] pieterh re
[11:56] pieterh djc: no reason except it was lovely weather and holidays
[11:56] pieterh ... will make the release later today
[11:57] pieterh drbobbeaty: I've one patch to apply, but that will mean delaying the release by a day or two
[11:57] drbobbeaty pieterh: thanks for letting me know. Very much appreciated.
[12:41] mato pieterh: why the extra .1 ?
[13:05] toni hi there. I m using pyzmq. Theres a method socket.send_pyobj(obj) which serializes the object and a socket.recv_pyobj(). Is there also something like socket.send_pyobj_multipart when explicit addressing is required? or does sending/receiving of objects only target on socket which do not use envelopes?
[13:06] guido_g simple message, no routing, no multipart
[13:07] mato send_pyobj() is simple code though, so you could always reimplement it yourself if you want to do multipart
[13:08] toni ok, thanks...
[13:10] toni mato: do you know where the code for send_pyobj lives?
[13:11] mato toni: hang on, they've moved things around since i last looked
[13:11] guido_g
[13:11] toni thanks!
[13:12] mato that's the one...
[13:26] pieterh mato: hi
[13:26] pieterh you missed something that passes for excitement in this little community
[13:30] pieterh
[13:31] pieterh if you have a better suggestion than sticking an extra '.1' at the end, let me know asap
[13:31] pieterh this was kind of the consensus solution here on IRC
[13:32] pieterh mato: ping
[13:42] hhummel Hi folks. Thanks for making this terrific product! I know this is a newbie question, but I can't figure it out. I'm trying to get started building a stripped down app, and am getting errors that indicate that I'm not linking in the library. How do I do that using g++ on Fedora? I'm trying: [harry@mars zeromq]$ g++ -Wall -L /usr/include/zeromq-2.1.4/src/.libs client2.cpp but it doesn't help. Can anyone help?
[13:43] NikolaVeber have you tried -lzmq ?
[13:45] hhummel Wow, that worked! What is -lzmq?
[13:45] NikolaVeber parameter telling g++ to use the 0mq library installed systemwide
[13:46] NikolaVeber you can probably remove the -L part completely
[13:46] NikolaVeber gcc -lzmq source.cpp
[13:48] hhummel That does the trick, thanks. I assume -lzmp is somehow introduced as part of the build process?
[13:49] NikolaVeber it links to the installed zmq libraries
[13:49] hhummel I really appreciate the hand. Thanks!
[13:50] NikolaVeber np :)
[14:11] mato pieterh: yes?
[14:11] pieterh mato: reason for the extra .1...
[14:11] mato pieterh: will look at it in a sec, thx for pointer
[14:12] pieterh ok, no hurry... would like to release 2.1.5 today but we have time
[14:12] pieterh btw czmq is a much nicer name, thanks for insisting
[14:12] CIA-75 jzmq: 03Gonzalo Diethelm 07master * rb5037aa 10/ (src/Socket.cpp src/org/zeromq/ (log message trimmed)
[14:12] CIA-75 jzmq: Added support for 0MQ 3.0. Everything should still work with previous
[14:12] CIA-75 jzmq: versions of 0MQ. Changes were:
[14:12] CIA-75 jzmq: 1. Added all missing socket options.
[14:12] CIA-75 jzmq: 2. Use zmq_sendmsg() and zmq_recvmsg().
[14:12] CIA-75 jzmq: 3. Added DONTWAIT as synonym of NOBLOCK.
[14:12] CIA-75 jzmq: Could use a lot of testing with different versions of the 0MQ library,
[14:13] mato pieterh: i'd just call it 2.1.6 and ignore 2.1.5
[14:13] pieterh mato: if you're happy with that, it's much easier for me
[14:13] mato pieterh: yes, i'm happy with that
[14:14] mato normal procedure, i've seen it elsewhere
[14:14] pieterh ok, so I can apply the patch martin sent today, and I'll mark 2.1.5 as "broken"
[14:14] mato yes
[14:14] pieterh nice to have you back
[14:14] mato ja, looks like i'll be more active in zmq again
[14:15] mato also time to do some work after lots of holidays
[14:15] pieterh I will probably be less involved after the summer
[14:15] mato btw saying that i arrive late to threads is a bit unfair :)
[14:15] pieterh well, it's kind of a pattern... :)
[14:15] mato it was only 5 days, *and* i was on a boat in the med
[14:15] pieterh diaf
[14:15] pieterh on a boat in the med...
[14:16] mato plus, referring to irc discussions is about the same as saying "we discussed it over beer last night, you weren't there" :-)
[14:16] pieterh mato: sincerely, this is how it goes, we have an issue, we discuss on email and irc, and move quickly to a solution
[14:16] mato yeah, did a circumnavigation of corsica
[14:16] mato pieterh: there's no "hurry" :)
[14:16] pieterh mato: IMO there is a hurry, yes
[14:17] mato anyway, nice we came to a good conclusion
[14:17] pieterh but mainly it's more efficient to make two, three fast cycles than one slow one
[14:17] mato sure, but there is no hurry regarding ZAPI i mean
[14:17] pieterh there is a hurry
[14:18] pieterh I've been struggling with an undocumented ad-hoc API in the Guide for half a year
[14:18] pieterh it's an unavoidable problem
[14:18] pieterh and everyone who learns & translates zhelpers is getting trapped
[14:18] mato well, C *is* a low-level language...
[14:18] pieterh not really
[14:19] pieterh i can write full *complex* apps in a few hundred lines of code
[14:19] pieterh it's all about sensible abstractions
[14:19] mato yes, but you invent your own object & type/container system on top of C
[14:19] mato this is a common pattern
[14:19] mato most substantial C project do that
[14:19] pieterh so, if we don't offer a reusable set of classes, people will reinvent their own
[14:19] mato which is *fine*
[14:19] pieterh i'd like to see more consistency between bindings
[14:19] pieterh it's important for human portabilty
[14:20] pieterh *portability
[14:20] pieterh e.g. how to set a socket option
[14:20] mato this is exactly my point; we are not out here to give people a C object/type/portability API
[14:20] pieterh well, the C API is just one of many but it is my gift, yes
[14:20] mato yeah, but you see my point
[14:21] pieterh let's say it gives me the authority to speak as a binding author
[14:21] pieterh I know what my users (who happen to be me) need
[14:21] pieterh (but not only me, even now)
[14:21] mato sure
[14:21] mato but you see the difference between what you're building and what the native API is there for
[14:21] mato right?
[14:21] pieterh have you seen the wiki page now?
[14:21] pieterh I wanted your review of that, I think it's getting much better
[14:22] mato hang on, doing multiple things at once
[14:22] pieterh this layering is essential, subtle
[14:24] mato well, i kind of see where you're going, libzmq + stuff = zeromq distribution
[14:24] pieterh ja
[14:24] pieterh this is what the user needs
[14:24] mato however that's not the case right now
[14:24] mato people equate libzmq with zeromq
[14:25] pieterh one step at a time... it's enormously hard to get this kind of change
[14:25] mato i'd be happier calling the distribution something else
[14:25] mato in fact
[14:25] mato that's a business opportunity hiding for you
[14:25] mato standard stuff
[14:25] pieterh the key is here is (1) community, (2) community, and (3) community
[14:25] pieterh "zeromq = libzmq" is bad for the community
[14:26] pieterh I think that's self-evident
[14:26] mato why?
[14:26] pieterh you really don't see?
[14:26] mato i'd like your view
[14:26] pieterh because it excludes about 90% of potential and actual contributors to what you call "zeromq"
[14:27] pieterh whereas the value to a contributor is to be part of the "zeromq" project
[14:27] pieterh not "the" but "a big part of the"
[14:27] pieterh you effectively turn all projects except libzmq into 2nd class citizens
[14:28] pieterh while at the same time making it *really* hard for people to contribute to libzmq
[14:28] pieterh surely you are aware of this?
[14:28] pieterh it is a toxic form of elitism that will backfire sooner or later
[14:28] mato hey, i'm also learning
[14:29] pieterh ok, don't mean to be critical, I've been down this road often before
[14:29] CIA-75 jzmq: 03Gonzalo Diethelm 07master * r623f6b9 10/ :
[14:29] CIA-75 jzmq: Merged pull request #36 from MrEvil/master.
[14:29] CIA-75 jzmq: Added SHELL variable to -
[14:29] pieterh how come jzmq commits are appearing here?
[14:29] mato and i am aware of these things, and hopefully they will change for the better
[14:29] mato they always were
[14:29] pieterh it's nice, we should expand this to more zmq projects
[14:29] mato and libzmq commits should also but i forgot to update CIA for the name change
[14:29] pieterh yeah
[14:30] pieterh my vision, which is always populist and anti-elitist is that we have a community of projects
[14:30] mato any project that has a VC repo can be monitored
[14:30] pieterh obviously layered, and obviously with different (earned) statuses
[14:30] pieterh anyone can join, anyone can compete, at any level
[14:30] pieterh the ability to compete is essential
[14:31] pieterh if I don't like implementation of X, I have the right (and information) to make X-prime
[14:31] pieterh thus, GPL license, proper layered specs, etc.
[14:31] pieterh any layer that escapes this competition will become corrupt over time
[14:31] mato of course. if you don't like the core API, you can fork it and users (may) follow with their feet
[14:31] pieterh yes
[14:31] pieterh precisely
[14:31] pieterh which is why we need proper WLP specs
[14:32] Guthur GPL?
[14:32] mato look, it's your modues operandi
[14:32] pieterh mato: well, the zeromq community follows my design, yes
[14:32] Guthur I thought libzmq was LGPL
[14:32] pieterh but I'm stating principles that are afaics scientifically provable
[14:32] mato but that's irrelevant
[14:32] pieterh Guthur: LGPL is GPL with an extra clause or two, same thing essentially
[14:33] pieterh mato: discussing the design ideology of a thing is relevant
[14:33] pieterh when we discuss that thing
[14:33] pieterh do recall that the whole 0MQ community was a conscious design from day 1
[14:34] mato over beer one day
[14:34] pieterh ok, but bear with me...
[14:34] Guthur GPL has a lot of stigma for some developers though
[14:34] pieterh what it means in terms of definitions
[14:34] pieterh is that "zeromq" is the community, not the core library
[14:34] pieterh this is perhaps hard to see, from some perspectives
[14:34] pieterh but it's my view, and it's based on experience of what will work and not work
[14:35] pieterh let me... give you some other examples
[14:35] pieterh of successful communities that have done this
[14:35] pieterh apache
[14:35] pieterh xmpp
[14:35] pieterh linux
[14:35] Guthur eek, just realised clrzmq2 seems to be published under GPL
[14:36] pieterh Guthur: if you're the sole author you can relicense any time you like
[14:36] pieterh bsd
[14:36] Guthur well, it's slightly complicated I orginally based it of clrzmq
[14:36] pieterh Guthur: and that was licensed as what?
[14:36] Guthur but it bears very little resemblance to that now
[14:37] Guthur pieterh: hehe I copied the license agreement from that
[14:37] pieterh Guthur: irrelevant, unfortunately, unless you can explicitly remove all possible derived code
[14:37] pieterh so clrzmq2 has to be GPL too...
[14:37] mato pieterh: sure, i guess this is not clear
[14:37] pieterh you can make a new clean-room implementation, presumably, quite straight-forward
[14:37] Guthur pieterh: well that is the probably with software development, there is only one way to do somethings
[14:37] mato pieterh: something to be discussed in brussels
[14:38] Guthur I can't possibly remove it
[14:38] pieterh mato: sure, much easier over beer... :)
[14:38] Guthur probably/problem
[14:38] mato pieterh: it's still more of a "beer" discussion than anything else
[14:38] mato ideology as opposed to "here is code that does stuff"
[14:38] mato it's fun to see the personalities at work
[14:38] Guthur expecially foreign function calls
[14:38] pieterh mato: put it like this, I'm responsible for the 0MQ community design and implementation, and I see it as "code that does stuff"
[14:38] pieterh things like legal frameworks are social code
[14:38] pieterh that's my lawyer hat
[14:39] mato you just like doing all this stuff :)
[14:39] pieterh mato: yes, that's true, and I also intensely dislike environments where it's badly done
[14:40] pieterh well, you hate those too
[14:40] Guthur the other authors where Martin and someone called Jeffrey Dik
[14:40] pieterh I don't need to remind you of weekly wednesday phone conferences...
[14:40] mato sure
[14:40] pieterh Guthur: I'd contact them all and ask if you can relicense their work
[14:40] pieterh they have to explicitly grant you a new license
[14:40] mato anyhow, i guess my thing is that the zeromq != libzmq thing is not clear
[14:41] pieterh mato: I know, but this should in theory become clearer over time
[14:41] pieterh renaming the project to 'libzmq' was a large step
[14:41] mato and when i see you discouraging use of the core API, i get annoyed since from my point of view (and users i have spoken to), the core API is *good*
[14:41] mato i realise you don't like it
[14:41] pieterh ah, mato... I didn't intend to discourage it
[14:41] pieterh look, it has some real flaws
[14:41] mato but you've been doing that quite persistently
[14:41] pieterh no, I've been trying to improve it quite persistently
[14:42] pieterh when my suggestions are persistently discarded, what do I do?
[14:42] mato make your own
[14:42] pieterh exactly
[14:42] mato which is fine
[14:42] mato but please do not tread on other people who actually like the current API
[14:42] mato that is not fair
[14:42] pieterh over-sensitive, maybe
[14:42] pieterh I don't think I treaded on anyone
[14:42] mato and it's abusing your position somewhat IMO
[14:42] pieterh if I did, apologies, sincerely...
[14:43] pieterh IMO you may be seeing intent where there is none
[14:43] mato possibly, but even sans the intent i see the resulting confusion
[14:43] mato which is why i spoke up now
[14:43] pieterh if I want to tread on the current API, trust me, I won't be subtle about it
[14:43] pieterh am I a subtle person?
[14:43] mato i would have earlier, but i would have got yelled at that there is no problem :)
[14:43] pieterh srsly
[14:44] pieterh I have a long list of things to improve with the current API
[14:44] pieterh what would you like me to do with that list?
[14:44] mato submit patches?
[14:44] pieterh sorry, can't do that (not competent)
[14:44] mato ask el sustrik, he is the boss on master
[14:44] pieterh I'm a user
[14:44] mato lol
[14:45] mato then do what you're doing now
[14:45] mato which is fine
[14:45] Guthur pieterh, Sorry to drift a little OT, but would it be possible for me un-derive the work, not that I actually want to
[14:45] pieterh Guthur: you need to make a clean room implementation
[14:45] Guthur or by looking at the original code has be thought process been tainted
[14:45] Guthur be/my
[14:45] pieterh Guthur: you document what the binding does, in written form, as detailed as you can but without code
[14:45] pieterh you give that to *someone else* and bribe them to make a new version
[14:45] pieterh at least a skeleton
[14:45] mato pieterh: you say you will be less involved after summer, why's that?
[14:45] Guthur lol
[14:46] pieterh mato: I'll explain over beer, not here
[14:46] Guthur legal stuff is bonkers
[14:46] mato incidentally, are you still planning on coming to bts before the brussels conf?
[14:46] pieterh Guthur: well, when you reuse someone else's code, that's the cost
[14:46] pieterh mato: no, after that
[14:46] mato right...
[14:46] mato btw how far is charleroi from brussels ?
[14:46] pieterh 1hour, it's a decent connection
[14:46] mato ryanair seems to be the only sane cheap option
[14:47] pieterh yes, it's sane and cheap
[14:47] pieterh except the return flight is bonkers, it's at 7.30 am or so
[14:47] mato it was showing me something during the day
[14:47] mato maybe they've changed
[14:47] pieterh I have a crazy idea... how about we drive back to BTS together?
[14:47] pieterh maybe changed...
[14:47] pieterh ah, that was perhaps the winter schedule...
[14:47] mato :-)
[14:48] pieterh 12 hours of argument :)
[14:48] pieterh fuuuuuuuuu.....
[14:48] mato night train works too
[14:48] mato 13 hours of argument *and* beer
[14:48] mato :-)
[14:48] pieterh otherwise train sounds excellent
[14:48] pieterh I'
[14:49] pieterh I'll add a travel note for those coming via Charleroi
[14:49] mato d'you have a floor i can sleep on in brussels? or is there a cheap-ish hotel in the vicinity?
[14:49] pieterh mato: yes, you can stay with us
[14:49] pieterh we have space, and beer
[14:49] pieterh you know you're always welcome in my home
[14:49] pieterh if you bring a sleeping bag that'd probably help
[14:50] mato i can do that
[14:50] mato i have one of those miniature summer ones
[14:50] Guthur pieterh: what happens in the case where an author becomes unreadable
[14:50] Guthur unreachable*
[14:50] mato Guthur: someone else takes over
[14:50] pieterh Guthur: you're unable to relicense, period
[14:50] Guthur that's pants
[14:50] mato oh, sorry, different discussion
[14:50] pieterh this is why you need to be pretty sure about this before over-investing
[14:50] Guthur what if he's dead
[14:51] pieterh Guthur: wait 75 years, or just 50 in Europe
[14:51] mato copyright goes to his inheritors
[14:51] pieterh welcome to copyright law
[14:51] mato doesn't it?
[14:51] pieterh it's magical!
[14:51] pieterh mato: yes, but 50+
[14:52] mato pieterh: ah, so you wait 50 years only to be told by surviving family to bugger off anyway :)
[14:52] pieterh Guthur: you're welcome to lobby to change the law but in the meantime your options are (a) accept the license under which you got the code (b) get permission to relicense (c) rewrite, clean-room
[14:52] pieterh mato: after 50 years the copyright on the original work used for the derivation has expired
[14:52] pieterh this is of course a bad joke, in software terms
[14:53] Guthur yep
[14:55] pieterh mato: I've added travel instructions from Charleroi to Brussels on the page
[14:56] mato pieterh: on the charleroi website it mentions some kind of direct shuttle to gare du midi for EUR 22 return
[14:56] pieterh I've added the necessary link
[14:57] pieterh the train is cheaper and more relaxing but takes longer
[14:57] mato oh, I see, the last sentence should be "If you can't get on the shuttle bus" or something
[14:58] mato pieterh: the return flight is now 12:20, much saner
[14:59] mato how does ryanair work? i take it they are anal about baggage or something being a lowcost carrier?
[14:59] pieterh mato: hmm, you're best taking only hand luggage and squeezing your clothes into that
[14:59] pieterh also print your boarding passes out beforehand
[15:00] pieterh and I tend to pay for the extra "early boarding", the best seats are in the very first row
[15:00] pieterh hang luggage has to be within standard dimensions
[15:00] pieterh that is I'm sure noted on the ryanair site
[15:00] pieterh if you can survive the relative brutality of the experience, it is quite pleasant
[15:01] pieterh extremely rapid boarding and deboarding
[15:01] pieterh it's the zeromq of plane flights, in some respects
[15:01] Guthur Ryanair are the carrier i detest the most
[15:01] pieterh hehe
[15:01] mato yes, well, but AUA want EUR 380 for a return flight to brussels, that's just insane :-)
[15:02] mato Ryanair for 88 EUR is cheap
[15:02] Guthur they are the only one that requires a passport of internal flights, let alone the charge for everything setup
[15:02] Guthur of/for
[15:04] jhawk28 I suspect that the pain of blocking would be minimized if there were a timeout version of send
[15:05] pieterh mato: yes, it's easily worth it IMO
[15:05] pieterh Guthur: all airlines ask for ID, no exceptions, to prevent ticket fraud
[15:06] Guthur pieterh: not a passport for travel on an internal flight
[15:06] pieterh UK internal you mean?
[15:06] Guthur yep
[15:06] Guthur Ryanair are the only one
[15:06] pieterh well, if the silly UKians had ID cards, that'd be easier
[15:06] mato pieterh: ok it means i get in monday evening and leave wednesday AM, which is ideal
[15:06] pieterh the point is to avoid ticket fraud
[15:06] Guthur i have plenty
[15:07] pieterh i.e. I buy a bunch of tickets at 1Euro and resell them for anonymous reuse
[15:07] Guthur driving license, and passport
[15:07] pieterh official ID cards, not driving license...
[15:07] guido_g howdy
[15:07] Guthur why would I need another
[15:07] pieterh mato: yup, neat
[15:07] Guthur driving license is quite official
[15:07] Guthur it's got from a government agency
[15:08] pieterh Guthur: you're right, I've no idea why I'm defending ryanair, they deserve your scorn, but however they're 1/4r the cost of alternatives, so... easy choice
[15:08] pieterh guido_g: hi!
[15:08] guido_g hi pieterh!
[15:09] pieterh I see you upgraded jzmq to 3.0, nice
[15:09] pieterh guido_g: sorry, I'm confusing my 'g's
[15:09] guido_g :)
[15:09] Guthur pieterh: it's quite alright, I fell foul of their passport requirement, and so I'm quite bitter
[15:09] Guthur I know...of course I should have read the terms...
[15:10] pieterh Guthur: see, if you lived in a civilized country you'd have an ID card and no stress
[15:10] Guthur pieterh: would I also need a driving license and passport?
[15:10] pieterh nope
[15:10] pieterh not unless you want to drive and cross borders outside schengen
[15:10] Guthur lol
[15:11] pieterh i can travel to/from UK with only my id card, in fact
[15:11] guido_g but these things you'll better do w/ a fake identity anyway ]:->
[15:11] Guthur those options are nice to have
[15:11] pieterh so it's valid across the whole EU afaics
[15:11] Guthur hehe everyone trusts the belgians
[15:11] Guthur never done nothing to no one, hehe
[15:12] pieterh except large parts of Africa, but nevermind...
[15:12] mato bah, they sure do add random bits to the fare
[15:12] mato 6 EUR "administration fee"
[15:12] mato wtf
[15:12] pieterh mato: yeah
[15:12] pieterh it's still so cheap as to be stupid
[15:13] pieterh I was buying tickets to BTS for 40 Euro all in at one point
[15:13] guido_g pieterh tweets in dutch? nice...
[15:13] mato yeah i checked the train, that would have been ~260 EUR return
[15:13] pieterh well, 'tweeten' just sounds weird in Dutch... :)
[15:13] mato so 108 EUR is still silly cheap
[15:14] pieterh plus it's really a nice little flight
[15:14] pieterh both small airports, little hassle
[15:17] Guthur where is BTS
[15:18] mato Bratislava, Slovakia
[15:18] Guthur ah, i honestly would never have got that
[15:19] mato pieterh: um, is there something vaguely like an invoice i can get off ryanair?
[15:19] mato the email receipt does not have e.g. the card details
[15:19] mato oh well
[15:19] pieterh mato: I usually print the email receipt, that's sufficient along with the visa card statement, at least in Belgium
[15:20] pieterh perhaps also a copy of the boarding passes
[15:21] guido_g anything a nerd should see in brussels?
[15:21] pieterh guido_g: how long will you have?
[15:21] guido_g *12:30
[15:21] pieterh hmm... we might organize a meetup on monday evening...
[15:21] mato that's what tuesday evening is for, no?
[15:21] guido_g cool... tweo meetups for the price of one!
[15:21] mato :)
[15:22] guido_g book nonw! limited offer! :)
[15:22] pieterh there's like the 'official meetup' and the 'secret' one
[15:22] guido_g ahhh...
[15:23] pieterh well, if you have time in brussels... there is an original lambic brewery, cantillion
[15:24] pieterh otherwise, just chilling in the old town with a laptop and wifi is fun
[15:24] pieterh actually right where we have the conference
[15:24] pieterh chilling = drinking orval and pretending to work while admiring the view
[16:12] phantomcircuit ok this is just ridiculous
[16:12] phantomcircuit i cant even get php-zmq to build on linux
[16:12] phantomcircuit autotools version mismatch
[16:12] Skaag built fine for me on Ubuntu 10.04
[16:13] Skaag with the standard build-essentials
[16:19] CIA-75 libzmq: 03Martin Sustrik 07master * r5816976 10/ (10 files in 3 dirs):
[16:19] CIA-75 libzmq: Message validity is checked in the runtime
[16:19] CIA-75 libzmq: Signed-off-by: Martin Sustrik <> -
[16:19] CIA-75 libzmq: 03Martin Sustrik 07master * re0246e3 10/ (55 files in 2 dirs): (log message trimmed)
[16:19] CIA-75 libzmq: Message-related functionality factored out into msg_t class.
[16:19] CIA-75 libzmq: This patch addresses serveral issues:
[16:19] CIA-75 libzmq: 1. It gathers message related functionality scattered over whole
[16:19] CIA-75 libzmq: codebase into a single class.
[16:19] CIA-75 libzmq: 2. It makes zmq_msg_t an opaque datatype. Internals of the class
[16:19] CIA-75 libzmq: don't pollute zmq.h header file.
[16:19] mato ok, that fixes that
[16:19] CIA-75 libzmq: 03Martin Sustrik 07master * r10fbc78 10/ src/fq.cpp :
[16:19] CIA-75 libzmq: Assert during SUB socket termination fixed.
[16:19] CIA-75 libzmq: Fair queueing algorithm was checking whether the current pipe
[16:19] CIA-75 libzmq: is not closed in the middle of reading a multipart message.
[16:19] CIA-75 libzmq: However, this is OK when the socket is closing down.
[16:19] CIA-75 libzmq: Signed-off-by: Martin Sustrik <> -
[16:20] jhawk28 mato: can you make it only do a single line per commit?
[16:20] mato checking...
[16:21] mato i've also enabled CIA on the pyzmq repo
[16:21] th pieterh: did you have a chance to look into this mixed multipart messages bug (#199) is there anything i can provide to help fixing this.
[16:23] phantomcircuit Skaag, yeah that's kind of the point actually, the WORKS FOR ME isn't a solution
[16:24] Skaag phantomcircuit: what OS are you compiling on?
[16:24] phantomcircuit gentoo
[16:24] Skaag oh god
[16:24] phantomcircuit yeah terrible
[16:24] Skaag your tools are probably too new and too bleeding edge
[16:24] phantomcircuit except everything else builds fine
[16:24] Skaag can you pastebin the error you get?
[16:25] phantomcircuit sure
[16:25] Skaag I don't hate gentoo, don't get me wrong
[16:25] Skaag please.
[16:25] Skaag I actually like it.
[16:25] phantomcircuit
[16:25] Skaag However, if I had a dime for every time a gentoo guy complained about something not compiling right... I'd definitely be able to buy a very fancy meal for everybody in this channel.
[16:26] Skaag I think it just stems from the fact very few developers test their build process on the latest gentoo.
[16:26] Skaag looking at your paste. hold on
[16:26] phantomcircuit it's definitely due to libtools version being too new
[16:26] pouete Hi all, a little question about the C bindig of ømq :) : why cant we reuse a msg that as already been initialized ?
[16:27] CIA-75 pyzmq: 03MinRK 07master * r65812f5 10/ README.rst : readme updates re: building pyzmq -
[16:27] CIA-75 pyzmq: 03MinRK 07master * r80e975d 10/ : (log message trimmed)
[16:27] CIA-75 pyzmq: add -lpthread on freebsd
[16:27] CIA-75 pyzmq: closes gh-101
[16:27] CIA-75 pyzmq: 03MinRK 07master * r02cc78d 10/ (zmq/tests/ zmq/tests/ cleanup a few more test sockets. -
[16:27] Skaag phantomcircuit: you tried to install the older libtools and get it to be used instead?
[16:27] Skaag or how about indeed recreating aclocal?
[16:27] CIA-75 libzmq: 03Martin Sustrik 07master * r5816976 10/ (10 files in 3 dirs):
[16:27] CIA-75 libzmq: Message validity is checked in the runtime
[16:27] CIA-75 libzmq: Signed-off-by: Martin Sustrik <> -
[16:27] CIA-75 libzmq: 03Martin Sustrik 07master * re0246e3 10/ (55 files in 2 dirs): (log message trimmed)
[16:27] CIA-75 libzmq: Message-related functionality factored out into msg_t class.
[16:27] CIA-75 libzmq: This patch addresses serveral issues:
[16:27] CIA-75 libzmq: 03Martin Sustrik 07master * r10fbc78 10/ src/fq.cpp : (log message trimmed)
[16:27] CIA-75 libzmq: Assert during SUB socket termination fixed.
[16:27] CIA-75 libzmq: Fair queueing algorithm was checking whether the current pipe
[16:28] Skaag wow lots of activity ;) go martin
[16:28] mato no, that's just me testing
[16:28] mato sorry for the noise
[16:28] Skaag oh ok
[16:28] Skaag I don't mind it
[16:28] mato jhawk28: header + first line of log message?
[16:28] CIA-75 libzmq: 03Martin Sustrik 07master * r5816976 10/ (10 files in 3 dirs): Message validity is checked in the runtime ...
[16:28] CIA-75 libzmq: 03Martin Sustrik 07master * re0246e3 10/ (55 files in 2 dirs): Message-related functionality factored out into msg_t class. ...
[16:29] CIA-75 libzmq: 03Martin Sustrik 07master * r10fbc78 10/ src/fq.cpp : Assert during SUB socket termination fixed. ...
[16:29] jhawk28 pouete: I think its because zmq used zero copy
[16:29] phantomcircuit Skaag, yeah let me see though
[16:29] mato jhawk28: ok, done, seems to work
[16:29] CIA-75 jzmq: 03Gonzalo Diethelm 07master * r8755367 10/ src/Socket.cpp : Use send/sendmsg and recv/recvmsg depending on 0MQ's version. -
[16:30] phantomcircuit Skaag, rebuilding aclocal.m4 didn't help
[16:30] jhawk28 mato: less noisy, better
[16:30] mato it's watching libzmq, rbzmq, pyzmq, jzmq, clrzmq
[16:30] Skaag phantomcircuit: I guess you need to install the older tools as well. They shouldn't interfere with the new ones, don't worry.
[16:30] mato anyone else wants some repos watched, shout
[16:31] phantomcircuit Skaag, yeah i guess
[16:31] phantomcircuit i hate autotools :|
[16:31] Skaag wish I could be more helpful...
[16:31] phantomcircuit Skaag, which version do you have
[16:31] phantomcircuit might save me some time...
[16:32] mato "only" 237 messages unread in zeromq-dev :-(
[16:32] mato time to go, see you all later
[16:32] jhawk28 pouete: there was some talk on the mailing list about it
[16:33] pouete jhawk28: kay, do you remember the approximate date ?
[16:33] pouete i mean month ?
[16:33] Skaag phantomcircuit: will tell you, sec.
[16:33] jhawk28 was just this month
[16:34] Skaag (GNU libtool) 2.2.6b
[16:34] Skaag automake (GNU automake) 1.11.1
[16:34] Skaag autoconf (GNU Autoconf) 2.65
[16:34] phantomcircuit weird i've got 2.2.10
[16:34] Skaag yah, too new I guess
[16:34] phantomcircuit you'd think that minor minor versions would be compatible :|
[16:34] phantomcircuit yay gnu :/
[16:34] Skaag Agree with you 100% - definitely not proper!
[16:35] Skaag you'd think any 2.2.x would play nice
[16:35] pouete jhawk28: cheers :) anyway, is this paper still right ?
[16:36] jhawk28 pouete: not sure, I haven't had much experience with C
[16:36] phantomcircuit lol sadly this box is hilariously slow
[16:36] phantomcircuit gogo pentium 4
[16:37] phantomcircuit compile like the wind!
[16:37] Skaag lol! pentium 4!!!
[16:37] phantomcircuit sigh
[16:37] Skaag :)
[16:37] Skaag I'm all for recycling and stuff, but this is actually worse for the environment...
[16:37] Skaag ;-)
[16:39] phantomcircuit sigh
[16:39] phantomcircuit there isn't another libtools package
[16:39] phantomcircuit there are higher versions but not lower
[16:39] phantomcircuit facepalm
[16:42] pouete jhawk28: who could give me the info ?
[16:42] jhawk28 pieterh or sustrik at least
[16:42] pouete thank you very much :)
[16:43] jhawk28 pouete: I think you need to clone the data before sending it
[16:44] jhawk28 in order to send it multiple times
[16:48] phantomcircuit this is like a bad joke or something
[16:49] pouete jhawk28: what i want to do is to loadbalance one pusher that receive data on different pullers that will process the data.
[16:51] jhawk28 pouete: so, kind of the worker pattern?
[16:51] pouete worker pattern ?
[16:52] jhawk28
[16:52] jhawk28 see the picture
[16:52] pouete yes, its this :)
[16:52] jhawk28
[16:53] pouete yep thats it
[16:53] jhawk28 thats the ZMQ way
[16:54] pouete okay, i take a look at that , thanks
[17:00] pieterh pouete: that zerocopy paper is accurate for libzmq/2.x
[17:35] Seta00 !setinfo <jhawk28> thats the ZMQ way™
[17:35] Seta00 ;)
[17:48] pieterh jhawk28: pouete: actually is closer to the Rabbit example
[17:48] pieterh but if it's just load balancing to workers, one PUSH and multiple PULL sockets is enough
[17:48] jhawk28 he wasnt wanting reliable
[17:49] pieterh jhawk28: yeah, wasn't sure
[17:49] phantomcircuit hmm
[17:49] jhawk28 I just remembered the rabbit picture
[18:00] pieterh guido_g: you around?
[18:08] mikko pieterh: renamed
[18:37] guido_g pieterh: yes
[18:37] guido_g what's up?
[18:50] pieterh mikko: thanks!
[18:50] pieterh guido_g: sorry, and excuse this ignorance, but I was wondering what parts of 0MQ you contributed to
[18:56] guido_g the first tests
[18:57] guido_g and a patch
[18:57] guido_g afair
[18:57] guido_g is this of any importance?
[18:58] guido_g do i get a better chair at the conf?
[19:20] pieterh guido_g: (/me is distracted), do you consider yourself a core developer?
[19:20] pieterh it's to add to the conference page, to indicate to visitors more or less who's who
[19:21] pieterh you get the best chair even without asking
[19:24] guido_g pieterh: no, i'm not a core dev
[19:25] pieterh guido_g: ok
[21:32] mikko pieterh: are people arriving on 9th ?
[21:32] mikko MAy
[22:29] iAmTheDave wait, there's a conference?
[22:44] th oh there was a patch for message atomicity for pub/sub? is that related to #199 which is about atomocity of xreq/xrep?
[23:05] lt_schmidt_jr Where is the May 9 conference going to be held?
[23:57] zedas ok i got an odd question: if I have an XREP socket, and I've got the idents from them, and those idents are the autogenerated ones, AND i store them to disk: can i just read them back and then send replies after a crash?