00:00:10 oh, that's what I do in rdfapi 00:00:11 PERSON and CONTACT are locally scoped in Python 00:00:19 the reason he couldn't do that is because of: 00:00:20 -- 00:00:27 bob.contact_address = Thing(None, 00:00:28 contact_street = "314159 There Street", 00:00:28 contact_city = "EveryVille", 00:00:28 contact_state = "NA", 00:00:29 contact_zip = "12345" 00:00:29 ) 00:00:30 -- 00:00:41 the problem is that those are evaluated at runtime as attribute lookups, so they would be slow 00:01:43 I guess the thingFactory could take namespace bindings. 00:02:52 bob[CONTACT.address] = Thing(None, { 00:02:53 CONTACT.street : "314159 There Street", 00:02:53 CONTACT.city : "EveryVille", 00:02:53 CONTACT.state : "NA", 00:02:53 CONTACT.zip = "12345" 00:02:53 } ) 00:03:12 er, that last '=' should be a ':' 00:03:29 Ooh, I hadn't thought of that! 00:03:56 but, that's where the RFC would come in, that all should be: 00:03:57 bob.CONTACT:address = Thing(None, { 00:03:58 CONTACT:street : "314159 There Street", 00:03:58 CONTACT:city : "EveryVille", 00:03:58 CONTACT:state : "NA", 00:03:58 CONTACT:zip : "12345" 00:03:59 } ) 00:04:47 and, you could even go back to: 00:04:47 bob.contact:address = Thing(None, 00:04:48 contact:street = "314159 There Street", 00:04:48 contact:city = "EveryVille", 00:04:48 contact:state = "NA", 00:04:48 contact:zip = "12345" 00:04:49 ) 00:05:11 because named parameters would "just work" 00:05:16 what would contact:city evaluate to? 00:05:32 i suggested to TimBL that he propose a URI type in Python, written 00:05:44 since he's giving a talk on webizing python 00:05:48 the tuple ["http://www.example.com/contact#", 'city'] 00:05:57 contact:state could evaluate to the appropriate URI. 00:06:18 there's also that interpretaion 00:06:43 and that's a good "literal" for URIs too: 00:06:58 bob. = ... 00:07:47 unfortunately, the RDF crowd need to figure out whether URI <-> two-part names is bidirection or unidirectional 00:07:59 We got the TAG to take that one. :) 00:08:26 hehe, schlep it off on someone else! 00:08:51 http://www.w3.org/2001/tag/ilist#rdfmsQnameUriMapping-6 00:08:54 besides RDF, is there *any* group that sees {NS}name as ? 00:09:14 not to my knowledge. 00:09:26 it's pretty funny because apparently they rushed XML namespaces out so that RDF could use them. 00:09:31 tansaku (~sam@n145-146.tokyu-net.catv.ne.jp) has joined #swhack 00:09:34 I thought so! yup, schlupping it off to some other group! 00:09:43 RDF does not see {NS}name as 00:09:54 it doesn't? 00:10:04 nope. It does not add a hash 00:10:12 oh, right. 00:10:24 understood, I was abbreviating 00:13:14 Hmm, this is really interesting: http://www.ietf.org/internet-drafts/draft-stlaurent-feature-xmlns-01.txt 00:13:50 now, given the proper interface (which can be approximated in current versions of Py and Perl), one should be able to have a range of interoperable RDF|node-property data stores that are very nicely bound to the languages 00:13:59 yeah, in that it dredges up all of the old "what is a namespace?" discussions 00:14:08 heh 00:14:38 i like nice bindings 00:15:07 gotta run: dinner 00:15:13 me too, ttyl 00:15:25 you too Sean! 00:15:30 c'ya! 00:15:35 kmacleod has quit ("Leaving") 00:51:15 BenSw has quit (Read error: 110 (Connection timed out)) 01:01:26 tav (tav@host217-34-70-43.in-addr.btopenworld.com) has joined #swhack 01:02:40 yes 01:03:16 [more off-log mumblings] 01:09:27 tav has quit (devlin.openprojects.net irc.openprojects.net) 01:09:27 chumpster has quit (devlin.openprojects.net irc.openprojects.net) 01:30:18 what did people do on the Web before Google, Alexa, and AudioGalaxy/Napster? 01:30:48 Who knows? 01:32:06 gotta run 01:32:19 c'ya 01:43:17 Hmm... on Dec 3rd. 2001 #, the top search result on Google for sha512 was http://lists.cryptix.org/pipermail/cryptix-team/2000-October/000295.html 01:46:20 * jeremiah is away: sleeping 01:55:54 tav (tav@host217-34-70-43.in-addr.btopenworld.com) has joined #swhack 01:57:09 kmacleod (~ken@kmacleod.static.iaxs.net) has joined #swhack 01:57:46 wb 01:58:15 whatever happened to the project to let me paste a URL as a login to sites, such that it picked up my info from the URL (at least as much as authorized) 01:59:26 erm... you mean a script to get the authority info from a URI, or something more fancy? 01:59:41 an identity server 02:00:10 authentication (or merely public info), not authorization 02:00:49 context: the reason I commented on Julian's article here is because of incremental pain of registering at yet another site 02:01:04 er, yep, sorry, wrong word 02:01:09 rillian (~giles@mist.thaumas.net) has joined #swhack 02:01:16 so naturally the idea resurfaced of an identity server 02:01:46 a little blob of RDF at the end of a URL, of which could nicely be a URI for referring to said person, too 02:01:49 metacookie or something 02:02:48 my grammar parser just barfed nuts and bolts on your sentence :-) 02:03:55 a common URI for "me" is mailto:ken@bitsko.slc.ut.us, but that's a lousy URL for getting info about me. something like http://identity.org/ken@bitsko.slc.ut.us would be much better 02:04:21 that first URI is the URI for your mailbox, not you 02:04:22 not only can that be a URI for "me", but also a resource containing info about me 02:04:47 understood 02:06:17 A URI can only denote one thing 02:06:19 the relationship between you and that mailbox is [ foaf:mbox ] . where [] is a bNode 02:06:24 [an existentially quantified variable] 02:06:30 but yeah, it's a good idea 02:07:00 yes. lack of preciseness on my part 02:07:14 however, you don't need to beg to some centralized server for identity 02:07:25 I mean, you have a server, I presume? Or you can use www-archive 02:07:32 so just put the RDF online 02:08:05 the missing piece is that anyone with the URL can paste the URL. needs "something you know" to do the final authentication 02:08:56 can you give an example? 02:09:12 that is to say, there needs to be a third-party authentication 02:09:19 yes 02:10:09 base scenario one: I go to VoidStar, enter my URL as my login, and enter comments 02:10:51 because I've not authenticated in any way, just pasted a URL, there's nothing preventing anyone else from pasting the same URL and forging comments 02:11:20 ah, right 02:11:20 makes sense 02:12:02 so really, the identity "URL" would be of more use when signing up for something 02:12:04 so, some kind of "token" needs to be added to the URL to positively verify it's me with the server that maintains the URL 02:12:32 I don't think you'd add it to the "URL" 02:12:40 you'd put it in as a separate form entry 02:12:53 and I don't think you'd use the URL to login 02:12:57 that's one way (the single paste method), otherwise you can type/paste the token in the password field 02:13:10 you'd just use a normal login name. But you could specify a list of preferred login names in your personal profile 02:13:21 so that when you sign up, it chooses one for you 02:13:52 the bigger picture is that I don't want to "sign up" and maintain a destinatiuon-local profile ;) 02:15:15 a destination-local profile? 02:15:16 again with VoidStar, here's my first time in, I find something interesting I want to comment on, I don't want to go through a registration, I just want to "log in" and comment 02:15:30 a profile maintained with the site I'm visiting 02:15:40 local to the site 02:16:10 Uh huh 02:16:34 yep, it sounds like a good idea - put you'd have to promote it 02:18:34 it's got problems of an RDF degree, that is: how can one control the amount of information given out 02:19:12 but the protocol should be really simple 02:19:53 GET http://identity.org/ken@bitsko.slc.ut.us HTTP/1.1 02:20:10 Authentication: Token=foo 02:21:12 P3P could probably be applied here 02:23:13 Would this be a bit like a simpler www.trusttoolbar.com ? 02:23:29 Which is for brands, companies. 02:27:03 heh, that sounds like spyware! 02:27:35 Yes, ugly, but the same idea, no? 02:28:32 Actually horrible! 02:28:52 in a brief review, I'm not seeing the comparison 02:29:21 it does seem to verify that the owner of the pages you're on is the same as the brand being advertized 02:29:41 Ah, yes. Correct. 02:29:41 or rather, the opposite, if you see a brand but no icon, then it's not authorized 02:29:43 the identitity URL was my idea... it didn't really go anywhere 02:30:21 Why ever not, I wonder - great idea! 02:30:33 yes, it's an old idea, it just hit me again is all, and made me think of the techniques of why it didn't work last go-round 02:30:39 and i had a way to do auth with it 02:31:37 it didn't work because no one implemented, i think 02:31:51 I don't think auth is very complicated, it should be a simple double-blind 02:32:25 and a simple responder, without reverse-authentication and filtering, should be a less than 50-line script 02:33:08 two important bits: simple protocol and trivial reference server code 02:33:32 trivial client code, for webmasters to incorporate would be nice too 02:33:39 i'm not sure what you mean by double-blind reverse responder 02:33:50 Morbus (~Morbus@s109.terminal3.totalnetnh.net) has joined #swhack 02:33:56 of course now-a-days we'll just get folks to use the plex 02:33:59 hey Morbus 02:34:06 can you chump http://www.oreillynet.com/pub/a/mac/2002/01/29/apache_macosx_four.html for us? 02:34:27 "double blind" is the wrong term for the type of authentication where there's no clear text password ever used 02:34:55 ah 02:34:57 i forgot how to chump here. what's the prefix? 02:35:01 it's @ 02:35:05 @http://www.oreillynet.com/pub/a/mac/2002/01/29/apache_macosx_four.html 02:35:05 it's very similar to challenge-response, except no challenge is given 02:35:10 err "@ " 02:35:14 @ http://www.oreillynet.com/pub/a/mac/2002/01/29/apache_macosx_four.html 02:35:29 I guess "one-time key" is the best term 02:35:39 i was thiinking that the server could redirect to your URL plus a challenge 02:35:47 Morbus! 02:35:49 WFM 02:35:54 and then you'd auth to the URL, and it'd respond with the answer to the challenge... 02:35:57 * Morbus hides. 02:36:04 to title: X:|blargh 02:36:14 Hmm... where'd chumpster go? 02:36:15 oh. chumpster appears to be missing. 02:36:33 just a se 02:36:57 I was thinking that the user would have a local connection to the auth server, such that they could just copy a unique pw at any time and paste it in a login field 02:37:29 chumpster (~chumpster@xcdfddb76.ip.ggn.net) has joined #swhack 02:37:36 @ http://www.oreillynet.com/pub/a/mac/2002/01/29/apache_macosx_four.html 02:37:38 A: http://www.oreillynet.com/pub/a/mac/2002/01/29/apache_macosx_four.html from Morbus 02:37:53 that'd be interesting 02:37:59 A:|Apache Web-Serving with Mac OS X, Part 4 02:37:59 titled item A 02:38:10 * kmacleod figures that if chump verbosity is ever gonna get fixed, he's gonna have to do it himself 02:38:12 you could do it with public keys... like please sign this message 02:38:29 heh, confirmation is good tho. 02:38:39 there are much bigger chump bugs, however. 02:38:41 depends on who wants the authentication 02:38:51 i thin ki've fixed many, with this syntax, tho 02:39:00 it's not so much the confirmation but how much is given 02:39:23 ah. 02:39:49 Bijan's original was really best so far in that regard 02:40:02 His new one is a verbose monster, I think. 02:40:11 it's based on chump ;) 02:40:22 true... 02:40:28 it gives each comment a URI :) 02:40:29 with RDF thrown in to make it worse ;) 02:40:36 yeah, exactly 02:40:45 we should get it set up here. ;-) 02:42:16 so, Morbus, what's the next installment going to be about? 02:42:26 mysql/php/postgresql 02:42:33 ah, maybe public-key is all you need in the generated URL. it's the destination-site that wants to verify that the paster-of-the-URL is really the-owner-of-the-URL, so http://identity.org/ken@bitsko.slc.ut.us?auth=0943093242309834098234023948 is just fine as long as that encrypted 02:42:37 i spent all this morning just researching the different distro's available 02:42:41 auth verifies against the public key from that url 02:42:56 wow... there are quite a few 02:43:07 well, specificlaly built for os x, i mean. 02:43:10 Hmm... database backed websites: a little RDF connection looming? 02:43:28 i know very little about rdf. i wouldn't do it justice 02:43:38 hmm, need a one-time thing worked into that somehow... 02:43:48 It's just a datbase format... but you could start using data from other people's sites 02:43:57 to prevent replay attacks (forged logins) 02:44:03 so, for example, news aggregation. You know about that, don't you? :-) 02:44:23 nah, i forgot all my news aggro knowledge to fit this damn db crap in there 02:44:33 heh, heh 02:44:52 I guess that an Amphy update has gone out the window, then :-) 02:45:08 heh, oh, they'll be one. 02:45:15 i have tons of ways forward, just no time to do them in 02:47:07 you said you were going to add threading last time we talked about it... it would be useful: the interface is a bit jerky 02:47:32 yeah, but the first thing i want to add is autocodeupdating, like RU has. 02:47:54 and as part of that, there'd be a wrapper binary around the plain text code, as opposed to the large binary now. 02:48:16 so, mac/win peopel could see the perl code, and mod as they want, they'd just have a single binary that would read it in. 02:48:40 read it in? 02:48:53 i was actually thinking of adding a "no gui" flag for mac/win, so that there wouldn't be a need for it. 02:49:03 the gui really presents nothing exciting to people after they've seen it a few times. 02:49:14 read it in > interpret at runtime. 02:49:48 well, it provides the status, when getting the new feeds... but yeah, I agree 02:50:06 well, it could be replaced with a dos window that disappears after finishing. 02:51:17 good idea 02:52:00 well, WinDump blows 02:53:08 sbp, you should get back into laerning perl so you can help me. 02:53:16 the other guy i had helping me disappeared. 02:53:24 i still talk to him frequently, but he has no time to code. 02:53:28 who was that? 02:53:32 he hasn't seen the code base since v0.24, i think 02:53:38 my good friend bill from monkeymind.net 02:54:16 cool. Well, I don't know how much help I'd be, especially at the rate I'm looking at Perl docs. 02:54:38 argh, I can't get a dump at all from WinDump! 02:54:39 ken: do you know if any work has been done on an xml::sax rss parser? 02:55:09 someone should come up with a Cygwin-lite tool, so that you can bundle, say, cygwin.dll, Perl, and your modules, and any other cygwin-based tools would incorporate any already-installed bits 02:56:00 kmacleod: my plan is similar. i'm using perl2exe to build ampheta now, so the desire was to write a simple wrapper that would really just set up /lib/ as an additon to @INC. 02:56:23 from there, all modules, cpan or otherwise, would get stuck in there, and the wrapper would just be the interpreter, with a hook to launch the main program. 02:56:31 no, I don't know if anyone's touched XML::RSS in a long while. Orchard's RSS module is based on the pre-{NS}NAME SAX2 02:56:39 so it'd be very easy to bring into conformance 02:56:49 last i knew, xml:rss wasn't being touched by rael. 02:57:11 the sad thing is i know nothing about sax. gimme an example, and I can go from there, but actually starting one is difficult for me. 02:57:25 plus, sax stuff isn't working perfectly under the 5.004 build of macperl. 02:57:29 SAX-in-Python really drives me nuts 02:57:32 so i'm kinda waiting for pudge to release a final. 02:57:34 is there a reason you need a sax-based one then? 02:57:42 that damn thing just refuses to work on so many levels... 02:58:02 kmacleod: not really, no. right now, i've got a bastard one built around expat, ::parser, and ::simple. but i'd like to reduce the need for expat. 02:58:13 sbp: really? I thought Py SAX was rock solid a long time ago 02:58:21 if i can ::sax and ::pureperl by default, and then use expat or another if it's there, then i can make installation a lot easier for a lot of people. 02:58:26 sbp just doesn't like event-based programming, i think 02:59:12 Morbus: ah, gotcha. yes, ::pureperl is the way to go for that, because everything you just mentioned is automatic 02:59:35 it automatically uses XML::Parser or SAX::Expat if they are installed 02:59:59 right. i would love that functionality. would make linux installs super easy, as well as making binary distributions smaller. 03:00:14 no, no, event-based programming is fine. The SAX methodology is great, but it just doesn't work 03:00:32 plus, i just want to use an event based sometime. at this point, ::simple is a memory hog on some of the large files I use. i didn't plan it out all too well. 03:01:36 sbp, what doesn't work? the cygwin distro? 03:01:38 looking at the ampheta code just makes me cringe now. 03:01:56 sigh. so much i want to redo. add a plugin api, modularize evertyhing like i shoulda, etc. 03:07:05 sbp has quit (Killed (NickServ (Ghost: SeanP!~sean@m125-mp1-cvx3b.pop.ntl.com))) 03:07:13 hm? 03:07:45 sbp (~sean@63.149.73.20) has joined #swhack 03:08:03 what happened there? 03:08:21 * kmacleod virtually whacks anyone who chmod +x/755 files that are never intended to be executable 03:08:33 Grpgmrpgmrpgprmgprmgprmgprmgpmph 03:08:34 dunno 03:08:48 Heh. 03:08:56 tav loves doing that. 03:09:13 * kmacleod then virtually whacks not the creators of color-ls, but those who made color the default 03:09:27 heh, heh. 03:09:29 you can unalias it 03:09:37 not a fan of bright green? 03:09:48 or red for gzip.s 03:09:54 nonetheless 03:11:21 anybody ever create multiple Apache server configs and try to keep them in sync? 03:11:42 I'm trying to figure out how much I need to replicate and how much can all point to common stuff 03:11:51 nope. incidentally, my ORA articles are related to Apache. 03:12:21 "point to common stuff" - with "Include"? 03:13:45 ah, I think I found it. 03:14:11 I'm setting up dev, test, and production servers and trying to duplicate as little as possible 03:14:53 the Apache I'm using comes with Oracle and has all sorts of install dirs, and I couldn't see which parts were our app and which were base install, but I think I figgered it out 03:14:57 do you want multiple servers runnign at a time? 03:15:28 essentially, yes. production on port 80, test on 8000, dev on 7000, that sort of thing 03:16:01 you could use one config file for them all. 03:16:13 and use Define's to pick and choose your production/dev stuff 03:16:14 and then command line which ones you wanted. 03:16:21 the kink in the plan is that the app is Java servlet based, and the server needs to be bounced every time the app gets updated 03:17:04 and that's on your prod server? 03:17:17 urgh. tcpdump crashed my machine 03:17:30 how do defines work in Apache configs? I was thinking of just using regex replacements 03:17:40 kmacleod: working on an example for ya. 03:18:57 [[[ 03:18:57 ... standard config crap ... 03:18:58 03:18:58 Listen 80 03:18:58 03:19:04 03:19:04 Listen 8000 03:19:04 03:19:05 03:19:06 Listen 7000 03:19:07 03:19:09 ... standard config crap ... 03:19:14 and then to start: 03:19:14 seems simple enough, thx 03:19:17 httpd -DPROD 03:19:20 httpd -DTEST 03:19:21 and so forth 03:19:38 ]]] 03:20:43 that gives you a single config, with if's based on what your command line is. 03:20:44 the hardest part is that this all got landed on me in a tarball of the /home/oracle directory, so I'm sifting through the rubble to find the real nuggets of what they consider the app 03:21:03 heh, i love crap like that. sigh. 03:21:09 * kmacleod wishes again that Oracle were available as an RPM 03:21:21 oracle is a MAJOR PAIN to install 03:21:26 i wasted 3 days on it. 03:21:31 yup. 03:21:36 oracle is a major pain in general, really. 03:21:54 it was like: how can we make a database to best torture petty humans 03:22:11 By comparison, IBM's DB2, which is packaged, took me 45 minutes the first time, and that's because 1) I did it from CDROM, and 2) Solaris's pkgadd is slow 03:22:18 s/torture petty humans/justify our support costs/ 03:22:43 or you can be cynical sure... 03:22:59 I've seen worse, th 03:23:00 tho 03:23:01 AaronSw: when have i ever been logical? ;) 03:23:34 damn, now I can't think of their name -- one the big DB third parties 03:23:46 postgres? 03:23:56 openbase? 03:24:16 no, top-tier. writes apps, dev tools, monitors for the big DBs 03:24:24 coldfusion? 03:24:39 hmm. 03:24:57 any case, not only were they not a package install, not only did they not have a Unix-based installer, noooo.... 03:25:15 it's in JAVA! 03:25:24 they had an NT-based installer that installed over the net using telnet/rcp 03:25:38 oh, that's better than i had to put up with 03:25:52 "Place this CD-ROM on an NT-Server, give us your root password, and we'll talk to you in a little while" 03:26:00 heh, heh 03:26:07 heh 03:26:45 i had this horrible java thing that took approx. 3 hrs to load and crashed every time it didn't like you, and turned out not to be comparible with the version of RedHat i had. 03:26:46 the only thing I hate worse is vender who *have no* standard install media, and place on-site consultants who install the software for you 03:27:32 anyone on a win machien with perl installed? 03:27:37 your left with, "uh, ok, and how exactly are we supposed to maintain backups, test/dev servers, etc." 03:28:11 heh, pay them, of course 03:28:12 no win here, only at work, and they only Cygwin, not ActivePropriState 03:28:47 i wanna replace my homebrew http daemon in ampheta with http::daemon. also want to add accept and fork, but dunno how well its supported. 03:28:51 I must be getting sleepy, that was a stupid slam ;) 03:29:04 especially since i have gui threads to listen for as well. 03:29:14 heh 03:29:29 sbp does, i bet 03:29:33 and tav 03:30:12 ActiveState: the quietest open-source fork ever accomplished 03:30:36 they forked? 03:30:57 for most intents and purposes, yes, they maintain their own distribution 03:31:11 with Larry as the nominal lead, mind you 03:31:17 Larry? 03:31:20 Oh... 03:31:35 heh. "Larry who?" 03:31:50 i was like, i don't know any larry who works on Python 03:32:03 :-) 03:32:05 heh 03:32:34 Morbus: how well does Perl fork (no pun intended) under Windows? 03:32:46 and/or thread 03:32:48 kmacleod: i don't know. 03:32:51 it doesn't thread. 03:32:58 i know i've run some POE stuff that forked. 03:33:04 but i don't know how well its supported. 03:33:12 thats why ampheta is single threaded. 03:33:30 it kinda uses a queue system, to shift between GETing, webserving, and gui refreshing 03:33:32 ya, I'm not sure how well it'd work for you. 03:34:20 * kmacleod needs a weblog module for MoinMoin 03:34:38 .google weblog moinmoin 03:34:40 weblog moinmoin: http://lists.infoanarchy.org/pipermail/p2pj/2001-March/000103.html 03:34:54 i didn't know infoanarchy hosted lists. hm 03:35:59 oh, they seem to be using epy which i think is twisted 03:36:28 jeez, i can't find any good simple example of a accept-and-fork http::daemon. 03:36:36 i have to be looking in the wrong place. 03:36:45 Aha! "��Iǻ��ǉ�C��f�b�N�X�y�[�WǕmod_snake�ǵǟ��ǵǇ�B" 03:38:30 * rillian 's client didn't display that properly 03:38:48 Me neither. 03:39:07 * kmacleod desires so badly to rpm -e mozilla, but too many useful utilities, like Galeon, depend on it. 03:41:01 argh. perlmonks is down. 03:42:29 kmacleod: this touches a bit on win32 forking. 03:42:29 http://www.webtechniques.com/archives/2000/03/junk/ 03:49:35 hum 03:50:10 .spell midevil 03:50:14 potential spellings for midevil are: medieval, mediaeval, Medieval, medievally, medievals, medially, mediative, mediaevals, mediately, Middle 03:50:26 [cough] jeremiah 03:51:24 :) 03:52:00 jesus 03:52:05 I am showing up on so many google searches now 03:52:21 bizzare 03:52:22 heh 03:52:30 no, it's the dave effect. 03:52:33 perhaps you're searching for the wrong things 03:52:49 http://referers.userland.com/staticSiteStats/referers?group=radio1&site=0001189 03:53:39 lol: http://radio.weblogs.com/0001015/images/2002/01/29/somethingfunnyhere.gif 03:53:41 probably the whole weblog effect 03:53:56 yeah 03:54:59