Semantic Web Interest Group IRC Chat Logs for 2003-07-27

02:15:39 <AaronSw> hello

02:44:26 <sandro> heyo, Aaron.

02:55:53 <AaronSw> ooh, new useful datetime module in python. and http://s.keim.free.fr/tz/doc.html

03:01:19 <DanC_g> G:|In-Room Chat as a Social Tool

03:01:25 <dc_rdfig> Titled item G.

03:01:28 <DanC_g> G:by Clay Shirky

03:01:28 <dc_rdfig> Added comment G1.

03:01:36 <DanC_g> G:12/26/2002

03:01:36 <dc_rdfig> Added comment G2.

03:05:05 <DanC_g> G:sounds familiar

03:05:06 <dc_rdfig> Added comment G3.

03:05:52 <AaronSw> G:I was just reading [another Shirky article|http://www.shirky.com/writings/group_enemy.html] where he talks about the power of conferene calls + IRC but gives credit to Joi Ito.

03:05:52 <dc_rdfig> Added comment G4.

03:08:54 <DanC_g> hmm... when did we start doing IRC+phone in W3C? I think it was 1995.

03:09:08 <DanC_g> that group_enemy article is good, though.

03:09:21 <DanC_g> I think Sandro and I read it a few weeks ago.

03:10:34 <DanC_g> G:we do that regularly in W3C meetings; e.g. [http://www.tbray.org/ongoing/When/200x/2003/07/23/TAG-0723|TAG in Vancouver]

03:10:34 <dc_rdfig> Added comment G5.

03:13:02 <DanC_g> ooh... gaim has a toolbar applet, kinda like iChat

03:17:17 <sandro> I use Psi for non-irc; it's got a nice toolbar thing.

03:20:20 <DanC_g> any reason you don't use psi for IRC?

03:20:42 <sandro> If it can do IRC, I don't know how.

03:21:24 <sandro> And the interface doesn't seem as good for it as XChat, but maybe....

03:22:01 <sandro> no tabbed windows, for instance. Harder to manage 12 conversations at once, I think, but maybe I just haven't really tried.

03:22:32 <DanC_g> I rely on tabbed windows.

03:22:47 <mdupont> hi DanC_g danbri sandro AaronSw

03:23:02 <DanC_g> gaim's defaults seem to have lots of windows, but I found preferences to make it act an awful lot like xchat

03:25:35 <sandro> Psi's protocol trace window almost makes up for every possible flaw (it's so cool: view source!). :-)

03:25:45 <DanC_g> I see.

03:25:58 <DanC_g> chupto, mdupont?

03:26:29 <mdupont> chupto? what is that? how are you doing, up also so early?

03:26:41 <DanC_g> chupto = "what are you up to"

03:28:13 <DanC_g> early? it's 10:30pm here... the beginning of the 10pm-2am maximum productivity window ;-)

03:29:34 <mdupont> hehe, i am up early, 10:30 pm, you must be ... 22:00 - 05:00 =~ 7hrs, east coast?

03:29:37 <DanC_g> why do so many blogs use microscopic fonts?

03:30:03 <mdupont> i wonder why my blog content got archived off so quickly

03:30:04 <DanC_g> I'm in the central US timezone... one west of east coast time

03:30:09 <AaronSw> I think they assume you use IE6 with default settings

03:30:15 <AaronSw> which is huge fonts

03:30:29 <sandro> Because so many people have big monitors they drive at low resolutions? Makes no sense. I have no idea.

03:30:38 <sandro> Ahhhh.

03:31:51 <DanC_g> IE --

03:32:43 * DanC_g wants to switch dm93.org from Zope to svn, but hesitates to get started, not knowing how long it will take and not wanting to stop half way thru

03:32:56 <sandro> content authors working around the user's (naively) stated preferences --

03:33:23 <sandro> svn?

03:33:25 <DanC_g> ooh... a subversion book online... http://svnbook.red-bean.com/html-chunk/

03:34:05 <AaronSw> why subversion? webdav support?

03:35:35 <sandro> how is svn a content-management system?

03:35:57 <sandro> like, compared to Zope.

03:36:14 <DanC_g> the main reason I use Zope is that it supports PUT

03:36:22 <sandro> Ah.

03:36:33 <DanC_g> I'd like to have both PUT and something like cvs commit

03:36:43 <DanC_g> If I understand correctly, that's what svn gives.

03:37:09 <DanC_g> Oh... the other Zope features I use are access control and ZWiki. But I expect I can set up work-alikes under apache with little hassle.

03:37:28 <AaronSw> you could also have used mod_put and mod_dav

03:37:31 <AaronSw> s/and/or/

03:37:51 <DanC_g> I tried mod_dav a while back... it seemed rickety. and I didn't see integrated undo support

03:38:50 <DanC_g> I meant: the main reason I use Zope is that it supports elephant-never-forgets PUT

03:39:03 <AaronSw> aha

03:39:10 <DanC_g> grumble... svn book sets fonts size to 12pt

03:39:46 <DanC_g> and serif. who the *heck* like reading online in a serif font?

03:39:50 <DanC_g> likes

03:39:58 <AaronSw> Tim Bray

03:40:54 <AaronSw> on a vaguely related note, i'm interested in recommendations for tools with which to build Web apps in Python

03:47:23 <mdupont> I would like to see these attributes added to foaf : ssh2PubKey ssh1PubKey gpgPubKey Task Location IrcID Occupation

03:48:06 <AaronSw> what's the use case for ssh pub keys?

03:50:48 <mdupont> well,

03:51:00 <mdupont> just to say, my ssh public key is located here

03:51:04 <mdupont> or has this value

03:51:27 <DanC_g> and then what?

03:51:51 <DanC_g> hmm... I've avoided administering an apache server thus far. I really loath the pseudo-XML config stuff.

03:52:21 <AaronSw> the security-hole-of-the-week is fun too

03:54:45 <mdupont> and then what?

03:55:02 <DanC_g> i.e. so you say your ssh public key is located here... then what?

03:55:06 <mdupont> well, lets say, somone wants to set me up an account

03:55:23 <mdupont> i submit a link to my foaf as an admission request

03:55:32 <mdupont> and it reads my public key from that location

03:55:56 <mdupont> that way, i dont have to type it in again

03:55:57 <DanC_g> and ... ?

03:56:14 <mdupont> then they can make a copy of that public key

03:56:48 <DanC_g> er... if anybody believes X is your public key just because your foaf file says so, let me know who it is so I can steer clear.

03:59:36 <sandro> I'm not sure it's as bad as that, DanC. If I wanted to permit some file to 20 of my friends, I think it would be reasonable to use the keys in their FOAF files. Obviously the key distribution is the weak point in the system, but that's often the case....

04:00:22 <DanC_g> if they're your friends, then they're not believing "just because your foaf file says so". They also have other data.

04:01:59 <sandro> I thinking along the lines of wanting to permit a file to someone who'se foaf file is at URL X. Who that person is may not be clear, to be sure.

04:02:40 <DanC_g> hmm... interesting... you can do diffs offline in svn (http://svnbook.red-bean.com/html-chunk/apas03.html)

04:03:51 <DanC_g> there's little point if using public key cryptography if you're going to import public keys without doing something like checking the fingerprint over the phone

04:08:39 <DanC_g> if you had a pgp signed foaf file, that would be a handy way to distribute ssh2 keys

04:09:02 <DanC_g> i.e. it's straightfoward to use a pgp trust relationship to establish an ssh trust relationship

04:11:36 <sandro> I think you're missing an area of appplication there. What you're describing is of course the normal though; I'm exploring more in part because that normal mode doesn;t seem to be catching on. (of course there are many many reasons for that.)

04:12:18 <DanC_g> # /etc/init.d/apache2 start

04:12:18 <DanC_g> Starting web server: Apache2Syntax error on line 1 of /etc/apache2/mods-enabled/dav.conf:

04:12:18 <DanC_g> Invalid command 'DAVLockDB', perhaps mis-spelled or defined by a module not included in the server configuration

04:14:13 <DanC_g> ^ what you get after: apt-get install libapache2-dav-svn

04:14:25 <mdupont> DanC_g i dont see the problem

04:14:25 <DanC_g> to be fair, it's documented as "alpha"

04:14:42 <mdupont> the foaf files will be submitted via a secure connection to the dotgnu server

04:14:58 <mdupont> the public key will be transmitted with the foaf file in one big signed package

04:15:02 <DanC_g> ok, if you've got a secure connection, that's different.

04:15:35 <mdupont> yes, the rdf webservices i have been protyping will be using the dotgnu authentication, and security system.

04:15:41 * DanC_g reads "Networking a Repository" chapter http://svnbook.red-bean.com/html-chunk/ch05s04.html

04:15:59 <DanC_g> any reason not to make your own sshKey property?

04:21:31 <DanC_g> hmm... this svnbook discusses apache 1, I think; debian sid uses apache 2

04:25:12 <mdupont> yes DanC_g i have my own

04:25:22 <mdupont> for dotgnu:foaf i added these properties

04:25:31 <mdupont> : ssh2PubKey ssh1PubKey gpgPubKey Task Location IrcID Occupation

04:27:47 <DanC_g> aha... "For Apache 2.0, you need to include mod_dav_fs" -- http://dav.lyra.org/pipermail/dav-dev/2002-January/003176.html

04:40:45 <DanC_g> wierd... now there are no error messages when I start apache, and *something* is accepting connections on port 80, but no bytes come back.

04:41:00 <DanC_g> maybe I'll try installing just apache2...

04:42:08 <mdupont> hey burtonator

04:45:31 <DanC_g> hmm... apache2 isn't working... log files are owned by root... is that right?

04:45:46 * DanC_g runs to #debian for help...

05:19:27 * DanC_g gets confirmation that there's a bug in apache2-mpm-perchild

05:36:53 <DanC_g> aha... work-around: use prefork. per http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=202557

05:37:02 <jordan> what do you think about http://www.prescod.net/groves/shorttut/

05:41:48 <DanC_g> argh! why do the apache docs only cover Basic? apache supports Digest. use of Basic should be LOUDLY warned against.

05:42:42 * jordan has no idea what DanC is talking about

05:42:52 <AaronSw> yeah, digest support in apache is bizarre

05:43:00 <DanC_g> said apache docs: http://httpd.apache.org/docs-2.0/howto/auth.html

05:44:17 <DanC_g> bizarre? how so?

05:44:21 <AaronSw> "This module implements HTTP Digest Authentication. However, it has not been extensively tested and is therefore marked experimental."

05:44:44 <AaronSw> and "Since digest authentication is not as widely implemented as basic authentication, you should use it only in controlled environments."

05:44:57 <AaronSw> it's like they're trying to _discourage_ it

05:46:53 <DanC_g> sigh... apache has been just about rewritten from the ground up, but they're still in the 1980s when it comes to authentication. sigh.

05:47:06 <DanC_g> this is the sort of malarky that sent me in the direction of Zope in the first place.

05:47:57 <DanC_g> There was some apacheweek article about "duct tape, bailing wire, and PUT". well, not really... but it showed that you had to write a custom PUT script (in, say, perl) to get apache to support PUT.

05:48:48 <DanC_g> sigh sigh sigh. even IE groks Digest.

05:48:53 <DanC_g> and the WebDAV specs mandate it.

05:49:04 <jordan> rdf:about="" should look to the xml:base of the document, right?

05:49:11 <DanC_g> yes, jordan

05:49:52 <DanC_g> google nominates this article on apache2 and Digest... seems pretty good... http://www.thethirdbridge.com/doItYourself/authenticationOnApache2.html

05:50:48 <jordan> if no xml:base is present, is the default ns considereD?

05:51:22 <AaronSw> jordan, no, the location of the document

05:51:44 <jordan> ah ok

05:53:22 <DanC_g> is there really no market for a lean little HTTP server that groks PUT?

05:53:41 <DanC_g> like for a church or small business?

05:54:27 <DanC_g> I guess with hosting services at rediculously low prices (did sandro point to a $10/year service?) there isn't.

05:54:47 <DanC_g> (must have been $10/month)

05:55:30 * DanC_g calls it a night.

06:16:08 <mdupont> has anyone thought of a wifi or wap support, wml and mobile authentication?

09:13:46 <mdupont> DanC dtd2xsd.pl is cool!

09:14:08 <chaalsNCE> night all

09:34:49 <mdupont>http://demo.dotgnu.org/wordnetws.dgmx

09:34:50 <dc_rdfig> A: http://demo.dotgnu.org/wordnetws.dgmx from mdupont

09:35:09 <mdupont> A:|Experimental wordnet RDF webservice

09:35:09 <dc_rdfig> Titled item A.

09:35:24 <mdupont> A:Allows clients to make statements using wordnet predicates

09:35:25 <dc_rdfig> Added comment A1.

09:36:13 <mdupont> A:[http://demo.dotgnu.org/wordnetws.dgmx#WordNet.Addantonym|Addantonym] for example, allows you to make an antonym for an subject

09:36:13 <dc_rdfig> Added comment A2.

10:37:07 <larsbot> larsbot is now known as lars|meet

10:44:20 <mortenf>http://www-106.ibm.com/developerworks/xml/library/x-rdfprov.html

10:44:20 <dc_rdfig> B: http://www-106.ibm.com/developerworks/xml/library/x-rdfprov.html from mortenf

10:44:34 <mortenf> B:|XML Watch: Tracking provenance of RDF data

10:44:34 <dc_rdfig> Titled item B.

10:44:48 <mortenf> B:By Edd Dumbill

10:44:48 <dc_rdfig> Added comment B1.

10:47:26 <mortenf> raises a question: when multiple instances (over time) of a single rss channel are merged, the elements of the rdf:Seq will each be assigned multiple items number - how is one supposed to handle that?

10:48:04 <danbri> normally one retracts and replaces an rss with it's successive re-descriptions...

10:48:17 <mortenf> yeah, but then there's no history

10:48:32 <danbri> well you can keep the old triples in there somewhere...

10:49:10 <mortenf> that's where my problem occurs, eahc time a channel is updated, an item drops off the end, and a new one with rdf:_1 arrives

10:49:11 <danbri> its like keeping back history of morten-foaf.rdf ... you probably want to hang on to it, while not trying to reason about the world as if all versions of it were simultaneously true

10:49:38 <mortenf> hmm, i guess

10:49:42 <danbri> (I'm not arguing its all simple, you're right there are practical issues here for ppl building aggregators)

10:51:10 <mortenf> but from a general (non-rss) view, it seems there's a logical flaw in the models somewhere; which is i guess why nobody likes Seq...

10:51:52 <danbri> its a different rdf:Seq each time, representing different orderings

10:52:03 <dajobe> different bnode, typically

10:52:12 <danbri> yup

10:52:22 <mortenf> that too, but it's still the same channel's rss:items property

10:52:52 <danbri> representing change is always tricky

10:52:57 <mortenf> and it seems silly to add a new property with 9 out of 10 list elements being the same

10:54:23 <dajobe> for rss apps, those are just structural parts in order that you can find the (list of items) that you care about

10:54:35 <dajobe> chcking how that list compared to any previous one is for the rss apps to do

10:55:06 <mortenf> yep, but it seems like it's a general issue, also re duplicates (even if they're strictly not duplicates)

10:55:51 <dajobe> there are lots of general issues. With redland-contexts, I addressed 1 approximately

10:56:24 <dajobe> for eg. if the backend store did transactions and rollbacks, that might have helped you with your possible worlds things

10:56:44 <mortenf> hmm

11:00:16 * danbri tries edd's Fraggle package, gets confused by runing it when it doesn't have #!/usr/bin/python at top

11:01:08 <danbri> first thing it does is run 'import RDF', which seems to be an X window screen capture to postscript utility...

11:01:22 <danbri> run properly, it works fine :)

11:01:23 <mortenf> heh

11:07:58 * mortenf now has some test programs for a mysql backed redland storage

11:08:24 <dajobe> ooh

11:08:46 <mortenf> mind you, not an implementation of storage yet...

11:09:21 <mortenf> but at least i'm getting the hang of the api

11:36:12 <arnarl> hi

12:56:19 <tav> tav is now known as tav|offline

15:12:49 <lars|meet> lars|meet is now known as larsbot

17:58:29 <jordan> The validator complains about: <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2003-07-17</dc:date> -- why isn't this allowed?

18:05:42 <danbri> looks ok to me

18:08:46 <jordan> validator says rdf:datatype is misused in this context

18:10:31 <danbri> does it parse ok without?

18:10:45 <danbri> ie. it realises dc:date is a property not a class, etc.

18:20:53 <jordan> Without rdf:datatype it returns Error: {E202} Expected whitespace found: '2003-07-17'.

18:21:29 <danbri> can you post a more complete example?

18:28:42 <jordan> The example is just <rdf:RDF namespaces...> <dc:date ...></dc:date> </rdf:RDF> -- that's all I'm feeding to the validator

18:30:16 <danbri> you'r starting with a property element not a node element

18:30:33 <danbri> <rdf:Description><dc:date>...</dc:date></rdf:Description> or similar is needed

18:31:34 <jordan> right

20:05:03 <tav|offline> tav|offline is now known as tav

21:03:01 * mattb wonders what the simplest python n3 parser to use is

21:03:19 <mattb> (for parsing the rdf model subset of n3)

21:03:20 <dajobe-lap> mattb: sbp's got a simpleish one

21:03:38 <mattb> dajobe-lap: tried that, it didn't cope with the [ ... ] syntax afaics

21:03:50 <dajobe-lap> pfft, well who needs that :)

21:04:22 <dajobe-lap> I think when you get to [] etc. you need a proper BNF/generator typically and the ambiguities start to pick up

21:04:56 <mattb> yeah, i got the complete "[ ... ]" subclause returned as an object literal

22:31:58 <burton_> burton_ is now known as burtonator

23:28:56 <danbri_lap>http://www.eblong.com/zarf/bookscan/index.html

23:28:57 <dc_rdfig> C: http://www.eblong.com/zarf/bookscan/index.html from danbri_lap

23:29:08 <danbri_lap> C:|Everything I Wanted to Know...

23:29:08 <dc_rdfig> Titled item C.

23:29:21 <danbri_lap> C:...about using the barcode scanner I just bought for cataloguing my books.

23:29:21 <dc_rdfig> Added comment C1.

23:29:57 <danbri_lap> C:I got as far as having a little Amazon lookup script, isbn.rb, and then hit the snag described here, ie. going from barcode to ISBN is non-trivial.

23:29:57 <dc_rdfig> Added comment C2.

23:30:00 <danbri_lap> C:Seems fixable though.

23:30:01 <dc_rdfig> Added comment C3.

23:34:15 <danbri_lap> C:Comes with a bunch of handy Python scripts.

23:34:16 <dc_rdfig> Added comment C4.