Mochabot log - CommonJS IRC channel: #commonjs on irc.freenode.net

2009-12-04:

[0:31] <Dantman> Wes-_, ;) for the record on the new server root is going to be the ONLY user with a password. Root has no authorized ssh keys. And ssh rejects root login and password login. So the ONLY ways to get into the terminal is with ssh and keys, or using the web based administrative console in the rackspacecloud account that controls all the servers (That's the only place the root password has any use)
[1:03] <Wes-_> there are always ways around keys
[1:03] <Wes-_> crowbars work well
[1:22] <MisterN> also, only criminals lock their computer.
[1:54] <JohnnyL> only siths believe in absolutes.
[1:55] <MisterN> JohnnyL: but never say never.
[2:04] <JohnnyL> I love this town!!!
[2:05] <keeto> ondras: turns out my current build of v8cgi works with SL too.
[6:44] <ondras> keeto: SL = ?
[6:45] <keeto> ondras: snow leopard.
[6:46] <keeto> btw, since you're here, could you help me with one issue?
[6:46] <ondras> yes
[6:46] <keeto> I'm trying to build a 64bit version.
[6:46] <ondras> hmm
[6:46] <ondras> so you have 64bit v8 already?
[6:46] <keeto> yeah.
[6:47] <keeto> gimme a second to get the output.
[6:47] <ondras> np
[6:48] <ondras> but bear in mind that I cannot get a 32bit version on mac to compile
[6:48] <ondras> so I am probably not the best person to aid with 64b :)
[6:48] <keeto> well, the main error is "ld warning: in /usr/local/lib/libv8.dylib, file is not of required architecture"
[6:49] <keeto> that's with a libv8 built with arch=x64
[6:49] <ondras> "file /usr/local/lib/libv8.dylib" ?
[6:49] <keeto> Mach-O 64-bit dynamically linked shared library x86_64
[6:50] <ondras> oukey doukey
[6:50] <ondras> so
[6:51] <ondras> try passing -m64 to linker
[6:51] <keeto> you'll have to tell me how to do that. x)
[6:53] <ondras> # default values
[6:54] <ondras> CCFLAGS = ["-Wall", "-O3"],
[6:54] <ondras> LINKFLAGS = []
[6:54] <ondras> try adding "-m64" to both these
[6:56] <keeto> nice, one step forward.
[7:00] <keeto> ondras: now I'm getting a bunch of undefined symbols.
[7:00] <ondras> show them :)
[7:01] <keeto> http://pastie.textmate.org/727031
[7:02] <ondras> this is basically what I get on 32bit version :)
[7:02] <keeto> haha.
[7:03] <keeto> I think the reason I'm still able to build is because of my old libv8.
[7:04] <ondras> hm
[7:04] * ondras is going to rebuild v8
[7:20] <keeto> ondras: so close! x)
[7:21] <keeto> I'm regressing through older v8 builds in my hd.
[7:23] <ondras> hehe
[7:24] <keeto> v8::Debug::EnableAgent(char const*, int)
[7:24] <keeto> that's the only error from one old v8 build.
[7:26] <keeto> baaah.
[7:26] <ondras> wow!
[7:26] <ondras> actually
[7:27] <ondras> I maybe know what this is
[7:27] <ondras> this is not in v8.h, but in v8-debug.h
[7:27] <ondras> but you get the error at link phase, not at compilation, hmmh :/
[7:28] <keeto> oh?
[7:30] <ondras> no, nothing
[7:30] <ondras> I just thought you might be not including the v8-debug.h
[7:30] <ondras> but that is nonsense
[8:01] <keeto> ondras: I'm giving up for the moment. x)
[8:01] <keeto> it'll be nice to get 64bit support, but I'll settle for what compiles now..
[8:05] <ondras> actually, I would be happy with 32bit support on mac :/
[8:11] <keeto> want the files? :)
[8:12] <keeto> http://keetology.com/_files/v8cgi.zip
[8:12] <keeto> there's also a compatible dylib on the server.
[8:13] <keeto> wget http://keetology.com/_files/libv8.dylib
[8:13] <keeto> I was thinking of turning it into a pkg.
[8:13] <keeto> for easy installation.
[8:14] <keeto> that's actually the reason why I'm trying to compile it for 64bit--so that the built-in apache would be compatible.
[8:24] <Dantman> Tch, damn, nameserver hostnames can't be CNAMES to A records containing the nameserver's IP
[8:25] <ondras> keeto: I want to build it with latest v8 :)
[8:26] <keeto> don't we all? x)
[8:26] <ondras> hehe
[10:43] <olivvv> join node.js
[12:04] <Achilles> Hi , I 'm newbie about commonjs.
[12:06] <evilstreak> hi
[12:08] <Achilles> Hi evilstreak. I frm Thailand.
[12:09] <evilstreak> good to know
[12:11] <Achilles> Today I installed node.js on my server. but I 'dont understand commonjs concept, How is the best way ?
[12:11] <evilstreak> best to ask about problems with node.js in the node channel
[12:12] <Achilles> Thank you evilstreak.
[15:37] <ashb> irakli isn't on irc is he?
[15:42] <MisterN> ashb: we should counter every "join node.js" with a "join flusspferd, actually" :D
[15:43] <ashb> MisterN: create an event loop and then we can
[15:43] <MisterN> ashb: i thought you wanted to create the event loop :P
[15:43] <MisterN> i'm busy creating pointless c++ apis :P
[15:43] <ashb> :D
[15:43] <ashb> yeah. creating an event loop for us isn't so trivial tho
[15:43] <ashb> since we need to support pluggable embedding too
[15:44] <ashb> which might well have its own loop already
[15:44] <MisterN> event loops in general are hard.
[15:44] <MisterN> especially if it should work on windows
[15:44] <ashb> yeah
[15:44] <MisterN> the asio stuff does work on windows, so maybe we could use it
[15:45] <MisterN> i hate thinking about windows
[15:45] <ashb> yeah i know
[15:45] <MisterN> to paraphrase Heine: "Denk ich an Windows in der Nacht, dann bin ich um den Schlaf gebracht."
[15:45] <MisterN> :D
[15:46] <ashb> heh
[15:46] <ashb> only had to loop up the last two words there :D
[15:47] <ashb> & got to prepare for presentation tomorrow
[15:47] <evilstreak> presentation on?
[15:47] <evilstreak> oh, LPW
[15:47] <ashb> carer development bullshit
[15:48] <ashb> and i need to got cut my beard into half-beard-of-awesome
[15:49] <evilstreak> left or right half?
[15:50] <MisterN> hmm will juice implement commonjs browser-side?
[15:50] <MisterN> that would be cool
[15:51] <evilstreak> fs-base browser-side? seems unlikely :P
[15:51] <MisterN> i mean mainly SecurableModules
[15:52] <MisterN> and then maybe you should do the same crazy static analysis shit that the... i forgot who was doing it, but somebody was
[15:56] <ashb> krisk
[15:56] <Wes--> you could do fs-base browser-side, sort of
[15:56] <Wes--> use the user local data area to store "files"
[15:58] <Wes--> BTW, is there any reason event loop cannot be implemented in pure JS? The flusspferd could just load up an extra module to bootstrap that type of thing.
[15:59] <MisterN> isn't there some local user storage in HTML5?
[15:59] <MisterN> that could be used to simulate a local fs :D
[15:59] <deanlandolt> MisterN: you mean this? http://dev.w3.org/html5/webstorage/
[15:59] <Wes--> I don't know about HTML5, but firefox and ie have user local data. Probably the other browsers too
[16:00] <Wes--> It's been possible to allocate arbitrary blocks of up to 64K from JScript since IE4
[16:00] <MisterN> deanlandolt: yeah i think this is it
[16:00] <evilstreak> the webkit web inspector has a lovely view to inspect all the local storage too
[19:11] <inimino> man, when did the mailing list get so high-volume?
[19:11] <ondras> after the ars article I believe
[19:14] <MisterN> ondras: you really should ban this shavon shomo guy from your list
[19:16] <ondras> I think I already did that
[19:16] <MisterN> ondras: dunno i just checked and had two v8cgi marked spam mails from him in my spam folder
[19:17] <ondras> yeah
[19:18] * inimino skips the entire templating thread
[20:41] <deanlandolt> Wes--: nice arguments for package-system-as-module
[20:41] <Wes--> deanlandolt: Thanks. This has been *really* bothering me.
[20:42] <deanlandolt> my only fear w/ that is that when (inevitably, at least at first) everyone creates their own package manager there'll be a million different require calls sprinkled throughout the internet that look nothing alike
[20:42] <deanlandolt> i guess that's not a big deal though...let the market work
[20:44] <deanlandolt> but it kinda /is/ a bit of a big deal in browser land when trying to determine dependencies
[20:44] <Wes--> well, yes and no
[20:44] <Wes--> presumably the web master will choose one package system
[20:45] <Wes--> and even if he doesn't, properly designed, they would work together anyhow
[20:45] <deanlandolt> well, what if you want to use a module packaged up differently (has different require semantics)? now you have two package systems
[20:46] <Wes--> deanlandolt: right, imagine how both of those package systems works
[20:46] <Wes--> one works, say, but searching for require("cjspan").require(XYZ)
[20:47] <Wes--> the other works, say, by searching for require("superDuperModules").require(ABC)
[20:47] <deanlandolt> good point
[20:49] <deanlandolt> it'd put a crimp in kriskowal's notions of a module transport format for sure, but perhaps if we standardized the package metadata that could still be worked out
[20:50] <Wes--> yes
[20:50] <Wes--> again, we are conflating implementation easy with wisdom of design :)
[20:50] <deanlandolt> :D
[20:51] <MisterN> hey we could abuse $ to host a require-autoloader *more bad ideas*
[20:51] <MisterN> so you could do $.system instead of require('system')
[20:51] <Wes--> I was thinking we could use three dollar signs
[20:52] <MisterN> using one dollar sign is standard
[20:52] <deanlandolt> we could release /another/ library that uses _ for that
[20:52] <MisterN> we should not deviate from the standard
[20:52] <Wes--> Yes, but three is better than one!
[20:53] <MisterN> i think three dollar signs would be UNPROFESSIONAL
[20:53] <MisterN> we should use no more than one
[20:53] <Wes--> deanlandolt: something else to consider, having packages outside of require like that might make it easier to use CDN for packages and still have local modules
[20:54] <Wes--> Actually, maybe we should use the euro symbol
[20:54] <MisterN> heh.
[20:54] <MisterN> unfortunately not a valid js identifier :(
[20:55] <Wes--> even when C strings are UTF-8?
[20:55] <deanlandolt> Wes--: oooh...i like...there could even be a package manager module that does everything via cdn
[20:56] <deanlandolt> yeah...i tried pasting ? into the console to no avail :-/
[20:56] <Wes--> deanlandolt: precisely!
[20:56] <deanlandolt> Wes--: that's pretty spectacular...and frankly, these package manager modules could be written /now/ (as you pointed out)
[20:58] <Wes--> right. And I think the secure sandbox argument is a red herring, but I have a tough time with all that academic-speak. ;)
[20:59] <deanlandolt> Wes--: heh...yeah, same here...don't tell anyone but i don't even have a computer science degree ;)
[21:00] <Wes--> I left a half-credit short myself. The work was just so much more interesting.
[21:02] <Wes--> I wouldn't mind going to back to school though to audit some grad-level language design courses (my alma mater has a language lab and some good profs in the area)... but I can never seem to find time, *sigh*
[21:03] <deanlandolt> same here...i actually work on campus of my alma mater and still haven't bothered taking any classes
[21:04] <Wes--> Well, I'm only a mile away, so not much better!
[21:15] <deanlandolt> Wes--: sounds like ihab's coming around..."I'm not [yet...] arguing that this is the correct approach"...
[21:21] <Wes--> It does... I'm going to have to spend some time parsing his message and arguments, though. :)

 

 

Logs by date :