Semantic Web Interest Group IRC Chat Logs for 2003-09-09

00:17:38 <danja> I believe sbp did Wiki in n3 - anyone else played with Wikis 'n' RDF?

00:20:36 <GabeW> redfoot... ?

00:21:13 <GabeW> wait, no thats not wiki

00:21:38 * GabeW was thinking of wrong social software modality

00:21:59 <GabeW> redfoot blog..

00:24:03 <danja> modality is alwaYS a worry

00:24:54 * sandro has thought a lot about Wikis and RDF, but not come to any conclusions. Some thoughts at http://esw.w3.org/topic/SemanticWiki ; feel free to update!

01:08:34 <ronwalf> danja: Bijan has been playing with MoinMoin and rdflib

01:17:07 <danja> sorry, miles away

01:17:34 <danja> ronwalf - thanks

02:15:21 <JimH>http://www.sis.pitt.edu/~dlwkshop/paper_miller.html

02:15:22 <dc_rdfig> A: http://www.sis.pitt.edu/~dlwkshop/paper_miller.html from JimH

02:16:07 <JimH> A:| Eric Miller's paper on Sem Web for Research

02:16:07 <dc_rdfig> Titled item A.

02:16:24 <JimH> A: This Miller guy seems to understand this Sem Web Stuff :->

02:16:24 <dc_rdfig> Added comment A1.

02:16:38 <JimH> A: seriously, very interesting paper by Eric M.

02:16:39 <dc_rdfig> Added comment A2.

02:29:54 <MikeM> haven't seen him here for quite a while though

04:15:25 <mdupont-ZZZ> What is the list to report bugs in cwm to?

04:15:31 <mdupont-ZZZ> mdupont-ZZZ is now known as mdupont

06:32:28 <GNUWookie> GNUWookie is now known as mdupont

07:30:16 <sbp> mdupont: if you're still around, my installation works just "fine". where "fine" indicates that I had to fix two lines before I could get it to work, as usual. but the errors that I got were just syntax errors, and easy to fix

07:30:19 <sbp> [[[

07:30:20 <sbp> $ ./cwm.py

07:30:20 <sbp> #Processed by Id: cwm.py,v 1.142 2003/09/05 23:04:40 timbl Exp

07:30:20 <sbp> ]]]

07:30:32 <sbp> I suggest that you get the latest version of cwm.py, and perform the same fixes that I did

07:30:53 <sbp> you can get the .tgz from http://www.w3.org/2000/10/swap/cwm.tgz

07:31:15 <sbp> the errors:-

07:31:24 <sbp> #1) change "nor" to "not" in the bit being raised

07:31:43 <sbp> #2) add another ")" to the end of the last statement being raised

07:32:22 <mdupont> hi sbp

07:32:27 <sbp> hey there

07:32:41 <mdupont> ahh, i did the nor

07:32:46 <sbp> chasing down syntax errors in supposedly stable versions is, unfortunately or not, part of the fun of using CWM :-)

07:32:47 <mdupont> but the ) i did not

07:32:52 <sbp> ah, okay

07:32:55 <mdupont> the bleeding edge

07:33:07 <sbp> hmm. but you shouldn't've got the import error still

07:33:09 * mdupont is running pysco on cwm.. it is profiling!

07:33:17 * mdupont updates cvs

07:33:59 <sbp> there's definitely a "def runNamespace()" in RDFSink.py

07:34:13 <mdupont> ok

07:34:15 <mdupont> let me look

07:34:39 <sbp> the latest is at http://www.w3.org/2000/10/swap/RDFSink.py

07:34:54 <mdupont> mdupont@introspector:~/cwm$ grep Id RDFSink.py

07:34:55 <mdupont> __version__ = "$Id: RDFSink.py,v 1.2 2001/09/26 02:46:48 connolly Exp $"

07:35:09 <sbp> ouch. that's an extremely old version

07:35:40 <mdupont> outch!

07:36:29 <mdupont> shit

07:36:33 <mdupont> i had a tag in there

07:36:44 <mdupont> to try and roll back the one file that was broke

07:37:03 <mdupont> now it taged all the files.

07:37:16 <mdupont> mdupont@introspector:~/cwm$ grep Id RDFSink.py

07:37:16 <mdupont> __version__ = "$Id: RDFSink.py,v 1.27 2003/09/02 00:13:41 timbl Exp $"

07:39:09 <mdupont> sbp: where do i add the ) to?

07:40:02 <sbp> er. hang on a second and I'll find it for you

07:40:49 <deltab> mdupont: I'd call that cvsid, not __version__

07:41:00 <mdupont> i just greped

07:41:03 <mdupont> :_)

07:41:19 <mdupont> i am not calling it version, it calls it version ;)

07:41:23 <deltab> ah

07:41:26 <mdupont> heheh

07:41:38 <mdupont> i think that is some python thingie...

07:41:49 <deltab> timbl: I'd call that cvsid, not __version__

07:41:55 <deltab> there :-)

07:42:04 <mdupont> and good for you deltab?!

07:42:12 <sbp> this, in query.py: [[[

07:42:13 <sbp> if verbosity > 40: progress("%s can affect %s because %s can trigger %s" %

07:42:13 <sbp> (`r1`, `r2`, `s1`, `s2`)

07:42:13 <sbp> break # can affect

07:42:14 <sbp> ]]]

07:42:24 <sbp> should have its second line replaced with:-

07:42:28 <deltab> mdupont: it's not

07:42:30 <sbp> [[[

07:42:33 <sbp> (`r1`, `r2`, `s1`, `s2`))

07:42:34 <sbp> ]]]

07:42:52 <mdupont> -break?

07:42:59 <sbp> no, add the )

07:43:13 <sbp> at the end of the line with the rn/sn things in it

07:43:23 <sbp> sigh. that should be using %r...

07:43:42 <sbp> the CWM code should be put on a Wiki!

07:44:02 <mdupont> yes

07:44:13 <mdupont> i have commente out the break as well.

07:44:16 <mdupont> is that ok?

07:44:21 <sbp> no, restore the break

07:44:24 <mdupont> ok

07:44:36 <sbp> it's only the lack of the ")" that is causing the problem

07:44:55 <sbp> if you add the ")" and run ./cwm.py you should see it running just fine (hopefully!)

07:45:04 <mdupont> great

07:45:06 <mdupont> thanks it work

07:45:07 <sbp> yay

07:45:11 <sbp> you're welcome

07:45:13 <mdupont> should I send this patch in?

07:45:24 <mdupont> sbp: you might be interested in this pysco stuff

07:45:35 <sbp> if you like, though I'll guess they know about it... DanC certainly does

07:45:57 <sbp> oh?

07:45:58 <mdupont> ok

07:46:22 <sbp> hmm. were you the one playing around with using Parrot for CWM?

07:46:28 <mdupont> yes pysco is a python profiler http://psyco.sourceforge.net/

07:46:44 <mdupont> yes I have gotten parrot for a large chunk of cwm

07:47:16 <sbp> that's excellent

07:47:22 <sbp> ooh, it can do it for arbitrary Python?

07:47:31 * sbp might well have to check this out for his own code...

07:48:12 <mdupont> yes

07:48:19 <mdupont> I have gotten in compiling

07:48:25 <mdupont> and have been running it now

07:48:33 <mdupont> it collects a full log of where the time is spent

07:48:38 <mdupont> and profiles your python

07:48:40 <mdupont> like this

07:49:08 <mdupont> 09:34:47.71 ______

07:49:08 <mdupont> #1 | 6.1 %| __getitem__ ...t/cwm1.82/llyn.py:186

07:49:08 <mdupont> #2 | 2.7 %| selectDefaultPrefix ...t/cwm1.82/llyn.py:816

07:49:08 <mdupont> #3 | 1.4 %| dumpNested ...t/cwm1.82/llyn.py:1126

07:49:08 <mdupont> 09:34:48.81 tag function: llyn.StoredStatement.__getitem__ %

07:49:09 <mdupont> 09:34:48.81 ______

07:49:11 <mdupont> #1 | 2.5 %| selectDefaultPrefix ...t/cwm1.82/llyn.py:816

07:49:13 <mdupont> #2 | 1.3 %| dumpNested ...t/cwm1.82/llyn.py:1126

07:49:15 <mdupont> in the last minutes where we are speaking

07:49:22 <mdupont> it is just dumpNested

07:49:39 <mdupont> so I guess that selectDefaultPrefix could be optimized alot

07:50:12 <sbp> hmm. just downloading it now

07:50:22 <mdupont> the compiling is a bit of a pain

07:50:27 <mdupont> but it works

07:50:37 <mdupont> the issue is that it produces only one object file

07:50:47 <mdupont> and it does not have a proper makefile

07:50:55 <mdupont> i can post the patch to cwm to get it to work

07:53:04 * mdupont prepares a mail

07:56:47 <mdupont> ok

07:56:49 <mdupont> sent

07:56:56 <mdupont> just responded to your mail

07:57:00 <mdupont> and tacked on the pathc

07:57:02 <mdupont> patch

07:57:33 <Wack> hmm, psyco is in the freebsd ports

07:58:39 <Wack> I thought psyco was a JIT like python vm?

08:01:04 <mdupont> yes

08:01:09 <mdupont> is it

08:01:15 <mdupont> a jit

08:01:33 <mdupont> but it is also a profiler

08:01:47 <mdupont> this dumpnested needs some speedup

08:02:36 <sbp> tacked on the patch: many thanks

08:02:39 <mdupont> gawd! 20 minutes to dump! what about a binary coredump that would be faster

08:02:53 <mdupont> ok you dont want the psyco patch yet :)

08:03:07 <sbp> yeah. I've been testing...

08:08:45 <mdupont> ok, thanks for the support

08:08:48 <mdupont> nice chatting

08:08:52 <mdupont> gotta run to work

08:08:56 <mdupont> mdupont is now known as mdupont-work

08:10:42 <sbp> you're welcome, and likewise

08:28:38 <arnarl> hi

09:15:42 <tav> tav is now known as tav|offline

11:25:58 <fidothe> morning all

11:26:28 <libby> hey fidothe

11:27:09 <fidothe> I was wondering if there was a clear RDF framework winner in the 'I'm using python, and I'm an RDF newbie' category

11:27:12 <fidothe> hey lib

11:27:30 <fidothe> I know that several exist

11:28:26 * sbp is working on one, but it's not quite at release stage yet... argh

11:29:21 * fidothe is flummoxed by whether to go for Redland/Raptor, 4RDF, or one of the others

11:29:30 <sbp> redfoot is highly regarded

11:29:31 <dajobe> fidothe: go with the ones that have been out for a few years and stable such as rdflib.net

11:30:29 <sbp> (rdflib is the RDF API for redfoot)

11:31:05 <dajobe> since you said framework, that rules mine into scope: http://www.redland.opensource.ac.uk/docs/python.html

11:31:17 <dajobe> but nobody's measuring "winners"

11:39:21 <fidothe> I'm mostly interested in something which is pretty 'pythonic' - something with interfaces which I can get my head round. Reasonable performance would be nice, but initially I need to be able to figure out what I'm doing

11:40:07 <fidothe> I'm a cold starter on the implementation side of things, but I'm fairly gung-ho and prepared to hack at something until it works

11:40:50 <dajobe> we've been bashing on redland's py api recently so it's much more pythonic

11:40:56 <dajobe> the pydoc shows the api http://www.redland.opensource.ac.uk/docs/pydoc/RDF.html

11:41:09 <dajobe> but it's better if you see example code

11:41:19 <dajobe> for statement in model:

11:41:19 <dajobe> print statement

11:41:26 <dajobe> that kind of thing is same as rdflib

11:42:43 <dajobe> here's one mattb did in http://www.hackdiary.com/archives/000036.html see http://www.hackdiary.com/src/bloginfo.py

11:43:03 <dajobe> or more recent http://www.hackdiary.com/src/tv.py

11:44:41 <fidothe> cool

11:44:59 <dajobe> there is also foafbot which is a lot larger

11:45:43 <fidothe> I've heard mattb talk about his picdiary stuff (did that use redland matt?) and like the idea of using something which other people are using on similar projects

11:45:58 <fidothe> kind of community-recommendations meets fitness-for-purpose

11:46:03 <dajobe> picdiary uses redland in places I think

11:47:02 <dajobe> the picdiary key idea is to use RDF (RSS1.0) everywhere. The technology that delivers it might vary

11:49:14 <fidothe> yeah, that's the beauty of actually- or virtually-distributed stuff I suppose...

11:49:32 <fidothe> as long as the joins between bits work, who cares...

11:50:40 <libby> fidothe, not sure if relevant, but did I show yu my image annotation thing, based on elemnts form mattb's stuff:

11:50:45 <libby> ...http://swordfish.rdfweb.org/discovery/2003/06/codepictjs/

11:50:56 <fidothe> yeah - i was thinking of stuff like that

11:51:04 <fidothe> this is all to do with the CLR bits...

11:51:28 <fidothe>http://www.centralletteringrecord.org/

11:51:28 <dc_rdfig> B: http://www.centralletteringrecord.org/ from fidothe

11:51:36 <fidothe> oh shit

11:51:44 <libby> it's ok... :)

11:51:56 <libby> do you mind people seeing it on http://rdfig.xmlhack.com?

11:51:58 <fidothe> how do I make it go away?

11:52:12 <fidothe> libby: no rdfig.xmlhack.com is fine

11:52:18 <libby> you have to change it to something else

11:52:33 <fidothe> might as well title it then...

11:52:40 <libby> yeah

11:52:55 <fidothe> B:|Central Lettering Record project homepage

11:52:55 <dc_rdfig> Titled item B.

11:53:21 <libby> Ah, course, I remember you talking about this

11:53:27 <fidothe> bitsko: the CLR project is an attempt to describe a large collection of images depicting public lettering

11:53:46 <fidothe> er... how do i keep adding comments?

11:53:56 <libby> tru without a space i.e. B:

11:54:07 <libby> xchat does autocomlpetion (which you can turn off)

11:54:07 <fidothe> oh, okay

11:54:24 <fidothe> B:the CLR project is an attempt to describe a large collection of images depicting public lettering

11:54:24 <dc_rdfig> Added comment B1.

11:54:41 <libby> the picures on the RHS are lovely

11:55:29 <fidothe> B:The descriptions include information about the form of the letters, their relationship to archetypal and evolved letterforms, and their place in space (and time)

11:55:30 <dc_rdfig> Added comment B2.

11:55:54 <libby> ooh space and time!

11:56:06 <fidothe> it's complicated...

11:56:33 <libby> yeah

11:57:55 <fidothe> we started from looking at a pure-XML approach (single schema) and are now realising that a single-schema approach is fine for a rendering of information about an instance of lettering, but the relationships are more complex than that and and RDF-style approach fits very well

11:58:31 <fidothe> it's all very graph-like

11:58:39 <libby> it sounds it

11:58:52 <fidothe> anyway, thanks for the advice about toolkits

11:59:10 <libby> does it have to be python?

11:59:32 <fidothe> I like python, and don't fancy going back to perl or learning java

12:00:00 <libby> right....just a thought - there's a ruby toolkit chumped yesterday that looked interesting:http://www.nongnu.org/samizdat/

12:00:23 <libby> i's perhaps more generic than you want, or too open

12:02:42 <fidothe> thanks, i'll have a look

12:03:08 <libby> ruby's a pretty ice language, though young

12:03:13 <libby> nice evn

12:03:16 <libby> nice even

12:03:17 <fidothe> heh

12:03:34 <fidothe> I'm kind of python-fixated at the moment

12:03:56 * libby wonders about chumping questions

12:04:02 <fidothe> python and i get along very well - so I'm kind of loath to look elsewhere

12:04:14 <libby> make sense

12:04:14 <fidothe> ack, I'm being summoned

12:04:17 <eaon|zZz> eaon|zZz is now known as eaon

12:04:36 <darobin> but ruby is *really* nice :)

12:04:37 <fidothe> lunch time - bbl, and thanx everyone

12:04:46 <libby> BLURB:question from fidothe: I was wondering if there was a clear RDF framework winner in the 'I'm using python, and I'm an RDF newbie' category

12:04:46 <dc_rdfig> C: question from fidothe: I was wondering if there was a clear RDF framework winner in the 'I'm using python, and I'm an RDF newbie' category from libby

12:04:49 <libby> bye!

12:05:07 * fidothe is away: I'm busy

12:05:15 <libby> C:"<fidothe> I'm mostly interested in something which is pretty 'pythonic' - something with interfaces which I can get my head round. Reasonable performance would be nice, but initially I need to be able to figure out what I'm doing"

12:05:16 <dc_rdfig> Added comment C1.

12:05:46 <libby> C:[some suggestions from the logs|http://ilrt.org/discovery/chatlogs/rdfig/2003-09-09.html#T11-27-09] more welcome...

12:05:47 <dc_rdfig> Added comment C2.

12:55:13 * fidothe is back (gone 00:02:26)

12:57:46 <darobin> darobin is now known as XRobin

12:59:49 <XRobin> XRobin is now known as darobin

13:13:10 <AaronSw> brooklyn?

13:14:33 * AaronSw tries to figure out how to interpret BRKLN

13:15:43 * danbri guesses so

13:16:02 * danbri changes his nick to danbriNWLNCRRGRTRN

13:31:42 <timbl> timbl is now known as timbl_away

13:32:32 * timbl_away horrified to see syntax error in cwm stuff despite rgression checks built into release. suspects some .pyc cache problems

13:33:09 * timbl_away scrolling back notes he does NOT catch comments to this channel, please send to timbl+n3bugs@w3.org

13:35:11 <reagleBRKLN> Aaron, yes Brooklyn

13:48:26 * sbp notes that today's comments were sent to www-archive+n3bugs and timbl...

13:48:32 <sbp> (along with patches)

13:48:45 <timbl_away> Thanks, sbp.

13:49:05 <sbp> no problem. did you catch it there, or does the +n3bugs to your personal address help?

13:49:24 <timbl_away> before I check something in, I run the regression tests., including running them with debug flag turned on.

13:49:48 <timbl_away> Last time, I found some bugs which turned up a few hours after I had done a release.

13:49:58 <sbp> hmm. if it's a .pyc caching bug, perhaps you can rm -f *.pyc first?

13:50:00 * dajobe thinks that sounds familiar

13:50:20 <timbl_away> Yes, I will try putting that in the test.

13:50:25 <sbp> thanks

13:50:31 <dajobe> I just found a book in Borders on releasing software. It had a section on the philosophy of version numbers

13:50:39 <timbl_away> I have also found that sometimes I make a change to source fiel and it just doesn't recompile.

13:51:20 <timbl_away> Philosophy of version numbers?

13:51:31 <dajobe> yeah, I laughed

13:51:44 <dajobe> customers prefer to see 1.6.0 after 1.5.3

13:52:24 <dajobe> and major versions such as2.0.0 versions better not be 2.0.5

13:52:36 <dajobe> more social than technical, I admit

13:53:53 <timbl_away> psychological

13:54:13 <timbl_away> I looked at how to make cwm 2.0 but couldn't figure out tyhe CVS doc in a few minutes.

13:54:19 <timbl_away> The on=bvious thing failed.

13:54:26 <dajobe> I've not figured it out in years

13:54:34 <timbl_away> cvs commit -r2.0

13:54:44 <dajobe> those numbers are internal stuff anyway

13:55:24 <timbl_away> But when you make a large chnage it would be nice to be able to incrememnet the major version

13:58:40 <sbp> btw, I note that CWM is still using rdf:parseType="Quote" on output (not as a qname, either); perhaps it'd be better to just use Literal and specify a "formula" datatype?

13:59:24 <timbl_away> Well, the spec now specifies that anything else necessarily means "Literal".

13:59:36 <sbp> I remember your comparison of formulae to literals in /swap/Primer, and I wonder that could be a good way of getting that into RDF/XML

13:59:40 <sbp> yeah

13:59:42 <timbl_away> Yes... the formula datatype is interesting.

14:00:43 <sbp> I've been working with an extended version of N-Triples for queries (N-Triples + ?x), and I've shoehorned the formulae into literals in that way. the problem is scoping of the variables

14:01:56 <libby> that's cool sbp, re extended ntriples...

14:02:08 <libby> is this in your new python stuff?

14:02:30 <sbp> yeah. it's good because I can just build it into the standard, and thoroughly tested, N-Triples parser that I have. I've got a "loose" parsing mode

14:02:58 <timbl_away> A forumla in N3 is only an RDF model plus a set of universal variables and a set of existential variables.

14:03:27 <timbl_away> You are rightt, sbp - using "?" isn't enough when you have nesting.

14:03:45 <timbl_away> it doesn't tell you the scope - you have to have something in the scope.

14:04:05 <sbp> I suppose if I just added @forAll and @forSome, it'd work, but it'd be a pretty major extension

14:04:22 <dajobe> ahem

14:04:57 <sbp> something up, dajobe? :-)

14:05:15 <timbl_away> Just do it.

14:05:49 * timbl_away has been playing with the inference engine instead of doing syntax upgrades

14:05:57 <sbp> oh. well, I meant of my code, not N-Triples. any extensions to N-Triples means you've gone beyond the spec so there's no probalem going further

14:06:29 <timbl_away> So long as N-Triples remains a subset of N3, I am happy.

14:07:31 <danbri> how would we know?

14:07:51 <dajobe> it's N3 that changes faster than N-Triples does :)

14:07:54 <timbl_away> compaing bnfs i guess

14:07:56 <danbri> (for fiddly corner cases etc...)

14:08:12 <sbp> there's no BNF to compare, though :-)

14:08:28 * sbp wonders what happened to yapps

14:08:33 <danbri> so the bnfs don't differ.

14:09:02 <danbri> rather, it isn't the case that there are inconsistent ntriple and n3 bnfs.

14:10:23 <sbp> \U needs to be supported. but that appears to be the only major difference (there are, perhaps, some other encoding problems)

14:10:23 <sbp> [[[

14:10:24 <sbp> notation3.BadSyntax: Line 1 of <file:/misc/tools/cwm/>: Bad syntax (bad escape) at ^ in:

14:10:24 <sbp> "<#x> <#y> "blargh^\U00000035" .

14:10:24 <sbp> "

14:10:26 <sbp> ]]]

14:11:09 * sbp might hack on that and send a patch

14:11:21 <dajobe> ^ isn't an escape

14:12:23 <sbp> ^ was added by the BadSyntax message printer; it signifies where the error in the input was. "blargh\U00000035" was the input

14:13:54 <sbp> oh, though \U00000035 isn't valid, if that's what you mean...

14:14:35 <dajobe> no, but you are right

14:15:03 * sbp repeats with U00010000 and gets the same problem

14:16:05 <sbp> hmm. I guess \U00000035 could be valid in N3 since it's a superset of N-Triples

14:16:25 <dajobe> N-Triples invented that

14:17:11 <sbp> yeah. but it apparently stole it from Python, which is what N3's apparently based on, so it's logical

14:17:11 <dajobe> and \U00000035 is not allowed

14:17:14 <timbl_away> NTriplesinvented \U?

14:17:23 <dajobe> well, I did

14:17:31 <timbl_away> :-)

14:18:01 <timbl_away> And \U00000035 i snor allowed because its not canonical?

14:18:04 <dajobe> somebody moaned that I mentioned it was fashioned after python so I removed that reference

14:18:04 <dajobe> yes

14:18:18 <sbp> [ is dc:creator of [ rdfs:label "N-Triples" ] ]

14:18:22 <timbl_away> Canonical is good.

14:18:35 <dajobe> if codepoint < 0x10000 then: sprintf("\u%04X", codepoint) else: sprintf("\U%08X", codepoint)

14:18:40 <sbp> removal of reference: shame

14:18:57 <timbl_away> "NTriple"^rdfs:label.dc:creator

14:19:02 <sbp> heh

14:19:09 <sbp> I thought ! was being used to forward traverse?

14:19:51 <timbl_away> I put in "." due to public request along the lines of "almost all programming langauges use it, don't be werid"

14:20:42 <sbp> hmm. I actually rather liked "!" since a) "." is already being used for something else with whitespace, and b) it harkens back to the ol' bang path email days

14:21:09 <sbp> allowing both may be more confusing, though. are you going to deprecate "!"?

14:21:26 <timbl_away> It is deprocated.

14:21:37 <sbp> okay

14:21:54 * fidothe is away: I'm busy

14:22:37 <sbp> oh, and whilst I'm thinking of things, a small proposal: $x for blank variables, to be used in conjunction with @forAll and @forSome. I'm hoping that using URIs and having them change into variables with the forAll and forSome syntax will be deprecated at some point way in the future

14:23:28 <sbp> so it'd be nice if instead of { @forAll <#x>, <#y>, <#z> . } one could do { @forAll $x, $y, $z . } At least then people would know out-of-context that $x is a variable, whereas with URIs there could be confusion...

14:23:59 <sbp> I just don't much like the thought of URIs morphing into variables due to a syntax trick :-) though I've found little practical problem with it, actually

14:24:33 <timbl_away> I was thinking of enforcing it in the code.

14:24:57 <timbl_away> Here's why I like qnmes for variables. There are times when you wnat to make sure that your variable names are not used elsewhere.

14:27:37 <sbp> hmm

14:27:52 <sbp> but if you're cutting and pasting from a file, then you don't know what the prefixes are mapped to anyway

14:28:04 <sbp> and if you're merging files, then you should have variable-local-name renaming

14:28:12 <sbp> (assuming you're using an RDF toolkit to merge)

14:29:58 <timbl_away> Also, it extends RDF less.

14:30:12 <timbl_away> Still, nodes can be URI,s its just you can qualify them.

14:30:24 <timbl_away> $x is interesting.

14:30:39 <timbl_away> Of course there is _:x

14:31:10 <timbl_away> which I assume one assumes has a scope of the local formula.

14:31:16 <sbp> though _:x is always scoped to... right

14:31:34 <sbp> hmm. is there any problem with "@forAll ?x", actually?

14:32:01 <sbp> then ?x would be defaulted to have its parent formula as the scope, unless there were an @forAll declaration to the contrary

14:32:09 <sbp> I guess that's obvious, actually. hadn't thought about it

14:33:46 <timbl_away> Hmmm. "unless"/

14:33:49 <timbl_away> ?

14:34:48 <sbp> so { ?x ... } would be different from { @forAll ?x . ?x ... } since the variable in the former is scoped to its parent, and the variable in the latter is scoped to the current formula

14:34:50 <timbl_away> Maybe. Right now, there is no unless

14:35:48 * twinkle thinks it's bad to have subtly different scopings based on small seeming syntactic changes....

14:36:07 <timbl_away> Some folks have suggetsed just having "?" for any variable, uni. or exi.

14:36:23 * timbl_away was wirrid about subtle sufferences too.

14:36:27 <timbl_away> worried

14:36:43 <twinkle> There's a good rationale for that

14:36:58 <twinkle> In general, variables can be bound or free

14:37:05 <twinkle> When bound, bound universally or exitentially

14:37:26 <twinkle> So, it doesn't clearly make sense to have a variable that's, in itself, existential or universal

14:38:10 <twinkle> Unless, of course, you have only one scope, then, well, ok, why not?

14:38:31 <twinkle> Also, you have to be careful of the logical equivalences between all and some

14:41:09 <timbl_away> In any one top-level expression, a variable is explicitly quantified with a given scope.

14:42:17 <twinkle> Hmm. But top-levelness isn't an inherent condidtion, is it?

14:43:29 <timbl_away> If two expresions have the same subexpression, which subexpressions are the same except for the scope of a variable is different, is there any use in regarding the subexpressions as equivalent, I wonder.

14:44:35 <twinkle> Er...

14:44:40 <twinkle> I think I need an example :)

14:44:56 <timbl_away> { @forall ?x {{ :a :loves ?x } a :Falsehood } and { {@forall ?x { :a :loves ?x } a :Falsehood }

14:45:34 <timbl_away> Is there are reason to regard the ?x as being the same, or knowing that the two expressions share a subexpression { :a :loves ?x } ?

14:45:35 <twinkle> And the difference betwee them?

14:45:38 <timbl_away> I don't think so.

14:45:43 <twinkle> I see soem apparently rouge {s

14:45:49 <twinkle> The latter yes

14:46:15 <twinkle> They clearly aren't the same

14:46:23 <twinkle> As I can rename one without renamign the other

14:46:34 <timbl_away> { @forall ?x {{ :a :loves ?x } a :Falsehood } } and { {@forall ?x { :a :loves ?x } a :Falsehood }}

14:46:34 <twinkle> They do have an intersubsitutable subexpression

14:46:54 <twinkle> I would count those as the same expression

14:47:04 <twinkle> And if you replaced the latter with a @forsome

14:47:09 <twinkle> I would say they had the same subformula

14:47:34 <twinkle> twinkle is now known as bijan

14:48:03 * timbl_away waves to bijan

14:48:13 * bijan needs aliases for his nicks so he can get all his memos

14:48:15 <bijan> Hiya

14:48:30 <bijan> Twinkle, c'est moi :)

14:49:27 * bijan now tries to remember how to read memos

14:51:52 <timbl_away> Would you say that { {@forall ?y { :a :loves ?y } a :Falsehood }} also shares a subexpression?

14:52:13 <bijan> Absolutely

14:52:32 <bijan> There's only one essential use of a variable in all of those

14:52:35 <danbri> how romantic

14:53:05 * bijan thinks we should use other constants than :a

14:53:25 <bijan> Since I keep reading that as "a is a falsehood that loves everyone" :)

14:53:27 <timbl_away> So you'd be in the conventional "use ?x syntax for all variables" camp.

14:53:48 <bijan> I just gave the arguement for why above, so yes

14:54:07 <bijan> I need a stronger motivatation for departing from the variables need marking, their scope needs a different marking

14:54:11 <timbl_away> yes. And it is trie that these things morph fro uni. to exi .

14:54:35 <bijan> Exactly, being quantified a certian way is NOT a feature of the variable itself.

14:54:53 <bijan> And, as such, seems tricky to make it depenent on the sytnax of the variable identifier

14:55:26 <timbl_away> The reaosn people want a shorhand is to avoid having to write out the quantifier.

14:55:37 <bijan> Well, there are other ways to do that, right?

14:55:39 <timbl_away> Like _:a syntax for an existential for example

14:55:51 <bijan> That works, just make the _: an operator :)

14:56:02 <bijan> But that makes scoping *VERY* tricky to read

14:56:05 <bijan> IMHO

14:56:28 <timbl_away> You have to remember that a loarge amount of N3 is RDF, which only has bnodes. A lot of the stuff people write is just data.

14:56:49 <bijan> _: works, to the degree it does, because there's only one scope

14:56:58 <timbl_away> So one could just make the use of _:a illegal except at the top level.

14:57:05 <bijan> Er...

14:57:14 <bijan> Hmm.

14:57:23 <timbl_away> Or, one could make it always be the local scope. I prefer that, so that you can cut and paset NTriples into a formula.

14:57:23 <bijan> You don't have to

14:57:33 <bijan> They jsut need to be alwasy globally scoped

14:57:44 <timbl_away> not globally.

14:57:45 <bijan> Hmm.

14:57:49 <bijan> Graph global

14:57:53 <bijan> graph/context global

14:58:00 <timbl_away> Ok. yes

14:58:00 <bijan> So the latter is consistent with rdf

14:58:04 <timbl_away> yes

14:58:07 <bijan> since you can take the "graph" to be the local scope

14:58:10 <bijan> But the former isn't

14:58:21 <timbl_away> former?

14:58:56 <bijan> timbl_awaySo one could just make the use of _:a illegal except at the top level.

14:59:07 <timbl_away> yes

14:59:09 * bijan maybe not sure what "top level" means

14:59:26 <timbl_away> In an n3 file, that not enclosed by {}

14:59:28 <bijan> You're still extending rdf with the latter, as you have "dark" parts of the graph :)

14:59:34 <bijan> but that's just a scope

14:59:38 <timbl_away> but i don't like that pproach anyway.

14:59:49 <bijan> I.e., there's an implicit {} around the whole thing?

15:00:21 <bijan> which approach?

15:00:37 <timbl_away> How do you mean? How do you test whether thre is an implict {} around the whole thing?)(

15:00:55 <timbl_away> s/(/^a(/

15:00:57 <bijan> I'm jsut asking about the semantics :)

15:01:12 <bijan> Hmm. Perhaps the behavior of log:semantics?

15:01:13 <bijan> Helps?

15:01:45 <bijan> I mean, that the Top Level behaves just like being inside a {}

15:01:58 <bijan> There aren't extra variable issues or evaluation differences, etc.

15:01:59 <timbl_away> I was agreeing that _:a should be construed as being existentially quantified in the scope of the local graph.

15:02:41 <bijan> Sure

15:02:42 <timbl_away> Yes, top level behaves just like being in a {}

15:03:17 <bijan> So, the question is whether bnodes are scoped only to the local scope and *not* nested scopes

15:03:24 <bijan> I suspect you want that situation

15:03:27 <timbl_away> ":f :g :h." log:parsedAsN3 { :f :g :h }.

15:03:33 <bijan> And that's a clear extention

15:04:15 <bijan> One wouldn't need to do it that way

15:04:21 * timbl_away distracted by what he should be doing

15:04:26 <bijan> For example, you could have them univicol unless expressely declared

15:05:12 <bijan> So {..._:x ...{..._:x...} ... {... _:x ...}

15:05:20 <bijan> There's only two distinct variables

15:05:33 <bijan> The appearce in the first {} vs. the appearence in the second

15:05:40 <bijan> (assuming it doens't appear in a parent

15:06:03 <bijan> Vs. {... _:x ...{@forsome :x. ..._:x...}}

15:06:06 <bijan> Which also has two distinct

15:06:19 <bijan> Because you explicitly set up the shadowin gin the inner scope

15:06:42 <bijan> eww. /me totally not proposing, just explorin gthe design space :)

15:06:58 <bijan> Ohhhhh, here's a grosser one!

15:07:08 <timbl_away> :)

15:07:10 <bijan> _:pathexp:x

15:07:16 * timbl_away is glad bijan not proposing

15:07:36 <bijan> So you can put the scope on the variable right by the variable :)

15:07:48 <dajobe> eww indeed

15:08:40 <bijan> Doesn't get much grosser than that

15:09:03 <dajobe> that sounds like a challenge I'd rather nobody won

15:09:05 <bijan> All the typeable clarity of hungarian notation with the semantic transparently of shadowable dynamic variables

15:09:21 <timbl_away> :)

15:10:02 <timbl_away> I am convinced abouyt having ?x on all variables, with _:x as a spoecial case in which each {} has its osnw distinct _: namespace.

15:10:25 <bijan> Hmm.

15:10:27 <timbl_away> I am leaning toward _:x as a spoecial case in which each {} has its osnw distinct _: namespace.

15:10:34 <bijan> Can the _: interact with other variables?

15:10:53 <timbl_away> I think that matches the RDF expectations. It means you don't have to look around for other occurrences very far.

15:11:04 <bijan> {@forall ?x ... ?x ... _:x}

15:11:25 <timbl_away> Well, if you convert a _: into a universallly quantified var, it has to have some representation.

15:11:34 <sandro> " _:x as a spoecial case in which each {} has its own distinct _: namespace" sounds pretty good to me.

15:11:47 <timbl_away> in that example, i'd say ?x and _:x are distinct. It that could be an error

15:11:56 <bijan> how about

15:12:09 <sandro> And unquantified ?-variables would be assumed to be universally quantified at the outermost scope? (as in KIF)

15:12:10 <bijan> {@forsome ?x ... ?x ..._:x...}

15:12:24 <dajobe> (why does n3 syntax design seem like the hunt for a yet unused ASCII symbol ...:)

15:12:56 <timbl_away> A requirement is that yuou can take an N3 file, and paseit into {} to make a valid expression with the same semantics. That means you can't have "at the top level".

15:13:09 <timbl_away> That's why currently ?x are by default scoped to the parent level.

15:13:37 <dajobe> ah

15:13:43 <shellac> dajobe: :-) any takers for €?

15:13:48 <timbl_away> dajobe: because you know I have held back $ and % and it bugs you? ;-)

15:14:01 <dajobe> I don't use n3 really, don't ask me

15:14:15 * timbl_away has resisted using inverted A for forall

15:14:57 <bijan> dajobe: because it eschewed s-exprs, of course

15:15:47 <timbl_away> design goal 1. be easier to write than RDF/XML.

15:15:56 <timbl_away> design goal 2. Be easier to write than s-expr

15:15:59 <bijan> in an xml editor...

15:16:02 <dajobe> where would I find an n3 doc that says " ?x are by default scoped to the parent level." I can't find it around swap/... Shortcuts|Tutorial|Primer

15:16:06 <bijan> in emacs?

15:16:42 <bijan> design goal 3. be easier to read than _:x.

15:17:01 <bijan> _:x a HarderToReadThanPerl.

15:17:10 <bijan> _:x a HarderToReadThanAPL.

15:17:29 <sbp> (shellac: aha! Euro sign. can't believe how long that took me...)

15:17:31 <bijan> _:x a HarderToReadThanK.

15:18:05 * mattb wonders if the subset of n3 that approximately equals the rdf model is defined anywhere independently of formulas and all the other features

15:18:16 <bijan> ntriples

15:18:24 <dajobe> I had a deeper look, can onl yfind logForall scoping discussion, not ?x

15:18:34 <mattb> heh, the larger subset than that

15:18:42 <bijan> And it's pretty easy to add the convenience features to ntriples :)

15:18:46 <danbri> mattb, yeah that'd be nice

15:18:47 <bijan> Add prefixes

15:18:50 <mattb> the maximal subset that still approximates RDF/XML

15:18:54 <bijan> Then add []

15:18:56 <sbp> "The definition of ?x is that it is quantified in the scope of not the immediate formula but the next enclosing formula." - http://www.w3.org/2000/10/swap/doc/Reach

15:18:59 <bijan> Then add ; and ,

15:19:04 <mattb> i do like the [ bNodes ], foo a Thing, etc syntaxes

15:19:07 <bijan> And you've got a pretty good set, actually

15:19:09 <bijan> Oh,

15:19:12 <bijan> add "a"

15:19:14 <bijan> too

15:19:16 <bijan> so

15:19:17 <dajobe> sbp: ok. Wasn't terribly obvious there

15:19:18 <mattb> but when i went looking for parsers they were hard to find

15:19:25 <mattb> cos mostly people were shooting for the full cwm set

15:19:35 <sbp> nope. took me a while to find, too. should be moved about. it'd be nice if it were all moved to the ESW wiki

15:19:38 <bijan> Ntriples + @prefix + qnames + "a" + [] + , and ;

15:19:49 <mattb> yup, that's the ticket

15:20:02 <bijan> logger, pointer

15:20:03 <bijan> See http://ilrt.org/discovery/chatlogs/rdfig/2003-09-09#T15-20-02

15:20:06 <mattb> jena/arp does it

15:20:14 <dajobe> + @encoding :)

15:20:16 <bijan> There you are, mattb, the place wher it's written up :)

15:20:20 <mattb> i found one by graham klyne that worked

15:20:34 <sbp> encoding is utf-8

15:20:36 <bijan> Pretty consistent with the general level of N3 documentation :)

15:20:36 <mattb> sbp's barfed on some nested [ ] or somesuch, bad me for not filing a bug report

15:20:54 <sbp> mattb: that's probably my old crappy parser

15:21:02 <sbp> afon.py is my latest, but that's fairly old too, now

15:21:33 <sbp> someone ported my old one into PHP recently... shame they didn't ask first--I could've directed them to the better one :-) oh well

15:21:36 <mattb> cwm didn't pass the "can i treat it as a plain n3 parser after 10 minutes' study" test

15:22:06 <dajobe> I got an n3 parser donation for raptor offered. It's c++ so I'd have to convert it to C.

15:22:16 <mattb> n3 in raptor would be ideal

15:22:31 <dajobe> but ntriples++very few things would be a lot easier

15:22:40 * mattb finding it useful as a hand-edit language for site content, eg http://www.hackdiary.com/rdf/31.n3

15:22:45 <bijan> dajobe, yep

15:22:55 <bijan> Perhpas an esw-wiki page should be set up

15:23:08 <dajobe> I'd stop at Ntriples + @prefix + qnames

15:23:08 <bijan> I think we could bash out even a BNF in a few minutes

15:23:19 * mattb wonders if jena has a grammar for it

15:23:22 <bijan> [] is very convenient, as are , and ;

15:23:33 <bijan> And map pretty naturally into some rdf/xml conveniences

15:23:56 <sbp> + lax whitespace handling

15:23:57 <dajobe> mattb: yes, probably has a javacc one

15:24:02 <dajobe> sbp: sure

15:24:05 <mattb> Antlr, it seems

15:24:09 <mattb> n3.g in Jena-1.6.0 source

15:24:11 <bijan> [dc:creator 'Bijan', 'dave', 'jim'; dc:title "What we did this Summer"]

15:24:25 <bijan> What nicer than the expanded equive

15:24:29 <dajobe> ugh, the implicit bnode

15:24:31 <bijan> Especially with good indented

15:24:32 <timbl_away> mattb, yes. It would be neat to have a BNF for a restricted langauge which if the most you can do which is still rDF.

15:24:40 <bijan> Hmm. I could give up the implicit bnode

15:24:45 <bijan> but , and ; are critical, really\

15:24:50 <mattb> n3.g in jena is 550 lines including copyright messages and much javaness

15:24:51 <danbri> i like the [] notation

15:24:53 <bijan> ; esp.

15:24:59 <timbl_away> --data in cwm strips the store down to rdf, except with literals as subjects

15:24:59 <mattb> i like [] a lot

15:25:11 <bijan> _:x a :Document;

15:25:18 <bijan> dc:author "Bijan Parisa"

15:25:26 <bijan> ; dc:title "foo" etc.

15:25:29 <bijan> That's workable

15:25:29 <danbri> someone write the bnf into the wiki and we'll vote for each production in the grammar ;)

15:25:56 <dajobe> voting for other people doing work, yeah right

15:26:03 <bijan> I think [] is important, upon reflection

15:26:06 * danbri doubts bijan wished to endorse danbri's insane approach

15:26:24 <bijan> So you can cut and paste foaf things, for exapmle witout manually renaming the bnode id

15:26:27 <mattb> aha, http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/jena/jena2/src/com/hp/hpl/jena/n3/n3.g?rev=1.9&content-type=text/vnd.viewcvs-markup

15:27:00 <shellac> mattb: just about to point to that

15:27:05 <bijan> so, i stick with my feature set as the most usuable (for authoring) superset of ntriples

15:27:19 <dajobe> maybe we should just chump the spec

15:27:21 <dajobe> s

15:27:39 * mattb sees nothing in his personal n3 creations that's not in bijan's subset spec

15:27:48 <timbl_away> Anyone care to edit Ntoation3.html and mail me one with a separat production for an RDF-compatibe sublanguage?

15:27:56 <dajobe> <bijan> Ntriples + @prefix + qnames + "a" + [] + , and ;

15:27:58 <timbl_away> What about paths

15:27:59 <dajobe> plus whitespace lax

15:28:18 <bijan> I'm not conversent with paths, myself

15:28:34 <bijan> But the above is pretty much what I see people currently using

15:28:48 * mattb wouldn't miss paths

15:29:01 * dajobe thinks a BNFy one is a better start for a lex & yacc approach

15:29:15 <bijan> Yes, this is nuts with out a real bnf :)

15:29:15 <timbl_away> Paths can be really neat ... and there are times when you just want a path eg in a IRc line.

15:29:36 <bijan> timbl_away: Not worth it this round, I'd have to say

15:29:58 <bijan> Just because I don't get them enough, haven't used them at all, etc.

15:30:30 <timbl_away> They are very simple. .dc:creator or ^dc:creator to traverse forwrad or reverse.

15:30:36 * mattb reads about paths

15:30:52 <mattb> i wouldn't miss them i think

15:31:16 <dajobe> ah, as usual, sandro has a lex lexer

15:31:30 <mattb> n3 wins for me cos of chaining with ',' and ';', shorthand for bnodes and rdf:type, and handy qname us

15:31:31 <mattb> use

15:31:41 <timbl_away> on paths: http://www.w3.org/2000/10/swap/doc/Shortcuts

15:31:44 <mattb> that's a small enough set of features to keep in my head

15:31:47 <bijan> Is there a natural place for this on the esw wiki?

15:31:56 <bijan> yeah, Ithink there's a conceptual tax to paths.

15:32:08 <bijan> Once you say "forward or reverse traversal" :)

15:32:11 <mattb> and . adds more line noise

15:32:25 <mattb> ,;[] enough

15:35:20 <timbl_away> Paths usecase: ( "check out" ?x.dc:title " by " ?x.dc:creator.contact:fullName " ).string:concatenation

15:36:10 <danbri> is that ok if there are two creators?

15:36:14 <timbl_away> trailing " is spurious

15:36:37 <timbl_away> 2 craetors would get you two matching graphs

15:40:56 * bijan off to meeting

15:45:25 <a> a is now known as aharth

15:48:54 <timbl_away> Anyone know an rdf vocabulary for BNF?

15:53:13 * bijan checks the 12th circle of hell

15:53:17 <bijan> Nope, nothing there :)

15:57:56 <sandro> I toyed with one 2 yrs ago, but.... nothing worth showing.

16:04:30 <sandro> it was really the RDF description of the C datastructures blindfold used to store/manipulate the grammar.

16:31:25 <bobo> bobo is now known as mhgrove

17:06:35 * sbp toyed with RDF-in-BNF, too

17:06:53 <sbp> and Len Kasday was thinking about it for EARL

17:06:56 <danbri> or BNF-in-RDF?

17:07:46 <sbp> er, whoops. yeah

17:08:03 <sbp> "I thought a bit about expressing BNF in RDF and decided that it wouldn't be worth the bother unless you adopted something like Sandro's pet project. [link to blindfold]" - http://miscoranda.com/10

17:09:04 <sbp> Len Kasday's idea: http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2000Dec/0043 - "Backus Naur Form (BNF) to XML"

18:00:27 <mdupont-work> mdupont-work is now known as mdupont

18:02:53 <mdupont>http://introspector.sourceforge.net/2003/09/cwm.log-psyco

18:02:54 <dc_rdfig> D: http://introspector.sourceforge.net/2003/09/cwm.log-psyco from mdupont

18:03:32 <mdupont> D:|Psyco trace of cwm running

18:03:32 <dc_rdfig> Titled item D.

18:04:02 <mdupont> D:This is cwm running a standard introspector file. notice how long it needs on emitting.

18:04:02 <dc_rdfig> Added comment D1.

18:04:10 <mdupont> D:Each function is timed.

18:04:10 <dc_rdfig> Added comment D2.

18:04:31 <mdupont> D:2MB file

18:04:32 <dc_rdfig> Added comment D3.

18:04:56 <mdupont> mdupont is now known as mdupont-pub

19:06:04 <_adam_home> _adam_home is now known as adamhill

19:15:05 <kendallc> kendallc is now known as tacky-k

19:51:01 <bijan> Ugh. secodn last call.

19:59:58 <danbri> edd++

20:15:38 <DanC_jam> 2nd last call?

20:16:44 <bijan> Oh

20:16:49 <bijan> Maybe not

20:16:57 <bijan> They're jsut WDs

20:17:10 <bijan> But that suggests a second last call

20:17:14 <bijan> To come

20:17:19 * bijan still a little jet lagged

20:17:37 <bijan> I guess I wasn't expecting any WDs before another LC

20:18:50 <DanC_jam> well, another LC does seem fairly likely... I'm curious about "ugh"

20:19:08 <DanC_jam> i.e. the 1st last call didn't result in "yup; looks good... you missed a typo"

20:19:29 <bijan> Yep

20:19:39 <bijan> It's not that I don't think it needed.

20:19:52 <bijan> I just think it's tiring, esp. for wg members

20:22:41 <danbri> def tiring, yeah

20:22:46 <danbri> but better to fix before rec than after

20:26:11 <bijan> Yeah.

20:26:17 <bijan> Hmm. That makes me want to raise issues )

20:26:18 <bijan> :0-

20:26:20 <bijan> Which is bad :)

21:01:05 <ndw> Off topic, anyone want to look at some slides on XML, RDF, RSS, and XSLT that I'm presenting to the US Govt XML WG tomorrow and offer comments in the next twenty minutes or so?

21:02:30 <danbri> how could i resist?

21:02:49 * danbri grabs bite to eat, back in a few

21:04:52 <steez> i'd like to look at them

21:05:25 * timbl_away wondered where the URI to norm's sldies went

21:09:00 <ndw> oh, all right, for the world: http://norman.walsh.name/scratch/slides.pdf

21:09:03 <ndw> :-)

21:09:12 * ndw notes that URIs in /scratch/ are not persistent :-)

21:10:36 * ndw waves to timbl_away, who maybe isn't. away that is :-)

21:11:22 * JibberJim likes "there's no standards org involved" "use it anyway" on rss.

21:12:11 <danbri> +1, no need to formally standardise everything

21:12:26 * timbl_away is definitely away

21:12:30 <danbri> though RSS community dev't process isn't one I'd go around recommending... rss-dev prety moribund

21:28:10 <AaronSw> C:[tramp|http://www.aaronsw.com/2002/tramp] makes rdf look like Python data structures

21:28:10 <dc_rdfig> Added comment C3.

21:31:00 <JimJibber> JimJibber is now known as JibberJim

21:40:34 <bijan> People have a funny view of standards.

21:40:41 <bijan> Clear specifications are clearly valuable

21:40:59 <bijan> And they don't need to be "standardized" in any sense

21:41:14 <bijan> You can standardize existing practice,which is perhaps the norm

21:41:26 <bijan> And generally involves smushing out disagreement :)

21:41:37 <bijan> I.e., a posteriori standards

21:41:48 <bijan> You can try to make a spec a standard in advance of practice

21:41:56 <bijan> I.e., a priori standardization

21:42:04 <bijan> Sorta on the motivation to avoid later standardization :)

21:42:18 <bijan> "Well, we'll have to agree some time...why not from the start?!"

21:42:25 <ndw> a priori standardization is "design by committee". often less than perfect

21:42:38 <bijan> a posteriori standardization is too

21:42:43 <bijan> And often less than perfect

21:42:47 <ndw> oh, yes.

21:42:51 <bijan> I prefer it, though :)

21:43:04 <ndw> it has different pitfalls.

21:43:08 <bijan> Because it does tend to be driven by existing use and experience

21:43:09 <bijan> Yep

21:43:24 <ndw> I do a lot of design by committee as my day job, so I should talk, huh? :-)

21:43:42 <bijan> We all have crosses :)

21:43:53 <dajobe> I just got an ntriples+what bijan wanted parser mostly working

21:44:02 <bijan> Or crescents

21:44:08 <bijan> Or some other iconof choice

21:44:21 <bijan> Cool, dave!

21:44:32 <bijan> B3!

21:44:33 * ndw has seen some beautiful icons in museums

21:44:43 <dajobe> at least the "," and ";" stuff works

21:44:55 <bijan> Nice

21:58:21 * ndw smells dinner; waves g'night

21:58:27 <ndw> ndw is now known as ndw^afgk

21:58:30 <ndw^afgk> ndw^afgk is now known as ndw^afk

22:00:32 <ndw> ndw is now known as ndw^afk

23:06:04 <adamhill> adamhill is now known as _adam_home

23:24:37 <dajobe> ok, ntriples+a little now works in raptor. i.e. from the rapper command line

23:24:41 <dajobe> just a bit of detail to sort out