Semantic Web Interest Group IRC Chat Logs for 2001-06-14

01:21:01 * AaronSw waves

01:21:10 <danbri> hi there!

01:22:12 * AaronSw is fighting with his mail client to give him the contents of his inbox

01:22:17 <AaronSw> ahh, for rdf here

01:26:40 <danbri> have a look at the IMAP Perl libraries maybe?

01:27:04 <danbri> I have an IMAP2RDF.pl perl script knocking around somewhere for RSS generation...

01:28:04 <AaronSw> no, that's the problem -- i need to get it out of my current client so i can get it into IMAP

01:28:16 <AaronSw> i'm doing some messy applescript coding

01:28:25 <AaronSw> thanks, though

01:28:47 <AaronSw> and i'll want that script when i do get them into IMAP!

01:31:00 <DanC> I've done a bunch of IMAP hacking in python; it's linked from 2000/01/sw/

01:31:28 <AaronSw> a search for imap loses

01:31:35 <DanC> look for email

01:32:10 <DanC> something like 2000/04/maillog2rdf/imap_sort.py

01:32:11 <AaronSw> aha, winning. yes, i remember this

01:32:26 <DanC> the most useful bit is mid_proxy.py

01:32:31 <DanC> i.e. mid: support!

01:32:59 <AaronSw> right-o -- i'm trying to get something like an rdf-based interwingle going

01:33:16 <DanC> well, today's victory: merging data from my palmpilot and the W3C list of WGs.

01:33:35 <AaronSw> really? how so?

01:33:43 <DanC> intertwingle: is that a Nelson-ism or a jwz-ism?

01:33:50 <AaronSw> jwz-sim

01:33:55 <AaronSw> err jwz-ism

01:33:58 <AaronSw> in this context

01:34:09 <DanC> how so: don't think I have time just now, and unfortunately, the data is confidential.

01:35:02 <DanC> family time approaches.

01:35:29 <DanC> another small victory: a teeny HTML forms-based editor for palmpilot data added to pdkb.pl

01:35:54 <DanC> i.e. you can edit the description and notes fields.

01:36:23 <DanC> it's sorta the wrong hack though: it goes straight from .pdb format to HTML forms, rather than using RDF in between.

01:36:25 * danbri catches up

01:36:47 <AaronSw> http://www.mozilla.org/blue-sky/misc/199805/intertwingle.html

01:36:48 <dc_rdfig> A: http://www.mozilla.org/blue-sky/misc/199805/intertwingle.html from AaronSw

01:36:52 <AaronSw> A:|Intertwingle

01:36:52 <dc_rdfig> titled item A

01:37:11 * DanC wonders if we can search for links to A in documents written by me.

01:37:37 <danbri> re intertwingle http://www.mozilla.org/rdf/doc/ ->

01:37:38 <AaronSw> A:by [jwz|http:///www.jwz.org]

01:37:38 <dc_rdfig> commented item A

01:37:52 <danbri> oh, you beat me to it :-)

01:38:06 <DanC> ArtB: see related noodling in [my research notebook|http://www.w3.org/People/Connolly/drafts/web-research.html]

01:38:14 <DanC> A:see related noodling in [my research notebook|http://www.w3.org/People/Connolly/drafts/web-research.html]

01:38:14 <dc_rdfig> commented item A

01:39:16 <DanC> hmm.. my research notebook mentions it but doesn't link to it.

01:39:24 <AaronSw> yes it does

01:39:40 <DanC> oh... right... there it is

01:41:43 <DanC> the results of merging my palmpilot data with the W3C WG stuff looks like {{{

01:41:46 <DanC> :13276350 a my:WeeklyAppointment,

01:41:46 <DanC> my:WorkEvent,

01:41:46 <DanC> pd:Indefinite,

01:41:46 <DanC> org:RemoteMeeting;

01:41:46 <DanC> my:groupMailbox <mailto:w3c-semweb-cg@w3.org>;

01:41:48 <DanC> my:home <http://www.w3.org/2001/sw/CG/>;

01:41:48 <DanC> pd:date "2001-03-26";

01:41:50 <DanC> pd:description "Sw cg" .

01:41:52 <DanC> }}}

01:42:11 <AaronSw> you do the match based on pd:description?

01:42:38 <DanC> no... I manually wrote an RDF document to link palm record IDs with group home pages.

01:42:59 <DanC> and those group home pages are used in our groups.rdf thingy, and away I go!

01:43:47 <DanC> so I did the group home page manually, but I get the group email address for free.

01:44:30 * DanC wanders off...

08:42:36 <ChanServ> This channel is logged and blogged. See: http://logicerror.com/rdfIRCWelcome

08:54:57 <ChanServ> This channel is logged and blogged. See: http://logicerror.com/rdfIRCWelcome

13:31:36 * sandro just realized that Canonical XML solves the problem of the RDF Spec BNF handwaving over XML syntax.

14:02:41 <danbri> http://www.w3.org/2001/sw/

14:02:42 <dc_rdfig> B: http://www.w3.org/2001/sw/ from danbri

14:05:04 <danbri> B:|W3C Semantic Web Activity

14:05:05 <dc_rdfig> titled item B

14:05:25 * danbri waves at joseph, demos the chump bot

14:05:37 <danbri> type B:blah blah....

14:05:48 <danbri> see http://rdfig.xmlhack.com/ for www pages

14:07:22 <danbri> B:The coolest thing since sliced bread (according to emiller :)

14:07:22 <dc_rdfig> commented item B

14:08:49 <reagle-ho> B: it is nifty

14:08:50 <dc_rdfig> commented item B

14:11:29 <reagle-ho> B: [http://www.w3.org/| W3C] is where I [http://www.w3.org/People/Reagle/work-style.html | work]

14:11:29 <dc_rdfig> commented item B

14:11:49 <timbl-away> Sandro, not sure how cannonical XML solves the RDF spec problem. Surely the XML spec does.

14:12:17 * timbl-away suspects the beeps are coming from the brown things (notices?) sent by chump and disables beeps on notices.

14:14:19 <danbri> joseph, the chump code is at http://usefulinc.com/chump/

14:15:17 <danbri> bijan?

14:19:08 <DanC_> timbl? help? raise internalError # We have something in an unknown state in the queue

14:19:25 <DanC_> I'm trying to use a rule:

14:19:26 <DanC_> { v:doc log:uri [ log:startsWith "http://lists.w3.org/Archives/Public/" ] }

14:19:26 <DanC_> log:implies { v:doc a :PubliclyReadable }.

14:35:01 <timbl-away> Got a file?

14:35:37 <timbl-away> or just that?

14:37:28 <timbl-away> If you could check in a test case into test/includes?

14:44:57 <sandro> TimBL, the problem I'm referring to is that the RDF spec has a BNF grammar for RDF, with productions like aboutAttr ::= ' about="' URI-reference '"' which means in RDF you can't have spaces around the equals sign, etc. Then there's human text saying the BNF isn't exactly real, that you need to understand XML to make sense of it. That's hard to formalize in code. But saying the BNF is for the canonical form of the XML is simple and precis

14:47:06 <sandro> The grammar does imply *a* canonicalization, but inferring that canonicalization algorithm from the grammar (on things like what is canonical order for attributes?) is hard. I was working on it, when I realized we already have a standard.

14:50:38 <sandro> (In particular, inferring the canonical ordering of attributes whose names are themselves not specified, as in the propAttr production, is particularly difficult.)

14:56:33 <timbl-away> Oh I see.

14:58:17 <sandro> good. :) (it is a pretty minor nit about the RDF spec, which probably only matters to me since I'm trying to make a machine read the spec instead of a human.)

14:58:28 <timbl-away> So you can build a parser by genenerating canonical RDF and parsing that.

14:58:51 <timbl-away> So long as your grammar is more general than the cannocical spec you are ok.

14:59:31 <sandro> if by "canonical RDF" you mean XML-canonicalized RDF, then yes.

14:59:49 <timbl-away> sorry yes i do.

15:00:19 <sandro> (I'm kind of hoping "canonical RDF" will mean something else one of these days. :-)

15:00:52 <timbl-away> thinks.. maybe one can let XML folks specify RDF in the same way... make an XML schema for RDF in one of its many serializations (a cannonical one?) and then just say "other RDF serializations are fine too".

15:01:28 <timbl-away> This way, the XML folks and RDF folks can ignore the vaguaries of each other's syntaxes.

15:03:10 <timbl-away> (Someone tried recently to make the point that XML was better than RDF because a very small device likes to know what order the data is coming in. I;m not sure if this makes sense. But anyway).

15:05:03 <sandro> I'm not sure I understand the first point, but it does seem like RDF's many equivalent XML forms make it hard to process as XML -- it has to be handled at the application layer anyway.

15:05:41 <sandro> I'm not sure XML makes sense for small enough devices either. :-)

15:06:02 <sandro> Hopefully people can stick with the model, even if they need to opitimized their syntax for particualr applications.

15:08:56 * DanC_ tries to catch up, looks for EricM...

15:10:10 <DanC_> yes, Sandro, we could use C14N to deal with some RDF grammar issue; i.e. we could give a BNF for RDF in canonical form, and say "any document whose canonical form is in the language generated by this grammar is an RDF document".

15:10:38 <DanC_> (by "RDF in canonical form", I also meant "RDF in xml-c14n form")

15:12:42 <DanC_> "to make a machine read the spec instead of a human" is an important QA excercise, in my experience.

15:12:50 <sandro> Yep -- it's probably an obvious thing to do when you're familiar with the parts.

15:13:45 <DanC_> if you have a moment, sandro, do send to www-rdf-comments your suggestion to use c14n to tighten up the RDF grammar; point to the "formal grammar" issue in the issues list if it's at all convenient.

15:14:03 <sandro> The interesting part, of course, is the specifying the semantics as well as the syntax. :-) [ right now I'm toying with the syntax for specifying the semantics. ]

15:14:27 <sandro> sure/

15:14:30 <DanC_> yes, TimBL, I should add a test for my bug.

15:17:58 <DrewMcD> DanC, can you answer a question or two about cwm?

15:18:45 <DanC_> maybe ;-)

15:19:03 <DrewMcD> Why is it called the "Closed World Machine"?

15:19:10 * sandro laughs

15:19:40 <DanC_> well, I was trying to get TimBL to appreciate the difference between "closed" systems like prolog and SQL on the one hand, ...

15:20:09 <DanC_> ... and access-limited-logic on the other. in SQL and prolog, you have to have all the data in the engine before you start processing.

15:20:30 <DanC_> in ALL, you can learn more as you process the query (as long as what you learn is monotonic).

15:21:04 <DrewMcD> Okay. Question 2: Does cwm use unification or a simpler notion of matching?

15:21:05 <DanC_> when TimBL first explained his design, it seemed like (yawn) yet another SQL/prolog thingy. So I said "it's a closed-world machine." and the name stuck.

15:21:16 <DanC_> since then, it's become a misnomer! cwm can learn stuff on the fly.

15:21:43 <DanC_> I think cwm's matching is simpler than unification, but I'm not sure.

15:22:14 * DanC_ noodles a bit...

15:22:29 <DanC_> I'm just not sure at all. maybe it is unification after all.

15:22:57 <DrewMcD> Of course, there's unification and unification. Most Prologs dont' do the full thing.

15:23:25 <DanC_> oh. I'm only familiar with one form of unification. (the one in the algernon source code ;-)

15:23:56 <DrewMcD> Prolog usually turns off the "occur check" for higher speed.

15:24:01 <DanC_> umm... unification is explained in SICP, no? (looking...)

15:24:17 <DanC_> occur check: oh yeah... I just read about that in the swi-prolog manual.

15:24:43 <sandro> The occurs check makes sure that things don't unify with things containing themselves, right?

15:24:50 <DrewMcD> Right.

15:25:35 <DrewMcD> Question 3: Does cwm do backward chaining, forward chaining, or both?

15:25:55 <DanC_> yes, SICP p. 459 explains unification ala simultaneous equasions. TimBL talks about simultaneous equasions when he explains cwm to me a lot. I usually don't follow.

15:26:20 <DrewMcD> What's SICP?

15:28:03 <DanC_> SICP is short for Structure and Interpretation of Computer Programs, by Abelson and Sussman. Not the book I learned from, but something of a canon, I gather.

15:28:18 <DanC_> I got a free copy for reviewing something a few years ago.

15:28:57 * DrewMcD wants to ask Question 3 again

15:29:09 <DanC_> oh... re 3: only forward chaining.

15:29:30 <DanC_> the --rules forward chains all rules it can find, once; --think iterates until there are no new conclusions.

15:29:41 <DanC_> --rules command-line option, that is.

15:30:34 <DanC_> Jos DeRoo has an implementation that does backward chaining. he explains it using the terms "robinson resolution" and "euler path detection" which I don't yet grok.

15:31:18 <DrewMcD> "Resolution" is basically unification + modus ponens. "Euler" I don't grok either.

15:31:45 <DrewMcD> I think that's all the questions I have just now.

15:31:48 <DanC_> his code is fast.

15:31:48 <sandro> And I have an implementation that takes the same input as cwm and translates it to tabled prolog for xsb. Some issues in comparision, since cwm has the different operating modes (eg --rules, --think, ...).

15:35:03 <DanC_> wow... the full text of SICP is online...

15:35:19 <DanC_> http://mitpress.mit.edu/sicp/full-text/sicp/book/

15:35:19 <dc_rdfig> C: http://mitpress.mit.edu/sicp/full-text/sicp/book/ from DanC_

15:35:40 <DanC_> C:|Structure and Interpretation of Computer Programs - 2nd edition

15:35:40 <dc_rdfig> titled item C

15:35:53 <DanC_> C:by Harold Abelson and Gerald Jay Sussman with Julie Sussman

15:35:53 <dc_rdfig> commented item C

15:37:16 <DanC_> C:explains, among other things, [Rules and Unification|http://mitpress.mit.edu/sicp/full-text/sicp/book/node100.html]

15:37:16 <dc_rdfig> commented item C

15:39:29 <DanC_> C:it would be nice to see the [swap/cwm rules engine|http://www.w3.org/2000/10/swap/] explained relative to the terms from this widely-read source.

15:39:29 <dc_rdfig> commented item C

15:39:53 <DrewMcD> I'm not sure it's that widely read.

15:39:59 <DanC_> no?

15:40:09 <DrewMcD> Correction: It's very widely read, but I don't think it's the standard source for unification.

15:40:54 <DrewMcD> In fact, I don't think there is a standard source.

15:40:55 <DanC_> but (a) it's available in full-text form online, and (b) I suspect/hope it uses terminology consistent with whatever the standard source is.

15:41:11 <DrewMcD> I'll take a look.

15:41:46 <bijan> I recommend PAIP.

15:42:01 <bijan> I believe the Warran Abstract machine tutorial is also on line.

15:42:08 <DanC_> PAIP? chump it, please?

15:42:25 <bijan> (PAIP is not on line, alas).

15:42:33 <DanC_> well, title/author?

15:42:47 <bijan> Yes.

15:42:58 * bijan digging...it's norvig.

15:43:33 <DanC_> Hi Eric. last night I succeeded in merging/joining my palmpilot data with the groups.rdf stuff. I've got a rule that tells which of the appointments in my palmpilot are governed by W3C Remote meeting process.

15:43:54 <em_mit> excellent!

15:44:01 <em_mit> sounds like you were busy :)

15:44:39 <DanC_> hmm.. the "rules and unification" section I cited above isn't the same as the section on p.459 that talks about simultaneous equasions...

15:45:02 <bijan> http://www.norvig.com/

15:45:02 <dc_rdfig> D: http://www.norvig.com/ from bijan

15:45:13 <bijan> D:|Peter Norvig's site.

15:45:13 <dc_rdfig> titled item D

15:45:57 <DanC_> C:actually, the [How the Query System Works|http://mitpress.mit.edu/sicp/full-text/sicp/book/node94.html] section is the discussion of Unification I was thinking of.

15:45:57 <dc_rdfig> commented item C

15:46:03 <bijan> D:Links to [Paradigms of AI Programming|http://www.norvig.com/paip.html] and [AI: A Modern Approach|http://www.cs.berkeley.edu/~russell/aima.html].

15:46:04 <dc_rdfig> commented item D

15:46:12 * sandro is very fond of Russell,Norvig

15:46:30 <DanC_> pls tell chump, sandro

15:46:30 <bijan> D:Both with extensive discussions of unification and implementing it.

15:46:32 <dc_rdfig> commented item D

15:47:12 <DanC_> ... so that later I can ask chump, "what was that book abut unification that sandro likes?"

15:47:40 <bijan> Russel/Norvig is AI: A modern approach.

15:47:55 <em_mit> http://boingboing.net

15:47:56 <dc_rdfig> E: http://boingboing.net from em_mit

15:48:10 <bijan> Warren's book 'Computing with logic' I believe I've mentioned before.

15:48:29 <em_mit> E:| Boing Boing - Articles from Mark Frauenfelder

15:48:29 <dc_rdfig> titled item E

15:48:48 <em_mit> E: Stay tuned for future articles regarding RDF and the Semantic Web

15:48:48 <dc_rdfig> commented item E

15:49:46 <bijan> http://www.isg.sfu.ca/~hak/documents/wam.html

15:49:46 <dc_rdfig> F: http://www.isg.sfu.ca/~hak/documents/wam.html from bijan

15:50:01 <bijan> F:|Warren's Abstract Machine: A Tutorial Reconstruction

15:50:01 <dc_rdfig> titled item F

15:50:37 <bijan> F:WAM was the first "high performance" Prolog/unification engine. Many systems are still based on a (modified) WAM.

15:50:37 <dc_rdfig> commented item F

15:54:23 * bijan notes that there is a classic "unification algorithm" paper...that he can't find at the moment.

15:54:31 <DanC> F:I have been looking for an explanation of the WAM that I can grok. Here's hoping...

15:54:31 <dc_rdfig> commented item F

15:55:15 <bijan> F:Eh...the Tutorial is widely lauded as making it understandable...it's still heavy going.

15:55:15 <dc_rdfig> commented item F

15:56:05 <bijan> F:The Tutorial is layered, so you start with a simpler system and gradually modify it into the full blown WAM. Adding things like tail-call optimization, naturally, complicates things.

15:56:05 <dc_rdfig> commented item F

16:03:08 <bijan> VIEW

16:03:23 <bijan> Er..

16:03:29 <bijan> dc_rdif:VIEW

16:03:31 <bijan> Dang

16:03:37 <bijan> dc_rdfig:VIEW

16:03:37 <dc_rdfig> Not understood: VIEW

16:03:41 <DanC_> it's dc_rdfig, view

16:03:47 <bijan> dc_rdfig view

16:03:54 <bijan> Witht he comma?

16:04:07 <DanC_> maybe a colon. use /msg dc_rdfig view to save us the noise

16:04:29 <bijan> That worked, merci.

16:05:02 <DanC_> de rien. [I think]

16:05:25 <bijan> C:From a brief skim, they're terminology doesn't jump out as particularly standard or widespread (in the literature).

16:05:25 <dc_rdfig> commented item C

16:05:37 <DanC_> no? bummer.

16:05:47 <bijan> Not really, no.

16:06:24 <bijan> D:PAIP takes you through some Prolog interpreters and compilers. (Written in Common Lisp.)

16:06:24 <dc_rdfig> commented item D

16:07:50 <DrewMcD> Unification is basically very simple.

16:08:00 <bijan> YEp. The algo is less than a page.

16:08:36 <DanC_> is the bit about matching things that contain themselves the only wrinkle? or is it like not, with zillions of subtle variations?

16:08:43 <DrewMcD> Just take the usual recursive equality algorithm and patch it so whenever an unassigned variable is tested against something else, you assign the variable

16:08:57 <DrewMcD> It's "self-fulfilling equality."

16:09:04 <bijan> Dan: You mean the "occurs check" problem?

16:09:17 <DanC_> occurs: yes

16:09:18 <DrewMcD> The occur check is the *only* wrinkle.

16:09:23 <DanC_> only: good.

16:09:28 <bijan> And it's a biggy ;)

16:09:45 <DrewMcD> It's a biggy when it's relevant. It slows things down when it isn't relevant.

16:09:51 <bijan> Yep.

16:09:55 <DanC_> well, I gather it's expensive, computationally, to check. But it's not very subtle to understand, is it?

16:10:05 <DrewMcD> Of course, there's a vast literature on extending the basic unification algorithm with all sorts of bells and whistles.

16:10:23 <DrewMcD> Understanding the occur check:....

16:10:33 <bijan> In general, speeding up unification, as with all optimization, can make things very hairy.

16:10:59 <DanC_> I hear about "AC unification" now and again, i.e. associative/commutative. that's a straight-forward engineering concern, right? nothing theoretically subtle, is it?

16:11:26 <DrewMcD> Alas, it's quite subtle.

16:11:49 * DanC_ wonders what Drew meant by "the *only* wrinkle".

16:12:31 <DrewMcD> I mean, ignoring extensions and efficiency, putting the occur check into the basic recursive hack is all you need to ensure completeness and correctness.

16:13:46 <DanC_> ok

16:14:07 <DrewMcD> I'll try to think of a clear example of when you need the occur check. Right now I have to go to lunch.

16:14:21 <bijan> Omitting the occur check is the tricky bit ;)

16:14:26 <DanC_> I saw an example in the sw-prolog manual, I think.

16:14:37 <DanC_> swi-prolog

16:15:13 * DanC_ noodles on how to separate the palmagent generic code from my personal PIM rules, from my data...

16:16:08 <bijan> WOw, this is suprisingly good: http://burks.brighton.ac.uk/burks/foldoc/11/81.htm

16:17:53 <DanC_> oh... when Drew wrote "Alas, it's subtle", he meant it=occurs check, not it=AC unification.

16:18:52 <bijan> http://www-lp.doc.ic.ac.uk/UserPages/staff/ft/alp/net/impl/occur.html

16:18:52 <dc_rdfig> G: http://www-lp.doc.ic.ac.uk/UserPages/staff/ft/alp/net/impl/occur.html from bijan

16:19:06 <bijan> G:|Interesting discussion of the Occurs Check Problem.

16:19:06 <dc_rdfig> titled item G

16:21:41 <bijan> I'm afarid I dont' know anything about AC unification. Pointers?

16:22:20 <DanC_> I don't know much about it either.

16:22:42 <DanC_> I'm not sure where I heard about it... somewhere near Boyer/McCune's reasearch, I think.

16:22:56 <DanC_> i.e. Boyer's research stuff and McCune's stuff.

16:23:11 <bijan> Ok.

17:08:53 <DrewMcD> Subtlety: I meant AC unification, not the occur check, which isn't very subtle.

17:09:36 <DrewMcD> Here's an occur-check example: Suppose I have (love ?x ?x) -- everybody loves himself

17:09:59 <DrewMcD> And (not (love (ex-spouse ?y) ?y)) -- Everyone's ex-spouse hates him.

17:10:09 <DrewMcD> Pretend that ex-spouse is a function, which it really isn't.

17:10:29 <DrewMcD> If the two "love" expressions unify, we have a contradiction.

17:10:42 <DrewMcD> They unify if x = y = (ex-spouse y)

17:11:05 <DrewMcD> But there is no term y such that y = (ex-spouse y). (This is equality of terms, not equality of their denotations.)

17:11:32 <DrewMcD> ... Possible exception: If you allow infinite terms, then there *is* a y = (ex-spouse y)

17:11:46 <DrewMcD> ... namely: (ex-spouse (ex-spouse (ex-spouse ....)))

17:12:03 <DrewMcD> But infinite terms are unpleasant algorithmically, not to mention useless.

17:12:37 <DrewMcD> With the occur check turned off, Prolog will cheerfully generate the infinite term, which will cause lossage the next time you do anything with it.

17:12:48 <DrewMcD> end of example

17:14:40 <DanC> nifty.

17:15:05 <DrewMcD> It occurs to me that since N3 has no terms, none of this will come up.

17:29:33 <ArtB> danbri - do you have any direct experience with the VRP RDF parser <http://139.91.183.30:9090/RDF/VRP/index.html>?

17:30:12 <danbri> Yes and no... hang on...

17:30:52 * ArtB is in no hurry ...

17:31:02 <danbri> me neither, just looking for a URI

17:33:48 <danbri> Odd... I made an adaptor a year or so ago, that plugged VRP in behind the old SiRPAC API that Janne and I invented. Will be obsolete now... Just as well as I can't find the files!

17:36:18 <danbri> http://ilrt.org/discovery/rdf-dev/rudolf/fax/FAXds.java

17:36:18 <dc_rdfig> H: http://ilrt.org/discovery/rdf-dev/rudolf/fax/FAXds.java from danbri

17:36:32 <ArtB> I haven't run it but was impressed by the (relatively) large number of error messages it produces http://139.91.183.30:9090/RDF/VRP/Errormsgs.html - that would be a significant improvement over SiRPAC ...

17:36:51 <danbri> H:|A wrapper for the VRP RDF parser to make it a SiRPAC datasource (rough cut, Oct 2000).

17:36:52 <dc_rdfig> titled item H

17:37:39 <danbri> H:VRP encapsulated behind a minor variant of the SiRPAC v1.14 API, which may well have changed subsequently.

17:37:40 <dc_rdfig> commented item H

17:38:57 <danbri> Yes, it does seem to be pretty well done. My only worry is that it (I believe) does its own XML processing, which may be a maintainability concern.

17:39:04 * danbri waves at bijan

17:39:19 <ArtB> Yuck!

17:40:06 <bijan> Hey dan!

17:40:58 <bijan> What's up?

17:49:45 <danbri> what's up?: playing with chump bot again, wondering about ways of extending it!

17:49:59 <danbri> how've you been? you kinda vanished from here for a while...

17:51:44 <bijan> Yep.

17:51:52 <bijan> Eh...lots of distractions.

17:55:00 <bijan> Right now, I'm enjoying the very mild flack following the current XML.com/

17:57:56 <libby> heh, nice intro

17:58:13 <bijan> Hey libby! How's it going?

17:58:27 <libby> not bad thanks?

17:58:33 <libby> what av=botu you?

17:58:46 * libby 's typing aint getting any better

17:58:48 <bijan> Ok. Can complain but don't feel like it ;)

17:58:55 <libby> heh

17:59:05 <danbri> hi! got to go to a meeting, good to see you though :)

18:04:02 <GlobalMessage> Attention Linux Junkies! David Diamond, co-author of the Linus Torvalds biography "Just for Fun", is chatting with OpenProjects users in #live. This event is sponsored by Linux.Com. /join #live to participate and enter our drawing for a free copy of the book!

18:08:51 <libby> see you bijan. interesting article

18:08:59 <bijan> Bye!

18:09:02 <DanC_> er.. I wanna reverse a list in python. isn't there a library function for it?

18:10:04 <bijan> Kendall says: "lst.reverse()"

18:10:53 <DanC_> doesn't seem to work. seems to fail silently.

18:11:02 <bijan> He also says that's an "in place" reversal.

18:11:23 <DanC_> ah. thx.

18:11:39 * DanC_ loves IRC

18:11:43 * GabeW loves Python's simplicity

18:12:02 <DanC_> I just wonder why I couldn't find list.reverse() in the docs.

18:12:38 <bijan> Where'd you look?

18:12:54 <bijan> It's in 2.1.5.2 of the library docs (py1.5.2)

18:13:00 <bijan> "Mutable sequence types"

18:13:01 <DanC_> I looked in http://www.python.org/doc/current/lib/built-in-funcs.html#built-in-funcs

18:13:11 <DanC_> and in http://www.python.org/doc/current/lib/typesseq.html

18:13:16 <bijan> It's not a function, it's a method.

18:13:16 <DanC_> ah... I didn't look under mutable.

18:13:44 <bijan> Plus, you weren't looking for a destructive one (which would only hold for mutable types).

18:14:12 <DanC_> sure would be nice to have a note in typesseq ala "if you want to reverse a tuple, see ..."

18:14:26 * bijan finds the organization of the python docs a frustrator.

18:14:33 * GabeW too

18:14:51 <DanC_> I'm mostly happy with the docs; they're quite well-written, as a rule.

18:15:08 <bijan> When they're good, they're very good.

18:15:23 <bijan> But their organization still sucks,e ven where they're good.

18:15:49 <bijan> It's just hard to find stuff, sometimes.

18:16:09 <bijan> And I tend to get in a loop where I forget that some stuff is *only* in the tutorial.

18:16:26 * bijan tends to ignore things called "tutorial" when looking stuff up.

18:16:49 * GabeW keeps a python interpreter open almost all the time to "try things out"

18:17:54 <bijan> Yep.

18:23:17 * GabeW is a little tweaked by that xml.com article..

18:24:28 <bijan> Hmm?

18:25:05 <GabeW> The "3 Myths" one...

18:25:25 <bijan> Yes, I figured :)

18:25:44 <bijan> It was an encouragin, "Yes, go one" sorta "hmm".

18:25:50 <GabeW> ah

18:26:05 <bijan> Alas, if only I'd had HumanML we wouldn't have misunderstoodiated each other! :)

18:26:05 <GabeW> Yeah, well, as one of the LegalXML founders, I obviously found it interesting.

18:26:17 <GabeW> And I think Kendall is actually technically correct about his points.

18:26:37 <GabeW> Its just that I don't think the majority of people *in* legalxml believe in the myths kendall talks about

18:27:38 <bijan> A not atypical reaction. I.e., on xml-dev someone basically said, "Yes, but *none of us* fall for those myths."

18:27:53 <bijan> I'm not sure it helps.

18:28:05 <GabeW> not sure *what* helps?

18:28:09 <bijan> Or rather, I'm not sure what the response is supposed to accomplish.

18:28:42 <GabeW> yeah, I think really, its the tone of the article thats annoying more than anything....

18:29:14 <bijan> Hmm?

18:29:26 <GabeW> and he's actually wrong about the effects that the fed gov using XML could have - there are substantial cost savings

18:29:35 * bijan isn't sure how *that* response "helps".

18:29:47 <GabeW> AND he's wrong about access to legal information - its really about lowering the cost of providing legal services

18:30:01 <GabeW> but the gist of what he's saying is right - XML isn't going to change the human experience

18:30:03 * bijan telling kendall to come on over.

18:30:08 <GabeW> hehe

18:30:26 <bijan> I don't see that as the gist.

18:30:35 <GabeW> OK, maybe I read it too fast.

18:31:11 <bijan> And it seems that both your "rebuttals" were not of things he said.

18:31:14 * GabeW wonders if this is the right forum to have this conversation..

18:32:50 * bijan notes that many of the points apply *mutatis muntandis* to RDF and the SW.

18:33:45 <GabeW> hey, he's here.

18:33:47 * kendall 's super intelligent Semantic Web irc robot notified me that I was being slandered here...

18:33:55 <kendall> hi Gabe, long time.

18:34:04 <GabeW> yeah, wondered if you would remember.

18:34:07 <kendall> of course

18:34:25 <kendall> i was just saying that Findlaw is still my favorite legal resource, and so it is. :>

18:34:38 <GabeW> Findlaw got bought by West Publishing, did you here?

18:34:43 <GabeW> s/here/hear/

18:34:43 <kendall> nope

18:34:46 <kendall> congrats

18:35:04 <GabeW> I haven't worked for findlaw for over 2 years and I didn't get much out of that deal

18:35:10 <kendall> sucks

18:35:12 <GabeW> Findlaw wasn't doing so hot

18:35:21 <GabeW> I mean, they were basically an ad-based company

18:35:42 <GabeW> Anyway, the thing that bugged me about that article is that I know you know the area you are talking about.

18:35:44 <kendall> yeah, well, utility and profitability are often, hmm, orthogonal. :>

18:36:04 <kendall> yeah, I do.

18:36:26 <GabeW> But it still bugged me because, well, I'm well aware of those "myths" and I think most of the people I deal with within LegalXML do as well.

18:36:37 <GabeW> Whether or not that "refutes" anything you said is another issue...

18:36:55 <kendall> a couple of folks have suggested that there are costs savings associated with XML that I don't "get" but that's really irrelevant to the point I'm making. I never said there was *no point* to using XML, only that it couldn't do impossible things.

18:37:32 <kendall> the head person of LegalXML that was quoted in Kotok's piece said things that struck me as very much buying into some of the myths

18:37:37 <GabeW> XML certainly (as you point out) is nothing particularily special except for the fact that its grease in the machinery to get people talking about technology

18:37:43 <kendall> i obviously can't speak to what *everyone* in LegalXML effort things or believes

18:37:48 <GabeW> I make no representations about the people currently on the borad of LegalXML ;-)

18:38:37 <kendall> yes, again, i'm only using humanml, legalxml as examples of the myths

18:38:56 <kendall> i don't mean thereby to denigrate the entire project (esp. not hte LegalXML which has a few interesting things going for it)

18:39:37 <kendall> if i hadn't included any examples, people would have said, "oh that's fine but no one really believes all that". :>

18:39:40 <GabeW> right, the thing is, LegalXML is just a stick to beat the legal/legislative world with to get them to do things in a more accessible way.. I'm just as excited about sunshine laws and all that, but I can't do as much there.

18:40:00 <kendall> yes, sticks to beat the powerful are good.

18:40:28 <kendall> deluding oneself that one's stick is actually a 12 gauge shotgun: not so good :>

18:40:31 <GabeW> I'm actually surprised by the OMB Watch guy saying what he said

18:40:39 <kendall> (woman, actually)

18:40:53 <GabeW> guy is my gender-neutral term for person - probably not very PC

18:41:02 <kendall> i've been dealing with FOIA requests to FBI for the past 18 months, and I found the FOIA points just inconceivably dumb

18:41:40 <GabeW> i feel sorry for you!

18:41:52 <kendall> anyway, i don't dispute that using XML might save the fed gov't some $$. That's pretty obvious, and nothing I deny.

18:41:53 <GabeW> OK, kendall, so I am over the emotional response to your article...

18:42:22 <kendall> and there are always some folks on the margins for whom electronic versions of laws and case filings will be helpful, but for *most* folks, it won't be relevant at all.

18:42:29 <kendall> GabeW: great :>

18:43:51 <GabeW> i've talked to a lot of lawyers (esp. working in government) who use findlaw rather regularly - I guess I'm not so into the XML part of things, but rather the fact that you don't have to pay certain cmpanies $3 a minute or whatever

18:43:57 <kendall> as you well know, having the law degree and all, access to law, court docs, legislative arcana is really the least of somoene's worries. understanding all that stuff is a skill that one acquires independently of the issue of mere *access*

18:44:31 <GabeW> My wife saved the City of Oakland hundreds of dollars a month herself when she would use findlaw instead of west

18:44:56 <GabeW> I'm not really expecting a guy off the street to use findlaw ;-)

18:45:35 <kendall> it got cut from the XML.com version (the fuller version is on Monkeyfist.com), but I complain that what citizens could really use is Naderesque democracy toolkits built *around* or using XML versions of the Federal Register. Now that would be useful.

18:45:44 <GabeW> The other big area is electronic filing (court filing) which advocates (I have no information here) hold up as a key to streamlining the courts

18:45:52 <GabeW> ooh - thats interesting sounding..

18:45:56 <kendall> I tried to do that last year, but the SGML versions are outsourced to some company that charged really high rates for FTPing them.

18:46:04 <GabeW> hehe

18:46:21 <GabeW> I used to deal with some company who sent out Supreme Court cases via **UUCP**

18:46:24 <kendall> again, a larger political issue, privatization of national information systems, that XML just cannot touch.

18:46:58 <kendall> I'm a bit skeptical that e-filing will unclog dockets.

18:47:07 <kendall> it is an interesting IT project (and huge!)

18:47:18 <GabeW> yeah, I've thought about the opportunity.

18:47:27 <kendall> of all places, I think New Mexico is the most advanced state court

18:47:41 <GabeW> yeah, thats for sure. although small trials are being held all over the place.

18:47:54 <kendall> yes, but NM has got nifty stuff in production.

18:48:15 <kendall> I tried to buy court records weekly from Texas state courts last year and they will only sell *mainframe* dumps on 9 track! :>

18:48:28 <GabeW> hehe - pdf's are all the rage these days

18:49:00 <kendall> i thought a bit about trying to contribute to some of the LegalXML working groups, but kind of lost the urge after a bit :>

18:49:39 <GabeW> its gone up and down and up and down and up and down

18:50:08 <GabeW> I think some of the working groups are starting to get traction - esp. the court filing and (it sounds) maybe the legislation groups

18:50:11 <kendall> i don't remember which ones, but 1 or 2 of the WGs seemed interesting

18:50:54 <GabeW> There seems to be a lot of interest these days in CMS (case management systems) as well - one of the first places ISV can make money I guess

18:50:58 <kendall> ISOGEN/DataChannel (office here in Dallas) is doing some big SGML integration thing with the Texas State Legislature, but I don't think it has public Web access even planned

18:51:33 <kendall> I suspect some of the EU efforts will pay off biggest, first, given that they have underlying political structures in place.

18:51:50 <GabeW> Oh yeah, the LexML effort has got a lot more traction recently

18:51:57 <GabeW> they are "affiliated" with LegalXMl

18:52:30 <GabeW> As a matter of fact, we (LegalXML) and LexML are working together/in parallel on a dictionary (and probably based on RDF)

18:52:52 <kendall> that sounds interesting

18:53:04 <GabeW> well, if you wanted to get involved..

18:53:43 <kendall> it's mailing list based? I have no institutional backing, so travel is out (unless serendipitious, of course, w/ other stuff I'm doing)

18:54:05 <GabeW> I've never travelled to a meeting - I've always been an individual contributor

18:54:17 <kendall> neat

18:54:28 <kendall> how would i go about getting involved?

18:54:46 <kendall> i saw email contacts on legalxml site, but which WG is this dictionary effort in?

18:54:50 <GabeW> Um, send me your email address - they are forming a working group now (the LegalXML folks)

18:55:06 <kendall> kendall@monkeyfist.com

18:55:14 <kendall> (and thanks)

18:55:16 <GabeW> Its actually supposed to be a "nontechnical" group.. but there will be some technical issues that need to get discussed somewhere.

18:55:45 <kendall> sure

18:55:46 <GabeW> Actually, you probably have to sign up on legalxml.org as a participant... but I will forward your email account to the person heading the wg effort

18:56:04 <kendall> ok, i'll go off and sign up today

18:56:47 <kendall> maybe i should use a pseudonym, given my criticisms? :>

18:57:28 <GabeW> Too late ;-)

18:58:11 <GabeW> No, really, participation is valued more than words ..

18:58:25 <kendall> i did find the "unified rap sheet" thing a bit creepola...

18:58:42 <kendall> though i'm not very knowledgable about the state of things in that area

18:58:55 <GabeW> LegalXML has a lot of different points of view and interests represented

19:02:35 <kendall> looking again at the WGs, it's a very ambitious project with lots of parts :>

19:02:45 <GabeW> kendall - I sent off an email requesting you to be added to the WG - and gave the wg owner the URL to the article

19:03:00 <kendall> GabeW: thanks much

19:03:46 <GabeW> kendall: thank you

19:04:02 <kendall> good talking to you again, Gabe.

19:04:08 <kendall> you still hacking Python and BXXP stuffs?

19:04:15 <GabeW> Yes, in fact, I am

19:04:34 <GabeW> I've gotten less ambitious - I'm just porting the IW java beepcore stuff "directly" to python

19:05:10 <GabeW> I've got my own consulting/outsourcing/other company now (with a partner)

19:05:13 <kendall> i'm gonna work on a SOAP frontend for the Monkeyfist CMS this weekend, I think.

19:05:22 <kendall> yeah, I think I saw that at some point.

19:05:33 <GabeW> its BEEP now, btw

19:05:41 <kendall> actzero's Python SOAP thing looks yummy

19:05:50 <GabeW> is that SOAP.py ?

19:05:50 <kendall> so BXXP->BEEP?

19:05:54 <GabeW> yeah

19:05:55 <kendall> GabeW: yes

19:05:59 <GabeW> RFC 3080 and RFC 3081

19:06:17 <GabeW> and mrose just put out the SOAP over BEEP draft

19:06:27 <kendall> interesting

19:06:48 <GabeW> yeah, there's a lot going on -- the most interesting is APEX

19:07:06 <kendall> what's that?

19:07:17 <GabeW> Ha! So you heard it here first

19:07:53 <GabeW> Its a message-passing infrastructure based on BEEP (messages relayed between any two endpoints)

19:08:14 <GabeW> I can't do it justice.

19:09:04 <kendall> i've been looking for good Python message passing stuff, but haven't found anything

19:09:37 <kendall> i wanna experiment with collaborative editing and content creation (for the IndyMedia sites) using a CMS built up from publish/subscribe

19:11:10 <kendall> got a URL for APEX?

19:11:17 <GabeW> http://www.ietf.org/html.charters/apex-charter.html

19:11:18 <dc_rdfig> I: http://www.ietf.org/html.charters/apex-charter.html from GabeW

19:11:35 <kendall> wow, rdfig has a Monkeyfist irc bot! :>

19:11:36 <kendall> neat

19:11:41 <GabeW> I: APEX - Application Exchange protocol built on BEEP

19:11:42 <dc_rdfig> commented item I

19:13:28 <GabeW> you want to look at the core doc first

19:13:42 <GabeW> its co-written by Graham Klyne, btw

19:14:01 <edd> kendall: ahem, you knew this place had a chump :>

19:14:34 <kendall> hey, i was trolling for compliments for Bij, from whom all chumpeth/churneth things flow! :>

19:14:54 <kendall> who's G. Klyne? Should I know that name?

19:14:58 <bijan> I was thinking of a new spinon, the daily churl.

19:15:03 <edd> They would be well-deserved, and indeed I have paid tribute to him here before.

19:15:16 <kendall> "Today's most churlish comment or person"? :<>

19:15:25 <GabeW> kendall: seen him on the rdf/sw email lists

19:15:28 <edd> kendall: Graham Klyne. You may know his name. He's an RDF wizard.

19:15:47 * kendall steps lightly around the subject of who knows whom's name...

19:16:12 <GabeW> Graham mentioned to me that he was thinking about using APEX for RDF/SW related stuff

19:17:36 <kendall> well, i hope i didn't bruise too many legalxml egos (irreparably, anyway)

19:17:44 <kendall> thanks for chatting w/ me about it, Gabe

19:19:02 <kendall> i'm off to read about APEX

19:19:04 <GabeW> its all good - I'm sure the article will be discussed and I'll forward you anything interesting people say (unless you want to subscribe yourself to some of the legalxml lists)

19:19:23 <kendall> hmm, if you see anything interesting, forward a url :>

19:19:30 <kendall> bye

19:56:14 * AaronSw returns

19:56:23 <AaronSw> man, looks like i missed a good chat... oh well

20:03:41 <GabeW> there's always an opportunity to have a better one tomorrow..

20:34:08 <AaronSw> hi Sean

20:34:29 <sbp> Hi Aaron

20:56:46 <danbri> http://www.thisisbristol.com/local_news_server/frames/newsframe_1.html

20:56:46 <dc_rdfig> J: http://www.thisisbristol.com/local_news_server/frames/newsframe_1.html from danbri

20:57:06 <sbp> gotta run...

20:57:14 <danbri> J:|e-carshare project in bristol (ILRT :)

20:57:15 <dc_rdfig> titled item J

20:57:27 <danbri> J:...another semantic web use case!

20:57:27 <dc_rdfig> commented item J

20:58:06 <jhendler> cool - be nice if the article had mentioned this...

20:58:10 <danbri> [[Martin Belcher, project manager for E-carshare at Bristol University’s Institute for Learning and Research Technology, said: “This route matching, combined with an easy-to-use web site, means that users of the system have a significantly increased chance of finding potential carshare matches for their journey to and from work.

20:58:11 <danbri> ]]

20:58:46 * sandro wishes he knew which language to write in (among C, C++, Java, and Python).

20:59:33 <AaronSw> J:Sounds suspiciously like a [barnraising usecase|http://groups.yahoo.com/group/barnraising/message/48]

20:59:34 <dc_rdfig> commented item J

21:00:56 <danbri> perl

21:01:03 <danbri> or mooscript

21:01:36 <AaronSw> python!

21:01:39 <GabeW> python

21:03:26 <sandro> I'm trying to write code like libwww -- code that can be used in the greatest number of applications, with the minimal number of people being kept away by platform constraints. so perl and mooscript are right out, and I don't think pythons really something people want to embed in their applications quite yet, though maybe that's just do to ignorance.

21:03:48 <sandro> s/do/due/

21:05:20 <AaronSw> well then it sure seems like C is the answer

21:06:17 <sandro> The question is whether C++ or Java has come far enough yet. Sometimes C gets soooooo ugly.

21:08:35 <sandro> (especially when I have a glaringly obvious class hierarchy....)

21:08:48 <AaronSw> what are you building? libsemweb?

21:09:21 <GabeW> python was built for embedding, btw...

21:09:55 <sandro> No, this generic parser thing -- to read any format/language as if RDF triples, given the annotated EBNF for the format/language.

21:10:16 <AaronSw> oh, cool!

21:10:33 <sandro> I hear you, GabeW.

21:10:35 * AaronSw wonders how well the compiled python->C stuff works...

21:11:24 <sandro> I guess Python and Java are not really on my table for this project right now. It's in C now, and I keep wanting C++ features.

21:11:33 <GabeW> oh, well, shoulda said so ;-)

21:12:04 <sandro> Sorry -- in the longer term it, I do wonder about it a lot. It is just a nice language......

21:12:34 <bijan> sandro: Try MostlyC!

21:12:38 <GabeW> Well, it really depends what your requirements are ..

21:13:05 <sandro> bijan, "MostlyC" ?

21:14:04 <bijan> Sandro: http://casbah.org/~kmacleod/orchard/#Mostly-C

21:14:12 <bijan> Part of Orchard

21:15:03 <bijan> Indeed, you should talk with ken, it sounds like what you're doing is similar in spirit to what he's doing with Orchard.

21:15:16 <bijan> And that there may even be code overlap possibilities.

21:16:03 <sandro> I see the similaries on the surface at least, bijan.

21:17:41 <DanC_> hi... I wonder if I could get some chump tech support... we've got this copy of chump http://dev.w3.org/cvsweb/2001/sw/chump/

21:17:53 <DanC_> and I'm trying to make it go; I'm getting os.rename(name,self._filename())

21:17:53 <DanC_> OSError: [Errno 2] No such file or directory

21:18:11 <DanC_> my command-line is python dailychumpbot.py -s irc.w3.org -p 6665 -n TestBot -c swcg -d `/bin/pwd`store/

21:19:30 <bijan> Oh yeah.

21:19:42 <bijan> That's a bug.

21:19:59 <bijan> Whose correction did not, I think , make it into the distro.

21:20:05 <bijan> Lemme ask kendall, he knows the fix.

21:22:15 <bijan> Dan: Kendall's coming over. i think he has a fixed version.

21:22:18 <bijan> THere here is.

21:22:27 * DanC_ can't wait.

21:22:54 <kendall> eh, i can't remember all the stuff i had to change

21:23:04 <kendall> and i hardcoded most of it because i was in a hurry

21:23:11 <bijan> os.rename is bogus, as I recall.

21:23:18 <kendall> but if you want a tarball of my version, which works on Linux, I'll email it to you.

21:23:25 <bijan> It fails on certain sets of paths.

21:23:30 <kendall> yes, part of it was os.rename() which isn't portable at all

21:23:46 <DanC_> why is it trying to rename something?

21:23:57 <bijan> It makes a temp file, then copies it over.

21:24:03 <kendall> iirc, it's a temp file thingie

21:24:24 <kendall> caching something or other

21:24:44 <kendall> DanC_: you want my tarball? you could do diffs of the 2 or 3 main files and get an idea

21:24:59 <bijan> When you chump something, url or comment,it writes the file to a temp file. After written it copies over the served file.

21:25:05 <DanC_> tarball: ok

21:25:09 <kendall> (though my tarball includes Bijan's semantic chump hackeries too)

21:25:28 <DanC_> ah... it's trying to rename from /tmp to /home . no workie.

21:25:43 <bijan> Yes.

21:25:53 <kendall> 'k gimme a few minutes

21:26:03 <GabeW> hey, why not use module tempfile?

21:26:18 <DanC_> I could just replace name = tempfile.mktemp() with name='./look-out-for-conflicts'

21:26:27 <jhendler> http://www.byte.com/column/BYT20010425S0016

21:26:27 <dc_rdfig> K: http://www.byte.com/column/BYT20010425S0016 from jhendler

21:26:43 <GabeW> oops

21:26:56 <jhendler> K: End of long column in Byte mentions Sem web - says someone should create software for the thing

21:26:56 <dc_rdfig> commented item K

21:27:04 <jhendler> K: perhaps someone should do that :->

21:27:05 <dc_rdfig> commented item K

21:27:13 <bijan> Yes, in one version I just deleted and wrote.

21:27:15 <AaronSw> K:|Writing the Semantic Web - Jon Udell

21:27:16 <dc_rdfig> titled item K

21:27:35 <bijan> Basically, if you change that, or use os.copyfile (I think) you'll fix it.

21:28:03 <bijan> Hmm. not os.copyfile...

21:28:12 <kendall> shutil.copy(), iirc

21:28:16 <bijan> Yes.

21:28:19 <bijan> That's it!

21:29:00 <AaronSw> K:last part of ["NNTP, IMAP, And The Semantic Web"|http://www.byte.com/column/threads/BYT20010425S0012]

21:29:00 <dc_rdfig> commented item K

21:29:03 <DanC_> er... I changed the name in stead: name = self._filename() + '.tmp'

21:29:25 <bijan> That should work.

21:29:31 <DanC_> it does.

21:29:36 <kendall> you can get a tarball of my hacked src directory (all you need, iirc) at http://monkeyfist.com/kendall/for-danc.tar.gz

21:29:37 <DanC_> er... I think...

21:29:52 <kendall> for whatever reason, i had to make several changes, and you may not need them all

21:29:53 <DanC_> got it, kendall

21:31:00 <sandro> jhendler, were you insulted by the Byte column? It is a little odd.

21:32:14 <jhendler> not insulted - too reasons - one, better to be mentioned than ignored, other - he's right, all the stuff we had about creating software was deemed to challenging by the editor and largely cut out. Lots of other articles talk mroe about that

21:32:22 <jhendler> s/too/two

21:32:31 <DanC_> ah... there we go... the -c option needs the '#' at the beginning.

21:33:24 <DanC_> ok... so... the chumpbot isn't an HTTP server; it just writes files?

21:33:47 <sandro> Yeah -- that just wasn't the point of the SciAm article. A little annoying to be taken to task for that.

21:34:18 <DanC_> I have in mind something that talks both IRC and HTTP, at the same time; kinda like the way Zope talks HTTP and ftp at the same time.

21:34:23 <AaronSw> but the sciam article is probably all most people know about the semantic web

21:34:39 <AaronSw> if that much

21:34:41 <jhendler> http://www.flutterby.com/archives/2001_Mar/14_Thesemanticwebishogwash.html

21:34:41 <dc_rdfig> L: http://www.flutterby.com/archives/2001_Mar/14_Thesemanticwebishogwash.html from jhendler

21:34:52 <sandro> But if you're going to be publishing an article about it, one might think one should learn a little tiny bit more. :-)

21:35:00 <jhendler> L: of course, not everyone likes the Sem web:

21:35:01 <dc_rdfig> commented item L

21:35:02 <AaronSw> ooh, dan lyke, another friend

21:35:13 <jhendler> L: Dan rants: The "semantic web" is hogwash

21:35:14 <dc_rdfig> commented item L

21:35:27 <jhendler> gee, you know all these guys!

21:35:49 <AaronSw> no, dan's right in that piece

21:35:57 <AaronSw> HTML4 is definitely not a semantic format

21:36:06 <AaronSw> i wonder who claimed that...

21:37:11 <jhendler> true, but actually his criticism of that stuff does apply to a lot of the "real" sw - he's better than the critics I've been facing on panels :->

21:37:14 <kendall> DanC: pretty easy to drop in Python's stdlib HTTP server

21:37:31 <kendall> but chump only speaks irc, yes.

21:37:34 <DanC_> but Python's stdlib HTTP server is a blocking server, no?

21:37:48 <kendall> i don't recall

21:38:13 <kendall> and i don't know what you need or want, so... :>

21:39:08 <kendall> hmm, Webware has AsyncThreadedHTTPServer, which I think is an enrichment of something else; but it's pure Py.

21:39:13 * DanC_ should probably read the SciAm article sometime.

21:39:38 <kendall> heh, i couldn't get past the "revolutionary" teaser :>

21:42:25 <kendall> DanC: you might want to look at http://webware.sourceforge.net/Webware/WebKit/Docs/Source/Files/AsyncThreadedHTTPServer.py.html

21:43:18 <DanC_> Async: well, thanks for the pointer, but I'm pretty happy with medusa's http_server.py

21:43:37 <kendall> nice, i was about to suggest that next, so you're golden. :>

21:43:40 <kendall> ok, i'm off

21:43:55 <DanC_> thanks for the help, kendall.

21:44:44 <jhendler> Was searching for a particular thing Tim wrote on Sem Web - google says

21:44:45 <jhendler> Searched the web for semantic web.

21:44:51 <jhendler> Results 521 - 526 of about 194,000. Search took 1.36 seconds.

21:45:10 <jhendler> 194K? yow!

21:45:46 <sandro> google on "semantic web" (in quotes) only says about 14,200. Still a lot.

21:54:49 <sandro> and on "broken links" it says 552,000. :-)

22:01:25 <lilo> [GlobalNotice] Good evening, all. The news is up on http://openprojects.net/news.shtml along with OPN project philosophy and status information. Please take a look. Highlights: Poll results, Gelhausen record, #geekswithguns, editor still looking for a job. Have a great evening!

22:07:23 <jhendler> Logging turned off

22:10:32 <jhendler> Logging turned on

22:10:32 * logger is logging

22:11:00 <jhendler> oops, never mind - had interupted the transfer w/o realizing

22:39:15 <timbl-away> timbl-away is now known as timbl

22:55:35 <timbl> DanC?

22:56:04 <DanC_> yes?

22:56:18 <timbl> Re your test case http://www.w3.org/2000/10/swap/test/includes/uri-startswith.n3

22:56:34 <DanC_> yes...

22:57:13 <timbl> cwm can't [read: I didn't make it] do anything with { ?x uri [startswith "..." ] } log:implies because ...

22:57:25 <timbl> I don't make it search all resources.

22:57:51 <timbl> It has nowhere to start.

22:58:07 <DanC_> but it shouldn't crash.

22:58:25 <timbl> Well, at the moment hte code is new enough that I wanted to flag when these things happen.

22:58:33 <DanC_> I see ;-)

22:58:37 <timbl> I could just drop it, or give a warning.

22:58:58 <GabeW> http://www.openp2p.com/topics/p2p/security/

22:58:58 <dc_rdfig> M: http://www.openp2p.com/topics/p2p/security/ from GabeW

22:59:08 <GabeW> oops

22:59:21 <GabeW> http://www.cs.caltech.edu/~adam/local/trust.html

22:59:22 <dc_rdfig> N: http://www.cs.caltech.edu/~adam/local/trust.html from GabeW

22:59:45 <GabeW> N: Section 5.2 discussing the need for security metadata for trust-based security

22:59:46 <dc_rdfig> commented item N

22:59:55 <timbl> re uri-startswith Or I could take it at face value and actually scan the store ...

23:00:06 <timbl> Maybe I should. After all, that was what you intended.

23:00:24 <timbl> The error was "I have looked for the easiest bit of this puzzle and it looks too hard".

23:00:59 <timbl> At the moment every rule uses an index, none loop over the whole store. It doesn't scale too well!

23:01:03 <DanC_> if I put more constraints in the antecedent, will it do somthing interesting?

23:01:13 <timbl> Yes.

23:01:28 <DanC_> lemme try that...

23:01:45 <timbl> Try telling it just one thing about doc which will help it find it.

23:02:26 <DanC_> oh... my test case reminds me... remember about the :- syntax? we don't need new syntax for the functionality I want...

23:02:45 <timbl> Oh good. I wantto weed that :- out anywy.

23:02:48 <DanC_> I just want to be able to say: the home page of this group is <http://...>, which has titile "abc"

23:02:53 <timbl> Oh good. I want to weed that :- out anyway.

23:03:10 <DanC_> we can do that with: <group> :homePage [ = <aPage>; :title "abc" ].

23:03:27 <DanC_> I'd like the n3 parser to special case = when it's the first thing inside a [.

23:03:49 <timbl> ALternatively you can use log:equalTo which is a builtin.

23:03:57 <DanC_> how so?

23:04:11 <DanC_> what I want is to avoid the gensym altogether.

23:04:36 <timbl> ummm... when looking for x log:equalTo y, cwm will strip it out and merge x and y.

23:05:22 <DanC_> but that won't help in the case, for example, when I'm just using n3 to write RDF files.

23:05:29 <timbl> How about <group> :homepage <aPage> which :title "abc". ?

23:05:53 <DanC_> why new syntax?

23:05:59 <timbl> why [ =

23:06:13 <DanC_> because it's already there, and it means what I mean.

23:06:15 <timbl> cwm does not understand =

23:06:25 <DanC_> notation3.py understands =

23:06:29 <timbl> except as a shorthand for daml:equivalentTo

23:06:38 <timbl> .

23:07:41 <DanC_> in RDF, I can write: <rdf:Description rdf:about="group"><homePage><rdf:Description rdf:about="page"><title>abc</title></></></>. But in N3, I have to write the identifier of page twice. I just want to get rid of that.

23:08:27 <DanC_> I often write stuff like: <> :transcribedFrom <mid:long-tedious-identifer>. <mid:long-tedious...> email:date "..."; email:subject "...".

23:08:38 <timbl> Fair enough.

23:08:57 <timbl> That was why I wondered about "which" as I have been in the same state myself.

23:09:29 <DanC_> so now I write <> :transcribedFrom [ <mid:...> email:date "..."; email:subject "..."].

23:09:35 <DanC_> in anticipation of this optimization.

23:09:45 <DanC_> oops... [= <mid: ...

23:09:51 <DanC_> or [ = <mid:...

23:10:25 * DanC_ thinks it might be easier to implement that to explain/justify ;-)

23:10:33 <DanC_> back to uri-startswith... lemme fidget some more...

23:10:41 <timbl> My problem us *not* with the shortcut. It is with using "=".

23:12:00 <timbl> You imply that :a p [= :a2; :q :c] is equivalent to :a :p :a2. :a2 :q :c .

23:12:22 <DanC_> which it is, no?

23:12:33 <timbl> Which implies that one should do that with "=". Now what do I do if I see "a = b,c" ?

23:13:16 <DanC_> pick one... say, the first. b.

23:14:15 <timbl> It parses to { :a :p _g1. _g1 = :a2 . _g1 :q :c} log:forSome _g1.

23:14:16 <DanC_> i.e. after recognizing a [, you look for =; if you find it, you parse the list of objects that follows, and you use the first one as the subject of the stuff after the ;

23:14:47 <timbl> I know what hack you want. I dispute that you can claim it means the same as using "=".

23:15:07 <DanC_> ??? the stuff you say it parses to means exactly the same thing.

23:15:13 <timbl> I can do it iff I smush on "=".

23:15:33 <DanC_> do what? I'm saying: don't introduce _g1 in the first place, in this case.

23:15:41 <timbl> The stuff I say it parses to means the same thing iff you build axioms about "=" into cwm.

23:16:16 <DanC_> you don't need to build axioms about = into cwm; you just need to avoid a gensym in notation3.py

23:16:18 <timbl> ("it" meant convert { :a :p _g1. _g1 = :a2 . _g1 :q :c} log:forSome _g1. to :a :p :a2. :a2 :q :c . )

23:16:52 <timbl> You ask me to read a = and not generate a = which I can only do if I have an axiom which allows me to get rid of the =.

23:17:00 <DanC_> in general, cwm isn't capable of simplifying logical formulas. that's life. I'm just saying: optimize away one complexity at parse time.

23:17:35 <DanC_> why do you need an axiom in order to *not* introduce _g1 in the first place?

23:17:43 * DanC_ has to go 17 minutes ago

23:18:08 <timbl> I would prefer to keep :- as a parser hack and = as DAML property.

23:18:22 <timbl> You ask me to read a = and not generate a = which I can only do if I have an axiom which allows me to get rid of the =.

23:18:50 <DanC_> keep isn't the right word; :- isn't there now -- not in this sense, anyway. I never checked it in.

23:19:08 <DanC_> You don't need an axiom to get rid of the = BECAUSE IT WAS NEVER THERE.

23:20:02 <DanC_> it's quite clear that you can avoid generating _g1 *without touching cwm.py at all*, no?

23:20:45 <timbl> BUT IT WAS IN THE SOURCE.

23:20:58 <timbl> An = was parsed.

23:20:59 <DanC_> yes, but we can decide what the source means.

23:21:15 <DanC_> i.e. we can decide that [ = x] generates no triples.

23:21:20 * timbl had to go 10 minutes ago

23:21:43 <bijan> For two guys who aren't here, your debate is quite vigorous :)

23:21:50 <DanC_> btw... adding more constraints to the premise didn't stop uri-startswith.n3 from making cwm.py crash.

23:22:01 <DanC_> now I get:

23:22:04 <DanC_> File "../../cwm.py", line 376, in intern

23:22:04 <DanC_> uriref2 = LITERAL_URI_prefix + uriref # @@@ encoding at least hashes?

23:22:04 <DanC_> TypeError: cannot add type "instance method" to string

23:22:23 <timbl> And I have to be careful if I WANT to say { :a :p _g1. _g1 = :a2 . _g1 :q :c} log:forSome _g1.

23:23:05 <timbl> typeerror - oops!

23:23:15 <timbl> alpha code ...

23:23:21 <timbl> later :-/

23:24:09 <DanC_> ciao

23:30:20 * bwm wonders whether he is the only one who can't access the w3 site

23:36:05 <bijan> Seems not to be responding for me either.

23:36:24 <bwm> k - thanks