Semantic Web Interest Group IRC Chat Logs for 2001-12-05

02:17:33 <AaronSw> <sbp> but people don't use the word "cheese" to denote things on the Web. URI-views are maintened so that they are compatable with URIs, that they can be used alongside them

02:17:41 <AaronSw> <AaronSw> Umm, no they're not. Back up that point.

02:17:41 <sbp> Grgpmrpgmrgh

02:17:43 <AaronSw> <AaronSw> sbp, want to take this to #rdfig?

02:17:48 <AaronSw> <sbp> they're defined in the URI specification, they're used in browsers, the xsd:anyURI datatype actually means any URI-references, etc., etc.

02:18:15 <sbp> not really, but it seems that I have no choice :-)

02:18:17 <AaronSw> I bet the terms MUST and MAY are defined in the URI specification too.

02:18:32 <sbp> they're not defined to work on the Web

02:18:42 <AaronSw> Well neither are URI-refs when you look at it.

02:18:45 <sbp> [[[

02:18:45 <sbp> The term "URI-reference" is used here to denote the common usage of a

02:18:45 <sbp> resource identifier.

02:18:46 <sbp> ]]]

02:18:46 <AaronSw> HTTP doesn't support them

02:18:53 <AaronSw> HTTP doesn't support them, etc, etc.

02:19:10 <AaronSw> You can't use them with WebDAV, you can't do access control on them

02:19:12 <sbp> so what that it doesn't "support" them; it's not supposed to. It would be utterly besides the point

02:19:19 <AaronSw> Why?

02:19:26 <AaronSw> Google doesn't support them, etc.

02:19:35 <AaronSw> When you change all that, let me know.

02:19:37 * bijan doesn't support them.

02:19:47 <AaronSw> see!

02:19:50 <AaronSw> Even bijan!

02:19:54 <sbp> bijan doesn't count

02:19:56 <bijan> But I don't support Aaron either.

02:20:10 <AaronSw> continuation of approx. http://blogspace.com/swhack/chatlogs/2001-12-05#T02-18-35, btw

02:20:12 <sbp> ah, now he does count :-)

02:20:36 <AaronSw> anyway, they don't identify Resources, do they?

02:20:38 <bijan> I always count. 1, 2, 3, 4,5, 6,2,3,6,10...see?

02:20:42 <AaronSw> So why should we use them as such?

02:20:59 <sbp> Well, they must identify soemthing

02:21:15 <AaronSw> Well, only as much as anything does.

02:21:34 <sbp> I think it's quite clear that they identify things, and that they have utility, otherwise they wouldn't have been defined; they wouldn't exist

02:22:25 <AaronSw> So you're saying "cheese" shouldn't exist unless it identifies Resources?

02:22:49 <AaronSw> That seems silly.

02:22:55 <AaronSw> It's in the spec for backwards-compatibility.

02:22:59 <sbp> Well, it does identify a resource, "cheese". The point is that it wasn't defined to work on the Web, and to interoperate with URIs

02:23:10 <AaronSw> Neither can URI-refs!

02:23:14 <sbp> It's in the spec. [period]

02:23:30 <AaronSw> I don't see how that's relevant.

02:23:35 <AaronSw> So is the author's address!

02:23:39 <sbp> well, I can put "http://example.org/#blargh" in an a@href, or use it as a namespace

02:23:48 <sbp> I can't do that with "cheese"

02:23:49 <AaronSw> That's a bug.

02:23:53 <sbp> bug! lol!

02:23:54 <AaronSw> the first part

02:24:03 <AaronSw> It's a well-known bug with the namespaces spec.

02:24:07 <sbp> it's a fact of life. Stop fighting it

02:24:13 <AaronSw> A fact of life?!

02:24:14 <bijan> Hmm. This sounds familiar...

02:24:22 <sbp> On the contrary, it was a careful decision

02:24:27 <bijan> only yesterday I was playing sbp and sbp was playing aaron.

02:24:40 <sbp> I don't think anyone said "Hmm... URI-refs as namespaces? let's flip a coin"

02:24:50 <bijan> Whew, it's bad enough i descend to sbp, but at least I'm too cool to become aaron.

02:24:51 <sbp> they chose, and they chose right

02:24:58 <sbp> lol, bijan

02:25:09 <AaronSw> They chose wrong.

02:25:27 <sbp> who says, and on whose authorty?

02:25:36 <AaronSw> I mean, just because others screwed up, we should too?

02:25:46 <bijan> yes.

02:25:55 <sbp> oh, please. it's too late to change anything anyway, even if I did believe that we should

02:26:00 <AaronSw> Exactly!

02:26:04 <AaronSw> You're not going to change HTTP.

02:26:05 <AaronSw> Or Google.

02:26:10 <AaronSw> or any of that stuff you'd break.

02:26:11 <sbp> URI-refs are used in RDF by everyone

02:26:24 <AaronSw> Why is that relevant?

02:26:26 <sbp> nothing is broken, RDF plainly works as it is

02:26:32 <AaronSw> Everyone watches stupid Saturday morning cartoons too.

02:26:37 <sbp> if it ain't broke, don't fix it

02:26:46 <sbp> Saturday morning cartoons are great!

02:26:48 <bijan> No nono. You're confused.

02:26:54 <AaronSw> Ok, fine, then stop claiming RDF is something that it isn't.

02:27:01 <bijan> The rule of technology is If it *is* broke, *then* fix it.

02:27:12 <bijan> Er..If it *isn't* broke...

02:27:14 <AaronSw> Rename it to the fragemntyflavored description framwork and i'll go away

02:27:16 <bijan> *then* fix it.

02:27:30 <AaronSw> But don't claim that it's about Resources.

02:28:00 <sbp> Let's call it Fred. I don't care; I just want to use it

02:28:03 <AaronSw> Ok, fine.

02:28:11 <AaronSw> I'll use my own thing.

02:28:12 <bijan> What *are* you arguing about?

02:28:19 <sbp> Fair enough, you do that :-)

02:28:31 <AaronSw> Let me know when you rename the spec Fred.

02:28:36 <AaronSw> Oh wait, you're not on the Working Group!

02:28:43 <bijan> "Fred, Model and Syntax"

02:28:52 <sbp> Thankfully

02:29:03 <AaronSw> We'll just go back to calling them Pumkins.

02:29:08 <AaronSw> Pumpkin Model and Syntax.

02:29:14 <sbp> Deeleys. I kept telling you

02:29:37 <sbp> Pumpkin is just fine, though

02:30:05 <AaronSw> OK.

02:30:12 <AaronSw> Just don't try and masquerade them as related to the Web.

02:30:24 <sbp> but seriously, I'll laugh quite a bit if RDF stops using URIs with fragments

02:30:33 <AaronSw> All the more reason to do it!

02:30:36 <bijan> Hmm. from what I can tell, Aaron's completely wrong.

02:30:37 <AaronSw> We'll bring joy to the world.

02:30:38 <sbp> they are related to the Web! How can you not accept that?

02:30:45 <bijan> sean is wrong too, but in a different way.

02:30:49 <AaronSw> They're related to HTML, but not the Web.

02:30:52 <sbp> a@href and xmlns are so pervasive

02:31:01 * AaronSw wonders if he should find out why he's wrong

02:31:03 <sbp> HTML and XML... pretty important bits of the Web, IMO

02:31:10 * sbp too

02:31:16 <sbp> go on bijan, tell us

02:31:25 <bijan> *But* you're all being wondering equivicol in your terminology, too.

02:31:27 <AaronSw> Right, and in HTML and XML they should stay.

02:31:40 <sbp> well, RDF is a format a lot like them

02:31:41 <bijan> "bits of the web" is a weird one, for example.

02:31:45 <bijan> "On the web" another.

02:31:46 <sbp> XML RDF is indeed XML

02:31:54 <AaronSw> OK, never mind, bijan.

02:32:12 <sbp> "bits of Web architecture"

02:32:12 <AaronSw> sbp, I mean, if you want to talk about bits of documents, then OK.

02:32:17 <bijan> But that's not why your wrong :)

02:32:22 <AaronSw> But that's not what RDF is using them for.

02:32:25 <sbp> s/your/you're/

02:32:38 <sbp> true, and that exposes a bug...

02:32:43 <bijan> Typo flames, *that's* why sbp is wrong.

02:32:49 <AaronSw> Heh.

02:33:04 <sbp> heh, heh

02:33:56 <bijan> I take it this is more hashing and slashing?

02:34:03 <sbp> yes

02:34:43 <bijan> is the aaron side, roughly, "Hashes aren't uris." and the sbp side, "Everyone uses them anyway?"

02:34:57 <AaronSw> Seems so, doesn't it?

02:35:03 <sbp> Hmm...

02:35:03 <bijan> I mean, is there any more meat to it than that?

02:35:19 <bijan> Heh. You can't have backward compatibility *both* ways :)

02:35:33 <sbp> clearly; when Aaron and I argue, we like to throw in the kitchen sink

02:35:35 <AaronSw> Heh.

02:35:45 <sbp> and when we stop arguing, we create stuff like XKitchenSink

02:35:48 <AaronSw> My kitchen sink only supports URIs, not fragments.

02:35:58 <AaronSw> It insinkerates fragments.

02:36:00 <sbp> My kitchen sink supports both. Go figure

02:36:10 <bijan> Well, there's a bit of a chain, I take it. I mean, aaron has to rely on something like "RDF identifiers aren't uri-ref",yes?

02:36:15 <sbp> so our sinks aren't compatable... I ain't washing up in yours

02:36:28 <bijan> Hmm. Or would you go on to say that they *Shouldn't* be?

02:36:41 <AaronSw> Well, they're certainly not uri-refs.

02:36:47 <AaronSw> That much is clear.

02:36:53 <sbp> RDF identifiers?

02:36:55 <bijan> Er...other way around.

02:36:55 <AaronSw> I would say they should be URIs.

02:37:08 <AaronSw> What is?

02:37:19 <sbp> RDF identifiers at the moment are URI-views; that's what RDF M&S 1.0 says

02:37:32 * bijan doesn't know this terminology well, if at all.

02:37:49 <bijan> URI-References are resolved to resource identifiers by first resolving the URI-reference to absolute form as specified by [URI] using the base URI of the

02:37:50 <bijan> document in which the RDF statements appear.

02:38:01 <bijan> If a fragment identifier is included in the URI-reference then the resource identifier refers only to a

02:38:01 <bijan> subcomponent of the containing resource; this subcomponent is identifed by the corresponding anchor id internal to that containing resource and the extent of

02:38:02 <bijan> the subcomponent is defined by the fragment identifier in conjunction with the content type of the containing resource,

02:38:06 <bijan> otherwise the resource identifier refers

02:38:07 <bijan> to the entire item specified by the URI.

02:38:11 <bijan> Wow, that's a scary bit.

02:38:21 <sbp> yes, you're right

02:38:24 <bijan> I thought I'd encountered all the scary bits of M&S.

02:38:33 <bijan> It continues to surprise me.

02:38:40 <sbp> I don't read M&S since it started giving me nightmares

02:39:27 <bijan> Let me test my understadning of that passage...unless you can figure out a way to give ancher ids to, say, car parts, it's only sensible to use hashs in RDF with documents.

02:40:17 <bijan> or ratehr http://somecar/#steeringWheel and http://somecar/#breaks both refer to the whole car? yeek!

02:40:22 <sbp> I think you're reading too much into the word "subcomponent"

02:40:44 <AaronSw> What I think it says, is that #breaks refers only to a subcompent of the document you get back at the URL, and not anything else.

02:41:06 <sbp> which makes sense

02:41:31 <bijan> Well, technically to a subcomponent identified by an anchor id.

02:41:32 <AaronSw> Well, it seems to contradict tim, I think.

02:42:18 <bijan> I mean, suppose you have an "anchor id" assignment function. And all there is to being a subcomponant is having an id assigned by that function.

02:42:35 <bijan> That's neither incoherent, nor, afaict, in conflict with M&S; it might even be useful :)

02:42:36 <sbp> no, I think it complents the saying exatcly. Subcomponent is not defined as being a physical part of some bag of bits

02:42:56 <AaronSw> Well, it is for most mimetypes.

02:43:00 <sbp> true

02:43:06 <AaronSw> The idea, i think, was for RDF to change that for the rdf mime type.

02:43:11 <AaronSw> but there isn't one yet

02:43:14 <AaronSw> so RDF is really broken

02:43:19 <sbp> yes... and that needs resolving

02:43:22 <AaronSw> since the core RDF terms refer to XPath node sets.

02:43:28 <sbp> broken: very much so

02:43:51 <sbp> the MIME type should have been registered along with the recommendation

02:45:07 <bijan> Hmm. Qua logical constants, it doesn't matter hash or slash. It's all just strings.

02:45:15 <AaronSw> yeah

02:45:21 <sbp> uh huh

02:45:31 <AaronSw> Perhaps all # should be encoded to %23s.

02:45:32 <bijan> The only interesting bit is if you want the URIs to have some semantics with them.

02:45:41 <sbp> Aaron: lol

02:45:51 <sbp> quite an intreguing solution, actually

02:45:56 <AaronSw> Yeah, isn't it?

02:45:58 * bijan is scared that that actually works.

02:46:09 <AaronSw> That's what my mail client does with them

02:46:26 <bijan> Er...spot of KIF, anyone?

02:46:27 <AaronSw> I'm going to write up this proposal.

02:46:34 <bijan> No way!

02:46:47 <AaronSw> Yeah, brb mailing to RDF Core

02:46:48 <sbp> I'm not going to second anything that your screwy email client has come up with :-)

02:46:53 <AaronSw> lol

02:47:15 <sbp> spot of KIF? Go ahead

02:47:30 * bijan wants it on the record that he is not in the *least way* responsible for *any* of this!

02:47:56 <sbp> I shouldn't have said that it was intreguing

02:47:59 <AaronSw> .py import urlparse

02:48:07 <bijan> Yes, it's sbp's fault. good.

02:48:24 <sbp> I meant to say "silly", honest!

02:48:30 <AaronSw> >>> urllib.quote('#')

02:48:30 <AaronSw> '%23'

02:48:37 <AaronSw> it's your fault

02:48:38 <AaronSw> too late now

02:48:59 <sbp> rats

02:49:26 <bijan> I think you should be more upset that all that wind could even *possibly* be resolved by such a solution :)

02:50:05 <bijan> Oh yeah!

02:50:11 <bijan> I got literals as subjects to work in rdf_db.

02:50:29 <sbp> I'm scared, but I will continue to use fragIDs against any odds

02:50:31 <AaronSw> sent.

02:50:57 <bijan> *Excellent*. It makes my parent/child example work.

02:51:11 <sbp> ah: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Dec/0006

02:54:28 <sbp> Interesting that Aaron's proposal will totally feck any hope of using fragURIs if implemented

02:54:58 <AaronSw> How would you use them?

02:55:11 <sbp> exactly, you can't

02:55:24 <sbp> well, I'd come up with a new serialization :-)

02:56:57 <sbp> the problem with this is that I'd have to start putting hundreds of silly %23 files on my site

02:57:15 <AaronSw> Or just fix your namespaces ;)

02:57:25 <AaronSw> we could replace %23 with ? i suppose.

02:57:33 <sbp> eek, no

02:57:58 <sbp> if you're going to screw up XML RDF, screw it up properly, please :-)

02:58:23 <AaronSw> Heh.

02:58:52 <sbp> well, I'll be interested to see what kind of feedback this gets

02:59:43 <AaronSw> Let me know, will you?

03:00:05 <sbp> Erm... huh?

03:00:57 <sbp> let you know what?

03:01:12 <AaronSw> What kind of feedback it gets.

03:02:35 <sbp> Well, you sent it to rdfcore, and you're subscribed there, but I'm not, so you should let me know :-)

03:04:58 <AaronSw> k have fun all

03:05:00 <AaronSw> i'm off again

03:05:41 <sbp> he went!

03:05:57 <sbp> isn't he meant to do "* AaronSw disconnects" or something?

03:06:50 <bijan> Oooo. Things will be *much much* easier with a working rdf_db.

03:11:17 <bijan> Hows that yapps thing comming along?

03:11:56 * tim-gone drops in fleetingly

03:12:02 <bijan> Hiya.

03:12:11 <sbp> I totally forgot about it, actually. Well, avoided doing it

03:12:12 <sbp> Hi Tim

03:12:13 <tim-gone> Sean - I added the forward functions to math.

03:12:25 <sbp> yep, I noticed those, and the math.n3 corrections; cheers

03:12:49 <bijan> Tim, out of curiosity, how tied is CWM to it's current in memory store?

03:13:14 <tim-gone> Hmmm.... good q.

03:13:20 <bijan> I was trying to figure out what one would have to override to hook it to something else, and I hit a point where I said, "And I guess the rest of them too"

03:13:36 <tim-gone> I would have to look at it.

03:13:53 <tim-gone> It is only llyn but that is most of it.

03:14:00 <bijan> Yes, I know ;)

03:14:09 <tim-gone> here is what I thought of doing --

03:14:15 <bijan> I got all excited when I saw llyn separated out...until I realized that it was nearly all of it :)

03:14:34 <bijan> I've actually successfully subclassed RDFStore...

03:14:35 <tim-gone> I was going to allow one to craete, if you like, a built-in on the fly.

03:14:42 <tim-gone> This would be a property which is externally implemented.

03:14:47 <bijan> ...but only the pretty printing :(

03:15:09 <tim-gone> Well done .... hey, if yo can improve the prettiness then send patches.

03:15:27 <bijan> Well, I needed something much less verbose for SemantiChimp.

03:15:38 <bijan> So it's not really general purpose.

03:15:47 <tim-gone> I read that chumped about e output alos -- seemed to be similar to llyn's.

03:16:00 <tim-gone> Ah.

03:16:26 <tim-gone> What I don't do is recursively return bits of text -- I spurt it out. So you can't revisit it.

03:16:36 <bijan> Hmm. I followed you til aorund 'that chumped about e output alos', AHA! Euler's pretty printing.

03:16:48 <tim-gone> If it worked by recursively returningbits of text one could munge it more -- put things which would fit ona line on a line.

03:16:55 <bijan> Hmm.

03:17:19 <bijan> Do you do any collapsing? E.g., inserting appropriate ;s and ,s?

03:17:53 <tim-gone> But that is sort of the last 20% which takes 80% of the time and I think they spirit of this era is to do the first 20% which gets 80% of the fun/results/understanding.

03:17:56 <tim-gone> Yes I sure do insert every , and ; I can.

03:18:05 <bijan> Require a take on the whole document i'd think.

03:18:11 <tim-gone> I start by ordering the statemenmts in order of s then p then o.

03:18:14 <bijan> Ah!

03:18:31 <tim-gone> I order it such that rdf:type < everything else.

03:18:49 <tim-gone> This is needed to get the type first when spitting out XML. And it is nice anyway in N3.

03:19:11 <tim-gone> So it is a set of kludges.

03:19:23 <tim-gone> necessary rather than beautiful.

03:19:27 <bijan> I think I'm clear on how think depends on apply_rules depends on try_rule depnds on match on query...

03:19:48 <tim-gone> But in practice, RDF is (as DanC and I were noting today) often sold or not on how readabvl eit is.

03:20:05 <bijan> But what I'm *not* clear on is how much query depends on stuff like Thing.

03:20:12 <tim-gone> matchis a bit vacuous.

03:20:22 <tim-gone> query is the meat.

03:20:27 <bijan> yes, but it's an awful lot for that bit of vacuity :)

03:20:51 <tim-gone> match is just a thinn layer on top of query which does th e work.

03:21:09 <bijan> I mean, could I just build a query that called out to redland or rdfdb or whatever?

03:21:29 <bijan> Or used gadfly or kwbuckets?

03:21:30 <tim-gone> I think I should go. [off] rice boiling over

03:21:46 <bijan> tim needs a rice cooker.

03:21:55 <tim-gone> yes -- clone query. but beware everywhere that occursAs and self._index are used , the indexes into the store.

03:22:08 <bijan> Ah, yes.

03:22:13 <tim-gone> gadfly? kwbuckets?

03:22:15 <bijan> That's useful to know!

03:22:22 <bijan> gadfly is a pure python RDBMS.

03:22:38 <bijan> It's built on kwbuckets which is a...er...data thingy.

03:22:40 <sbp> .google gadfly Python RDBMS

03:22:41 * tim-gone goes [iff] to do sinner

03:22:41 <tim-gone> ooops

03:22:42 * tim-gone goes [iff] to do dinner!!!

03:22:43 <xena> gadfly Python RDBMS: http://www.chordate.com/gadfly.html

03:22:51 <bijan> Heheheheh.

03:22:53 <sbp> heh, heh

03:23:31 <bijan> "kjbuckets is a Python extension module that provides fast set, graph and mapping operations"

03:23:51 <bijan> kjbuckets is a C extension to python which defines three Python data

03:23:51 <bijan> types kjSet, kjGraph, and kjDict, implemented using a fast and space

03:23:51 <bijan> efficient hash table strategy. The types are tightly coupled and may

03:23:51 <bijan> be combined and manipulated using a collection of fast "set at a time"

03:23:52 <bijan> operations written in C.

03:24:08 <bijan> A SHORT EXAMPLE MODULE USING KJBUCKETS: SIMPLE RELATIONAL ALGEBRA

03:24:08 <bijan> The following is a "relationally complete" implementation of

03:24:08 <bijan> relational algebra, restricted to equality predicates. It

03:24:08 <bijan> is both short (102 LOC) and fast -- I mean FAST. For example,

03:24:08 <bijan> the natural join uses a hash-join algorithm which couldn't

03:24:10 <bijan> be much faster without the use of precomputed indices.

03:24:15 <bijan> There you go, sean, your next project.

03:24:45 <sbp> Hmm...

03:34:00 <bijan> Heh, now of course all my inference predicates are munged to deal with the rdf_db.pl bugs :)

03:35:26 <bijan> Damn, back to the infiit loop.

03:56:50 * ronwalf doesn't quite get cwm...

03:57:27 * ronwalf is having an argument with cwm about what his n3 rules should be doing

03:57:38 <bijan> Join the club!

03:58:50 <ronwalf> bijan: The "don't get" or the "argument with cwm" club?

03:59:02 <bijan> More the former, but I'll embrace the latter :)

03:59:19 <ronwalf> You're working on a prolog version?

03:59:25 <bijan> Er..yes.

03:59:48 <bijan> "Working on" is a bit strong, but essentially :)

04:01:14 <ronwalf> I take it I shouldn't hinge my final grade in Hendler's class on it, should I :)

04:01:26 <bijan> Er...well, depends.

04:01:32 <ronwalf> It works?

04:01:40 <bijan> Oh, no. Not at all.

04:01:56 <bijan> But if you wanted to finish it, I'm sure you'd do well :)

04:02:20 <ronwalf> Hehe, no... not in the two weeks I have left.

04:02:28 <bijan> Pishaw! No biggie!

04:02:50 <bijan> Be bold! very bold. Very very bold!

04:13:11 <sbp>http://infomesh.net/2001/12/terms/

04:13:12 <dc_rdfig> A: http://infomesh.net/2001/12/terms/ from sbp

04:13:22 <sbp> A:|Assorted SW Terms

04:13:22 <dc_rdfig> titled item A

04:13:46 * bijan notes that his minion recruiting technique STINKS!

04:13:50 <sbp> A:Actually, not a serious namespace. It is a test/demonstration for "what goes at the end of an RDF 'namespace'"

04:13:51 <dc_rdfig> commented item A

04:14:39 <sbp> A:Try viewing it in IE, and then Mozilla. IE shows the source XML, with color highlighting. Mozilla shows the page as XHTML, with working HyperLinks

04:14:39 <dc_rdfig> commented item A

04:14:56 <ronwalf> bijan: Maybe later- get Hendler to pay me for it!

04:15:24 <sbp> A:The trick? Use abbreviated RDF, and then put some XHTML in as a literal. Mozilla sniffs the namespace, and displays the XHTML

04:15:24 <dc_rdfig> commented item A

04:17:54 <ronwalf> I'm running across an odd interaction bug with cwm.

04:18:03 <ronwalf> When I call it from the command line, it works fine

04:18:30 <sbp> A:Heh, the [http://www.w3.org/RDF/Validator/|RDF Validator] barfs a bit on the huge literal, forgetting to quote the <

04:18:31 <dc_rdfig> commented item A

04:18:48 <sbp> ron: what's the problem?

04:18:55 <ronwalf> When I call it from within one of my python scripts using os.system, it errors out

04:19:08 <sbp> er... but what's the error?

04:19:10 <ronwalf> Traceback (most recent call last):

04:19:10 <ronwalf> File "/home/ronwalf/bin/cwm.py", line 598, in ?

04:19:10 <ronwalf> doCommand()

04:19:10 <ronwalf> File "/home/ronwalf/bin/cwm.py", line 461, in doCommand

04:19:10 <ronwalf> p.load(_inputURI)

04:19:10 <ronwalf> File "/home/ronwalf/lib/pythonlib/sax2rdf.py", line 479, in load

04:19:12 <ronwalf> self.loadStream(f)

04:19:14 <ronwalf> File "/home/ronwalf/lib/pythonlib/sax2rdf.py", line 487, in loadStream

04:19:16 <ronwalf> raise SyntaxError()

04:19:18 <ronwalf> SyntaxError

04:19:53 <sbp> what are you trying to do to it? Print out the command line that you're sending it

04:20:33 <ronwalf> It's a little hacked around with, since I was trying everything to make it work but ....

04:20:35 <ronwalf> if os.system('./foo.sh') != 500:

04:20:36 <ronwalf> ...

04:20:38 <ronwalf> foo.sh:

04:20:47 <ronwalf> #!/bin/bash

04:20:47 <ronwalf> cwm.py -rdf temp.daml -n3 database.n3 -filter=searchrules.n3 > tempout.daml

04:21:00 <ronwalf> running foo.sh by itself is fine.

04:21:22 <sbp> Hmm... that is odd

04:21:44 <ronwalf> It almost has to be an enviroment variable.

04:22:20 <sbp> are you sure that the .sh file works? It should have "python cwm.py", AFAICT

04:22:45 <ronwalf> I run ./foo.sh, and everything pops out all right.

04:22:58 <sbp> I have a .sh in my /bin, that goes something like "#!/bin/sh\npython /home/cwm/cwm.py $*"

04:23:02 <ronwalf> I've modified cwm.py to point to my version of python

04:23:08 <sbp> ah

04:24:09 <sbp> perhaps you can modify sax2rdf a bit to find out what it barfs on

04:24:44 <ronwalf> Possibly. I'm checking the env variables for the moment.

04:26:13 * sbp wonders if it's O.K. to have a central file defining lots of terms pointing at tons of RDF documents using rdfs:isDefinedBy

04:27:30 <sbp> I suppose that's one of the purposes of rdfs:isDefinedBy :-)

04:27:54 <ronwalf> Heh, that works. As long as your tools can take it

04:28:13 <ronwalf> Hm, I doubt it is the env variables. They are nearly identical.

04:31:15 <ronwalf> The only thing that is different is the shlvl

04:32:28 * sbp reckons it's great that so many people are using CWM now; even if they can't get it to work :-)

04:33:03 <ronwalf> Hehe. I don't like the last part, but it is too true.

04:33:16 <ronwalf> The function that is throwing the error is

04:33:22 <ronwalf> def loadStream(self, stream):

04:33:22 <ronwalf> s = xml.sax.InputSource()

04:33:22 <ronwalf> s.setByteStream(stream)

04:33:22 <ronwalf> try:

04:33:22 <ronwalf> self._p.parse(s)

04:33:23 <ronwalf> except xml.sax._exceptions.SAXException, e:

04:33:25 <ronwalf> raise SyntaxError()

04:33:27 <ronwalf> self.close()

04:33:45 <ronwalf> It's raising that error.

04:34:55 <sbp> add " print 's: '+str(s)" after the "self._p[...]" bit

04:35:11 <sbp> and then run it once using the .sh, and once through Python os

04:35:16 <sbp> compare the result :-)

04:35:48 <ronwalf> It shouldn't make it past the self.p part: that's what is throwing the error.

04:36:10 <sbp> er, sorry, above the self._p bit...

04:36:39 <sbp> then again, there shouldn't be a difference; they should both be instances of InputSource

04:36:47 <sbp> the error probably lies in the stream

04:36:51 <sbp> print that out too

04:36:52 <ronwalf> How about printing the error?

04:37:30 <sbp> if you think it will help :-)

04:39:23 <ronwalf> Hah, it killed my bot!

04:39:39 <ronwalf> oh well, I'll fix that part later

04:39:43 <sbp> oops!

04:40:10 <ronwalf> the error is: <unknown>:1:0: no element found

04:40:27 <ronwalf> The two s thingies look the same.

04:41:44 <sbp> how about stream?

04:42:09 <sbp> that indicates to me that it's not loading the document for some reason

04:42:15 <sbp> [pretty obvious

04:42:16 <sbp> ]

04:46:24 <ronwalf> stream: <addinfourl at 136045380 whose fp = <open file '/home/ronwalf/class/cmsc498x/edas/search-bot/temp.daml', mode 'rb' at 0x824f9f0>>

04:46:33 <ronwalf> Why is mode rb?

04:46:56 <ronwalf> That shouldn't matter for unix though

04:47:29 <sbp> was that for the shell, or for the Python os?

04:47:58 <ronwalf> for Python os. Shell is:

04:47:59 <ronwalf> stream: <addinfourl at 136424588 whose fp = <open file '/home/ronwalf/class/cmsc498x/edas/search-bot/temp.daml', mode 'rb' at 0x821bce8>>

04:48:20 <sbp> heh, that's pretty weird, then

04:48:54 <ronwalf> This makes absolutely no sense...

04:49:26 <ronwalf> Why should it run differently? Is it a matter of stdin and stdout being redirected?

04:49:33 <ronwalf> I wouldn't think so.

04:50:44 <sbp> I honestly don't know. I've ran files through CWM using os.popen before, and it's been fine

04:52:16 <ronwalf> I guess I will try that...

04:58:09 <ronwalf> It seems to work...

04:58:13 <sbp> great!

04:58:43 <ronwalf> wait... spoke too soon.

04:59:36 <ronwalf> Nope, it doesn't work.

04:59:41 <sbp> Ugh

05:07:36 <ronwalf> Ah well, I will poke at it later.

05:07:40 <ronwalf> 'night

05:08:29 * sbp does some more SWN stuff

05:08:32 <sbp> c'ya ron

05:08:47 <sbp> hope you manage to fix it :-)

05:13:19 * ronwalf jumps out of bed and starts smashing his head on the monitor

05:13:29 <ronwalf> How do you flush a file you've just written to?

05:14:44 <ronwalf> os.close

05:15:51 <sbp> yep

05:15:54 <sbp> os.close()

05:16:53 <ronwalf> Hm, it wants an int?

05:17:19 <ronwalf> Got it

05:17:28 <ronwalf> filename.close()

05:17:43 <ronwalf> That's not what it says in the docs...

05:17:47 * sbp smacks self

05:18:06 <sbp> that closes the file, it doesn't delete the content

05:18:39 <sbp> to truncate it, you'll either have to write nothing to it, or open it again with truncate

05:19:00 <sbp> (write nothing assuming you're not appending)

05:19:05 <ronwalf> No, I just needed the output flushed to disk

05:19:23 <ronwalf> The problem was that I was writing to a file and immediately trying to read it

05:21:23 <sbp> ah yeah, you have to close and reopen it (which is annoying: there should be a shortcut, IMO)

05:25:12 <ronwalf> filname.flush() should do it

05:25:17 <ronwalf> filename, even

05:30:45 <ronwalf> Off to bed again

05:32:31 <sbp> 'night :-)

06:27:05 <tav`> tav` is now known as tav``

06:28:18 <tav``> tav`` is now known as tav

15:48:34 <dajobe>http://www.w3.org/2001/10/stripes/

15:48:34 <dc_rdfig> B: http://www.w3.org/2001/10/stripes/ from dajobe

15:48:48 <dajobe> B:|RDF: Understanding the Striped RDF/XML Syntax

15:48:48 <dc_rdfig> titled item B

15:48:55 <dajobe> B:by Dan Brickley

15:48:55 <dc_rdfig> commented item B

15:48:58 <DanC> bwm, did you write that entire P3P schema by hand? you transcribed the entire P3P base data schema by hand?

15:49:00 <dajobe> so I can find it again

15:50:15 * bwm looks shamefaced and admits to not having perl skills to do it otherwise

15:50:15 <DanC> did you get a chance to peek at my grokPolicy.xsl and the machine-generated forShoppers.rdf, by the way?

15:50:43 <DanC> "the bane of my existence is doing things that I know the machine could do for me" -- me

15:51:06 <DanC> well, I applaud your persistence. that was a heap-load of work!

15:51:10 <bwm> Not yet - doing now - there is never the time to learn the skills that will save time

15:51:22 * DanC takes a call...

15:51:32 <bwm> I started xslt but ran into problems

16:48:20 * DanC returns from call...

16:48:36 <DanC> hm... I wonder if there's a pile of P3P test policies that I could use to test my grokPolicy.xsl

16:50:52 <DanC> BLURB:interpreting P3P as an RDF vocabulary

16:50:53 <dc_rdfig> C: interpreting P3P as an RDF vocabulary from DanC

16:51:24 <DanC> C:I wonder if there's a pile of P3P test policies that I could use to test my [grokPolicy.xsl|http://www.w3.org/2000/10/swap/p3p/grokPolicy.xsl]

16:51:25 <dc_rdfig> commented item C

16:51:38 <bwm> I believe there are some

16:51:39 <DanC> C:hm... [P3P developer resources|http://www.w3.org/P3P/develop.html]

16:51:39 <dc_rdfig> commented item C

16:52:09 <DanC> C:phpht. the [safezone example|http://www.w3.org/P3P/safezone.html] looks like not-well-formed XML.

16:52:09 <dc_rdfig> commented item C

16:52:17 <bwm> I was wondering whether your intention was the grokpolicy would produce output conforming to the schema I've been working on

16:52:30 <DanC> not sure.

16:52:55 <DanC> in some ways, I'm just writing down whatever seems staightforward, then comparing it with what you've done.

16:53:09 <bwm> so long as I'm not wasting my time

16:53:31 <DanC> no, on the contrary; I'm much more able to understand P3P now that you've written in th RDF!

16:54:02 <DanC> I'm also more interested in this as a cwm use case than a P3P modelling task.

16:54:38 <bwm> I think thats complimentary - I've found as i've done it I've wanted to assume a fairly powerful schema processor

16:55:37 <DanC> C:see also [bwm's work-in-progress of 27Nov|http://lists.w3.org/Archives/Public/www-archive/2001Nov/0080.html]

16:55:38 <dc_rdfig> commented item C

16:56:11 <DanC> C:oops... rather [bwm's 2Dec stuff|http://lists.w3.org/Archives/Public/www-archive/2001Dec/0006.html]

16:56:11 <dc_rdfig> commented item C

16:56:55 <DanC> where you broke contact.telecom.telephonenum.intcode into 2 parts looks hard to automate. I see you've gone away from that in your latest stuff, yes?

16:57:18 <bwm> yes - I'm flattening it

16:57:24 <bwm> more typing

16:57:48 <bwm> ex 3.2 is in your inbox if interested

16:57:50 <DanC> I'm assuming there's going to be data produced in P3P's xml format, and I want to machine-convert it to RDF, I guess.

16:58:12 <DanC> when you mail me stuff, I'd much appreciate it if you'd copy www-archive

16:59:20 <DanC> re-thinking P3P's model is something that might be worth doing, but that would involve lots of collaboration with the P3P work, and I'm having more fun collaborating with the machine ;-)

16:59:33 <DanC> i.e. P3P is fun to work on cuz I can Just Do It.

16:59:48 <bwm> archive: ok

17:00:17 <bwm> I think rigo said there are a bunch of test cases somewhere

17:00:41 <DanC> hmm... I don't see why you use rdf:value in <p3p:contact.telecom.telephone.intcode rdf:value="1"/>

17:01:10 <DanC> [you probably already explained, but I'm not necessarily reading what you wrote in order ;-]

17:01:59 <bwm> so I can do <p3p.contact.telecom.telephone.intcode rdf:value="1" rdf:type="&p3p;Contact"/>

17:02:28 <DanC> why do that?

17:02:50 <DanC> also... why is p3p:ResolutionProcedure-independent-organization not just p3p:independent-organization ?

17:02:50 <bwm> look at the end of example 3.2

17:03:30 <DanC> I'm looking at the end of example 3.2; I see dynamic.cookies. go on...?

17:03:33 <bwm> hey - I like typing - don't you? the names all need cleaned up

17:03:53 * DanC very much dislikes typing any more than necessary ;-)

17:03:54 <bwm> it includes category information - which I've been modeling as type on the bnode

17:04:18 <DanC> I see what you've done (types on bnodes). I don't see why.

17:04:51 <bwm> looking ..

17:04:57 <DanC> hmm... maybe I'm starting to see...

17:05:02 <bwm> from ex 3.2 p3p form: [[

17:05:02 <bwm> <DATA-GROUP>

17:05:02 <bwm> <DATA ref="#dynamic.cookies">

17:05:02 <bwm> <CATEGORIES><state/></CATEGORIES>

17:05:02 <bwm> </DATA>

17:05:02 <bwm> <DATA ref="#dynamic.miscdata">

17:05:04 <bwm> <CATEGORIES><preference/></CATEGORIES>

17:05:06 <bwm> </DATA>

17:05:08 <bwm> </DATA-GROUP>

17:05:10 <bwm>

17:05:12 <bwm> ]]

17:05:19 <DanC> logger, pointer?

17:05:19 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-05#T17-05-19

17:05:34 <DanC> C:see also: [discussion|http://ilrt.org/discovery/chatlogs/rdfig/2001-12-05#T17-05-19]

17:05:34 <dc_rdfig> commented item C

17:07:20 <bwm> I need something to hang the category info on to - can't do literals as subjects yet.

17:09:10 <DanC> the reuse of p3p:DATA syntax in different contexts is... icky.

17:09:35 <DanC> sometimes it's a template, sometimes it's filled in.

17:09:41 <bwm> just so

17:11:04 <DanC> hmm... it seems you're connecting a p3p:STATEMENT to an rdf:Statement via a p3p:data arc; makes a certain amount of sense... why do you use rdf:value, then? why not rdf:object?

17:11:47 <bwm> What are you looking at? the mail I sent?

17:12:06 <DanC> er.. yes

17:12:49 <DanC> er... hm... in the case of p3p:entity, you just cut out the <DATA> glorp altogether. Why use rdf:value, then?

17:12:52 * bwm figures out how to get eudora to show attachements of mailed messages

17:14:13 <DanC> sigh... this P3P data-schema stuff really is a re-invention of RDF. what a massive coordination failure.

17:14:50 <bwm> [ ]--p3p:entity->[ ]-p3p:business.name->[ ]-rdf:value->"CatalogExample"

17:15:09 <DanC> just can't bring yourself to use N3, huh? ;-)

17:15:31 <DanC> [ p3p:entity [ p3p:business.name [ rdf:value "CatalogExample]]].

17:15:32 * bwm is shamed again by DanC

17:15:42 <bwm> :)

17:16:07 <DanC> in fact, N3 had support for --> thingies in the syntax. it was recently declared vestigial/obsolete.

17:16:30 <DanC> but why not just [ p3p:entity [ p3p:business.name "CatalogExample"]]. ?

17:16:35 <bwm> dajobe beats me up if I use it and he is closer than you

17:17:06 <dajobe> hey!

17:17:18 <DanC> sigh... folks are happier using ascii-art with no automatic translation to RDF/xml than they are to using N3?

17:17:19 <dajobe> just stay away from vague things like {}

17:18:04 <bwm> Could do that in entity but not in data. same p3p structure in both so i've kept same structure in rdf. maybe too purist.

17:18:37 <DanC> well, the purist instinct is perhaps consistent with ease-of-automation. not sure.

17:18:56 * bwm wonders whether the collection of things to do/learn should be a queue or a stack

17:19:24 <bwm> think I prefer a stack

17:19:30 <dajobe> there's been lots of bad N-Triples on rdf-core-wg - people can't even get that right

17:20:12 <DanC> but at least there is a formal sense in which they're right or wrong. i.e. you can tell them, with appeal to code, that they're wrong.

17:20:31 <DanC> with ascii-art, it's never clear whether they made a mistake or whether you just don't quite understand their use of the notation.

17:20:37 <bwm> An old Oxford don used to order his mail in a stack - answered the top ones first. Some folks thought he was a really good correspondent, others thought he was really cr*p.

17:20:45 <dajobe> yes, but I didn't since would have distracted from the general flow of the email

17:21:11 <DanC> I use a combination of reading email most-recent-first with periodic meetings.

17:22:47 <DanC> back to p3p... ah: I have a comment in grokPolicy.xsl:

17:22:49 <DanC> <!-- this is not what Brian did... -->

17:22:49 <DanC> <p3pr:data rdf:resource='{concat($DataPfx, $ref)}'/>

17:23:29 <bwm> I don't mind changing names to make translation easier

17:25:51 <DanC> ok... let's see if I can get grokPolicy.xsl to spit out your forShoppers thingy...

17:26:30 * DanC is hungry...

17:30:19 <DanC> brian, you've put the <data ref="#business...."> stuff in the P3P namespace; it belongs in the base data schema namespace, no?

17:30:51 <DanC> and actually, if folks specify a different <data-group base="...">, then it goes in that namespace, no?

17:31:26 <bwm> p3p and data schema namespaces are the same?

17:32:05 <bwm> actually I now believe we need p3pr: a separate namespace for the rdf translation

17:32:43 <bwm> base= does change applicable namespace - matters for xslt, but not from the rdf

17:32:43 <DanC> I had to use a different p3pr: prefix in the XSLT thingy because of the # vs no-# distinction

17:33:00 <DanC> the default base= is not the p3p: namespace, though, see?

17:33:14 <bwm> oh bu**er

17:34:10 * DanC wonders what could substitute for ** to make it offensive

17:36:06 <bwm> f, g, m, n, v

17:36:44 <DanC> a... g.

17:38:45 <DanC> oops: <#business.name

17:39:13 <DanC> hmm... what I actually need to do is take the value of the ref attribute, URI-combine it with the base, then split it into nsname, localname.

17:39:35 <DanC> I think I'll kludge for now and throw an exception when I see stuff I don't handle...

17:41:38 <DanC> ok... getting there:

17:41:41 <DanC> <p3pr:entity rdf:parseType="Resource">

17:41:41 <DanC> <business.name xmlns="http://www.w3.org/TR/P3P/base#">CatalogExample</business.name>

17:44:05 <DanC> C:official [P3P syntax|http://www.w3.org/TR/2001/WD-P3P-20010928/#Appendix_schema] as an XML Schema

17:44:05 <dc_rdfig> commented item C

17:44:56 <DanC> C:what does the base= attribute mean on a DATA-GROUP inside an ENTITY?

17:44:56 <dc_rdfig> commented item C

17:45:49 <DanC> C:or, for that matter, CATEGORIES on DATA in there? or the optional attribute?

17:45:50 <dc_rdfig> commented item C

17:46:11 <DanC> C:TODO: ask the P3P WG. (i.e. cook up an example; send to their comments list)

17:46:11 <dc_rdfig> commented item C

17:53:01 <DanC> hm... I had skipped over disputes-group altogether. not sure I want to implement that yet. But I'll need it if I want rules of the form "if trust-e backs them, they're OK"

17:53:42 * bwm decides that the new data schema must be machine generated - beyond doing by hand - wonders if can get away without it - just describe the general pattern

17:54:08 <DanC> I started grokDataSchema.xsl , in case you're interested

17:54:47 <bwm> I'd have to parse the '.' structure of the attribute values - is that doable in xslt

17:55:08 <DanC> it's doable, but not straightfoward.

17:55:23 <DanC> it involves thinking like a scheme programmer.

17:56:12 <DanC> I implemented the abs+rel=abs algorithm for URIs in XSLT... it involves going character-by-character, using recursion for iteration.

17:57:07 <DanC> cf http://www.w3.org/2000/07/uri43/uri.xsl

17:57:37 <bwm> I'm doodling a java datastructure

17:57:56 <DanC> java's probably much easier.

17:58:03 <dajobe> bwm, DanC: I was wondering if we should think about warning w3 comms about potential syntax WD around 17th Dec

17:58:44 <bwm> :)

17:58:46 <DanC> if the timing is critical, then yes, you should try to reserve some comm team time. Is the timing critical?

17:58:53 <dajobe> nope

17:58:59 <dajobe> but I won't be around after 20th

17:59:31 <bwm> might be best to hold off so you are around to handle feedback

17:59:56 <dajobe> eh?

18:01:52 <DanC> it can be sorta off-putting to say "send us your feedback." then the feedback comes, and there's nobody home to respond to it.

18:02:24 <DanC> not critically bad, but not ideal.

18:02:54 <dajobe> do people expect instance responses over holiday periods?

18:03:11 <DanC> people almost always prefer instant response.

18:03:28 <dajobe> I knew you would type that

18:03:45 <dajobe> reason is to get another WD out end Jan with revisions, in time for f2f

18:03:50 <DanC> if you don't intend to deal with feedback during the holiday period, why solicit it?

18:03:54 <dajobe> or at least progress towards it

18:04:09 <DanC> ah... if it's more internally focussed, that's reason enough to press ahead.

18:10:00 * DanC hunts for P3P implementation reports, test data...

18:15:06 <DanC> umm... dajobe, if you want to publish anywhere near Dec 17, yes, contact the comm team and get in line. A bunch of XML stuff seems queued up for that time.

18:16:55 <DanC> well, I see various implementation reports in progress. I don't see anything that suggests that all these extension mechanisms have been thoroughly tested.

18:25:27 <DanC> C:sent a [comment about DATA-GROUP in ENTITY|http://lists.w3.org/Archives/Public/www-p3p-public-comments/2001Dec/0001.html]

18:25:28 <dc_rdfig> commented item C

18:42:41 <DanC> hmm... remedies...

18:44:54 <DanC> hmm... use parseType="daml:collection"?

18:45:18 <DanC> I think I'll do it just to force the issue...

18:47:15 <sbp> do DAML recognize daml:collection as a QNAme yet?

18:47:46 <DanC> well, the spec hasn't changed.

18:47:52 <DanC> if that's what you mean.

18:48:23 <sbp> I mean: will it change? Are they willing to tell people that the parse type should be a QName?

18:49:25 <sbp> Hmm... rdf-tracking doesn't seem to mention the issue of parsetype extensions

18:49:38 <DanC> what issue?

18:50:40 <sbp> "Other values of parseType are reserved for future specification by RDF." - RDF M&S, and yet people are starting to reserve them anyway: DAML as strings, CWM as QNames, etc.

18:51:11 <DanC> oh yeah; I completely forgot that RDF M&S doesn't define parseType as a QNAme.

18:51:25 <DanC> pls send to www-rdf-comments

18:51:29 <sbp> O.K.

18:56:04 <DanC> C:it's kind of a pain that N3 qnames can't have periods in them.

18:56:04 <dc_rdfig> commented item C

18:57:52 * sbp finds http://lists.w3.org/Archives/Public/www-rdf-comments/2001AprJun/0130.html

18:58:44 <sbp> root: http://lists.w3.org/Archives/Public/www-rdf-comments/2001AprJun/0127

19:07:25 <DanC> would folks flip out if '.' was allowed in N3 qnames? you couldn't write { :fido a :Dog. }. you'd have to write { :fido a :Dog .}

19:08:17 <DanC> hmm... maybe allow '.'s except at the end of a qname? getting uglier...

19:41:25 <oierw``> oierw`` is now known as oierw`

19:41:44 <mnot> just stumbled across: http://www.lumeta.com/firewall.html

19:42:08 <mnot> would be interesting to see firewall rules expressed as n3 rules; then one could do all sorts of interesting things

19:44:27 <bijan> DanC, one could have a quoted qname form.

19:44:34 <bijan> use single quotes or some such.

19:52:37 <tim-gone> You could allow "." inside a qname but then I was thinking of going the oethr way and switching to "." for namespace separator. Advantages: distinguish from statement separator because the latter always fllowed by w/s or punc; (b) closer to python if you wanted to merge the languages ...

19:53:24 <tim-gone> We have the "-" problem already. See DesignIssues/Notation3 note about encoding xml "-" as N3's "__".

19:53:32 <tim-gone> (new)

19:53:53 <tim-gone> Hmmmm. could alway sencode "." as "___"

19:53:59 <sandro> what would you use as the statement separator, if not ".", since you've already claimed ";"?

19:53:59 <tim-gone> (ugh)

19:54:20 <sandro> or as %2c or something.

19:55:00 * tim-gone jealously guards unused punctuation in a small wooden box for important uses later

19:55:41 * sandro thinks it's okay to change the language when you're changing the language.

19:55:52 * bijan whimpers.

19:56:06 * tim-gone every now and again opens the box a crack, peers cautiosly in, and smiles...

19:56:12 <dajobe> I think there's something out of kilter with n3/ntriples use of _:0 (n3 OK, ntriples not) - I've not checked recently

19:56:23 <sandro> As long as we tag language versions properly, supporting old versions is not a problem. (A major reason for blindfold.)

19:56:43 <tim-gone> You mean starting with 0-9?

19:56:49 <dajobe> yeah

19:57:02 * sandro gets a little nervious when Tim smiles.

19:57:03 * tim-gone gets a feeling of deja vu

19:57:13 <dajobe> I was wondering about that solving some local naming thing that came up a few days back here

19:58:03 <tim-gone> N3 should constrain localnames to start with alpha.

19:58:08 <tim-gone> Doesn't it?

19:58:19 * dajobe thinks he saw _:0 generated

19:58:29 <bijan> depends which grammar you're looking at ;)

19:58:40 <dajobe> I'm looking at the output of code

19:59:33 <bijan> Hmm. The design issues grammar doesn't actualy define localnames.

20:00:32 <bijan> Though it does do prefixes.

20:00:53 <bijan> Hmm. Though it seems to let prefixes begin with a number.

20:01:08 <bijan> Ah! I think that's a typo.

20:01:12 <bijan> prefix

20:01:12 <bijan> alpha

20:01:13 <bijan> alphanumeric*

20:01:19 <bijan> when it's meant to be

20:01:21 <bijan> prefix

20:01:25 <bijan> alpha alphanumeric*

20:02:03 <tim-gone> Yes, an omission.

20:02:11 <dajobe> where?

20:02:19 <dajobe> in the HTML docs?

20:02:28 <bijan> tim, would you fix prefix too? It's a little confusing.

20:02:36 <bijan> dajobe, design issues

20:02:56 <dajobe> give me a URL

20:03:12 <bijan> Here: http://www.w3.org/DesignIssues/Notation3.html

20:03:21 <tim-gone> localname.

20:03:23 <dajobe> I was looking at the python modules for various regexes, trying to work out which one was used

20:03:35 <sandro> Tim, do you want you BNF to be formal (machine checked & stuff)? I started working on that, but never really talked to you about it.

20:03:51 <tim-gone> yes, locanme def is an omission. fragid is defined and not used and was probably what I indended

20:04:10 * bijan recommends the Yapps grammar, if debugged.

20:04:17 <bijan> Since you could actually use that with CWM.

20:04:27 <bijan> And thus have less gap between them.

20:05:01 <tim-gone> DanC, why is it a pain that "." i not allowed? Just for conversion?

20:05:01 <tim-gone> from XML

20:05:41 <dajobe> too easy to confuse with the terminating sentence '.', maybe?

20:06:07 <DanC> "." is used all over the place in p3p names.

20:06:17 <tim-gone> One could require that space be used between terminating "." and a following identifier.

20:06:28 <bijan> Yick.

20:06:39 * DanC catches up...

20:06:41 <tim-gone> I'm not very keen on letting people put a lot of punctuation into identifiers.

20:06:52 <DanC> tell that to the P3P group, then, please.

20:07:01 <tim-gone> Python get away with a lot of yick -- to great effect!

20:07:24 <tim-gone> I suppose dots re better than minus signs.

20:07:34 <tim-gone> Minus signs are very connonly operators.

20:07:38 <sandro> You can do all the P3P stuff if you just use <uri> identifiers instead of ns:qname identifiers, right?

20:07:39 <tim-gone> commonly

20:08:19 <tim-gone> Do the P3P folk just use "." or do they use "_" and "-" as well?

20:08:47 <sandro> To my eye, 3-1 and 3 - 1 are both subtraction. But x.y is (x's y) while x . y is perl's strcat(x, y).

20:09:13 <DanC> the yapps grammar is formal. I hope TimBL will invest any future BNF-work there.

20:09:40 <tim-gone> tim-gone is now known as timbl

20:09:41 <bijan> I've been trying to direct folks toward it :)

20:10:08 <DanC> yeah, the <uri> syntax is fully general. cwm resorts to it for P3P sutff. But it's a pain to look at. I hacked notation3.py to allow "." and it looks much nicer:

20:10:24 <sandro> Since I'm fine with mappings between formal grammars, I don't much care which EBNF one uses.

20:10:25 <DanC> :entity [

20:10:25 <DanC> p3dr:business.name "CatalogExample";

20:10:25 <DanC> p3dr:business.contact-info.postal.street "4000 Lincoln Ave.";

20:10:25 <DanC> p3dr:business.contact-info.postal.city "Birmingham";

20:10:25 <DanC> p3dr:business.contact-info.postal.stateprov "MI";

20:10:59 * bijan notes to DanC that there are places where the Yapps grammar diverges from Design Issues and Notation3.py (e.g., single quote delimited string literals)

20:11:01 <DanC> the yapps grammar is handy because there's code to walk it. witness the automatically generated .html version

20:11:06 <bijan> Yes.

20:11:10 <bijan> I *love* that.

20:11:13 <dajobe> afaik, there is an optional space before the final terminating ., so isn't it ambiguous to find it, if . can be inside the qname?

20:11:16 <sandro> URI?

20:11:38 <sbp> the space'll have to be made non-optional

20:11:43 <sandro> That P3P use of "." sure looks like the ('s) operator, even if they are treating it as one symbol.

20:11:56 <DanC> see http://www.w3.org/2000/10/swap/rdfn3-gram.html <- http://www.w3.org/DesignIssues/Notation3

20:12:23 * timbl doesn't find http://www.w3.org/2000/10/swap/rdfn3.g ultrareadable and would be frightened of breaking it.

20:12:32 <DanC> yes, the P3P use of "." is somewhere in the middle between symbol-character and punctuation.

20:12:41 <timbl> Doe sanyone have a parser built from http://www.w3.org/2000/10/swap/rdfn3.g ?

20:12:52 <DanC> yes, tim.

20:13:06 <timbl> DanC, when you made your hack, I assume you checked for trailing alphanumeric after a "."?

20:13:16 <bijan> DanC, the yapps grammar looks like it has all the semantic actions need to replace Notation3.py as the input parser...is there any barrier to dropping it in?

20:13:23 <sbp> I think that "_" needs to be taken out of the PREFIX production, although CWM actually allows you to set it, and override the bNode construct

20:13:24 <bijan> Into cwm?

20:13:28 <DanC> in the hack, I just allowed ".". I didn't do anything reasonable about ambiguities.

20:13:48 <timbl> Good point, sbp.

20:13:59 <bijan> timbl, look at the html: http://www.w3.org/2000/10/swap/rdfn3-gram.html

20:14:00 <timbl> Dan, isn;t eth XML constaintthat a prefix should start with alpha?

20:14:20 <bijan> It hides all the semantic actions, which can be a bit overwhelming if you don't know the syntax.

20:14:27 <timbl> No, sbp the prefix in notation3.html is required to start with alpha.

20:14:27 <DanC> tim, the rdfn3.g compiles to python; output: http://www.w3.org/2000/10/swap/rdfn3_yapps.py

20:14:31 <bijan> Oops., not quite, it subscripts them.

20:14:39 <bijan> Still, a bit easier to follow, IMHO.

20:15:25 <bijan> Hmm. It doesn't quite subscript them, it subscript the thing which is passed between productions or something.

20:15:31 <DanC> tim, if rdfn3.g looks scary, try taking all the {{ }} semantic actions out.

20:15:37 <DanC> or just look at the .html version ;-)

20:15:48 * bijan feels either behind or ahead of the curve :)

20:15:51 <timbl> PREFIX ?

20:15:54 <DanC> or consider that it's no scarier than notation3.py ;-)

20:16:03 <DanC> token PREFIX: r'[a-zA-Z0-9_-]*:'

20:16:04 <timbl> :)

20:16:20 <bijan> Tokens don't show up in the html.

20:16:22 <timbl> Ah . tokens elsewhere.

20:16:39 <bijan> At the tope of rdfn3.g.

20:16:49 <sbp> QNAME also has to start with [A-Za-z] - shouldn't it be the same as PREFIX?

20:17:11 <DanC> I'd think so, sbp. seems like a bug.

20:17:25 <timbl> They are different things, it isnice to have them the same so we can contemplate changing them seperatly

20:17:38 <timbl> Oh sorry QNAME.

20:17:58 <DanC> I intend to change the treatment of 'this' in the grammar; I don't think "this log:forAll :x" should be parsed as a statement. it should have its own production.

20:18:27 * bijan was thinking that way at one point.

20:18:31 <timbl> I would call that changing the tratment of "forAll".

20:18:45 <timbl> this is cleaner than forSome.

20:19:04 <sbp> it'd be great if forAll was defined as a built-in term... because it's not a property

20:19:15 <DanC> ah... good point; it's more forAll that needs special treatment. hmm.

20:19:28 <sandro> What does "grounded" mean, in phrases like "grounded in URI-space" ?

20:19:37 <timbl> yes, theer are two pseudoproperties which re not rdf:Properties.

20:19:41 * timbl hungry

20:20:11 <DanC> grounded has two meanings: (a) syntactically grounded, in the way <dc:title> is just sugar for http://purl.org/dc/elements/1.1/title ...

20:20:45 <DanC> and (b) grounded in the sense that it takes on meaning by use; you can actually use that identifier in the web and get some interesting information about it.

20:20:58 <bijan> dc:titel is sugar for <http://purl.org/dc/elements/1.1/title>?

20:21:00 <bijan> dc:title

20:21:23 <DanC> given the relevant xmlns foo, yes, bijan; this is news to you?

20:21:36 <bijan> Er..i was refering more to the placemen tof the <>

20:21:41 <DanC> oops... sorry... I meant <dc:title> in RDF/xml

20:21:47 <bijan> Ah!

20:22:04 <bijan> No, *that's* not news to me :)

20:22:13 <sandro> I'm trying to think of what to call XML languages which have formally defined semantics. GroundedXML is one term that came to mind, but I'm not quite sure.

20:22:19 <bijan> I was getting a little worried about my n3 understanding though :)

20:23:17 <DanC> well, sandro, one distinction is whether the XML language has any propositional content at all.

20:23:17 <sbp> GXML

20:24:35 <sandro> I've been able to read propositional content into all the XML I've seen (of which I had some human understanding).

20:26:00 <timbl> Grounded - see http://www.w3.org/DesignIssues/Meaning

20:26:12 <timbl> 'We can say a document is "grounded" if its meaning is completely defined because every term used is explicitly, directly or indirectly, an explicit direct or indirect referece to its definition in a document on the Web. Clearly a definition of "grounding" depends on the set of documents one considers acceptable definitions. '

20:26:22 <sandro> another kind of term is "joinable XML", saying that this is stuff you can merge into the one true RDF graph.

20:27:10 <bijan> So, uh, we're kind of departing from the use of "grounded" meaning, roughly, "without variables, at least when fully expanded"?

20:27:11 <DanC> sandro, I wonder about propositional... what I mean is: "4+5" isn't a proposition. "4+5=9" is. "<p>xyz</p>" isn't a proposition. it's more like "4+5".

20:27:57 <sbp> One important point raised therein is that XML schemata should have mechanisms for enabling people to state how langauge should merge with one another: and you can only do that by grounding semantics using URIs

20:28:06 <DanC> well, bijan, not only without-variables, but also: the symbols used in the language have, as their scope, the global conversation that is the web.

20:28:24 <timbl> well, when you fully expand a term you go to its definition and expand that.

20:28:29 <timbl> It is the same general sense.

20:28:40 <bijan> ok

20:28:54 <timbl> If you find a term which doesn't ahev a URI, or for which there is no spec, then the process breaks.

20:29:04 <sbp> But Aaron and I tried doing that with one of our languages... we started inventing a whole new system for mixing languages. It's very difficult... XML Schema goes some of the way by defining things such as substn. groups; but they're syntactic only

20:29:41 * timbl goes [off] to find food

20:29:42 <DanC> P3P is really a fascinating excercise in semantic-anthropology ;-)

20:30:01 <sandro> I take every XML document to be a description of an object corresponding to its root element. Sometimes the stuff you care about is burried deep inside that description, but it's still linked. And <HTML> element is roughly a <WebPage> and the XML contains knowledge about that web page, which a browser can use to learn how to display it, etc.

20:31:03 <sandro> Tim, my definition of grounded basically matches yours -- but is that definition something the world at large follows?

20:31:05 <sbp> I found it rather interesting that Mozilla will render XHTML that it finds inside a parseType="Literal" in XML RDF - just by sniffing the namespace. It probably shouldn't do that; XML RDF doesn't define what to do

20:32:11 <sandro> In the GroundedXML semantics I'm drafting for RDF, that's incorrect behavior.

20:32:44 <bijan> Er...if it's undefined, is any behavior acceptible?

20:32:51 <sbp> I think it's "incorrect" for XML in general :-)

20:34:08 <sandro> Yeah -- when the outer elements are not understood, you shouldn't go poking inside, --- but browsers are supposed to try hard to display something, so I understand why it does.

20:34:21 <DanC> world at large: stay tuned for TAG discussions of self-describing web etc.

20:34:23 <oierw> but not xhtml

20:34:36 <oierw> xhtml is supposed to be valid or dumped

20:34:45 <oierw> no guessing what should be done

20:35:05 * sbp is thrilled in anticipation of TAG :-)

20:35:13 <sandro> DanC, should I sit back and watch, or go ahead and try to implement? :-) (I kind of need that for WSDL containing SOAP and XMLSchema objects)

20:36:36 * sbp is pleased to see more EARL tools from Jim/Nick

20:39:58 <sandro> Sean, my plan is: you mix languages by saying "at this point, there can be a grounded XML element", and then specifying what to do with the RDF graph & object you get back from parsing it.

20:41:34 <DanC> C:[P3P-happy sites|http://www.w3.org/P3P/compliant_sites] provides happy test data

20:41:34 <dc_rdfig> commented item C

20:43:11 <sbp> So it involves always mapping an XML tree onto an RDF graph? What do you do about the QName identification problem?

20:45:15 <sandro> Which QName identification problem? I assume zero about the XML except that I get to a formal anotated grammar from it's namespace name (by web retreival or prior configuration)

20:47:36 <sandro> Indeed, DanC. Nice XML.

20:49:20 <DanC> sit back: no. I hope TAG discussions will benefit from your implementation experience.

20:49:30 <DanC> hmm... what to do with <DATA ref="#dynamic.miscdata"><CATEGORIES><physical/></CATEGORIES></DATA>

20:49:39 * DanC re-reads Brian's ideas...

20:50:00 * edd wonders if TAG election results been announced publicly anywhere yet?

20:50:12 * timbl catches up (sandro, dunno what the world at large thinks.)

20:50:18 <sbp> edd: not yet; Ian seems to be blanning to

20:50:24 <sbp> s/blanning/planning/

20:51:30 <sbp> QName-identification: the fact that the structure of QNames are often undefined, and that you can't really identify them as easily as some people would like to in XML RDF. You can model them... IF you're just getting stuff from the namespace, that's O.K. though

20:56:22 <sandro> Yeah, without external information, XML is unusable.

21:02:10 * DanC chews over AT&T's privacy policy...

21:03:11 <DanC> nope, that model won't work...

21:06:29 <DanC> well... it's easy enough to map XML syntax into an RDF graph, I guess.

21:06:58 <DanC> it's much tougher to tell whether the resulting relationships are supposed to be 1-1, 1-many... whether lists are supposed to have the "and that's all!" feature, and so on.

21:31:28 <DanC> this huge P3P spec is kinda painful to deal with. I'd rather it were split: P3P examples and guiding principles; P3P developer details

21:32:04 <bijan> 3P3PPs? Ah, only 2.

21:33:21 * em waves and catches up with log

21:39:11 <bijan> into account the needs of Python developers as Python matures. XML, on the other hand, though strongly impacted by the ideas of a small

21:39:16 <bijan> cadre of visionaries, has grown from standards committee roots.

21:39:23 <bijan> Oops! Wrong window.

22:12:32 * DanC waves to EricM

22:16:04 <DanC> hmm... ok, so now this grokPolicy.xsl thing is sort a working...

22:16:13 <DanC> now I'm trying to get the non-trivial APPEL rules do go.

22:17:02 * DanC pauses to commit recent work...

22:18:13 * em waves back

22:22:06 <DanC> ok... grokPolicy.xsl etc. checked in

22:24:14 * timbl finds out where OE's "turn off scripting" bit is hidden (In IE's "Internet Options/Security/Internet/Customize/ActiveScripting") and turns it off

22:25:06 * edd fumbles for mutt

22:25:43 * em wishes he had more time to look into evolution 1.0

22:25:50 * em waves to edd

22:25:57 <edd> heya em

22:26:09 <em> howdy edd! :)

22:26:17 <edd> evo 1.0 is the best thing *ever*, imnsho

22:26:52 <DanC> yup... evolution recently gave me hope of abandoning NS4.7

22:26:59 <DanC> but then it crashed or something.

22:27:08 <edd> heh. and windows never does that, right?

22:27:20 <DanC> I wouldn't know if windows ever crashes. I don't use it.

22:27:36 * edd misread NS as NT, doh <shamed/>

22:28:28 <edd> i still have high hopes of extracting rdf from evolution, i've done the basic proof of concept interface stuff. the component model seems quite sensible.

22:29:11 <em> oh! do tell!

22:29:55 <edd> there are several projects i've sort-of got going, based around contacts lists and calendaring

22:30:14 <edd> it's easy to extract libby's calendar rdf stuff out of evolution.

22:31:04 <edd> also handy is that there's plentiful vCard/iCal usage inside evo.

22:31:46 <edd> the ultimate requirement would be a calendar server of some sort.

22:32:03 <edd> do you know if the RDF calendar folk have something of that in mind?

22:32:37 <edd> my problem with develoing this stuff is that i keep getting down to the lower layers, away from RDF

22:33:07 <edd> so i've been learning about OBEX and various IrDA protocols, and am currently on a Bluetooth adventure

22:33:21 <edd> ultimately i'll piece it all back together again and expose data that Useful Things can happen with

22:34:53 * DanC went on an OBEX adventure a while ago...

22:35:08 <DanC> we seem to cover a lot of the same ground, edd, but we seem to be a couple months out of phase.

22:35:09 <edd> did it result in any code?

22:35:15 <edd> heh

22:35:29 * sandro is astounded that google "evolution 1.0" worked, but it did.

22:35:41 <DanC> yes, I found some python OBEX code that allowed data to go from my nokia cellphone to my vaio laptop, as i recall. But not the other way

22:35:53 <edd> my phone has a calendar and speaks OBEX: i want it to sync like my palm

22:36:25 <DanC> edd, me too. cf http://www.roads.lut.ac.uk/lists/gnokii/2001/10/0024.html

22:36:28 <edd> aha, i took the C route (presume we're both referring to the openobex stuff). i even found a bug or two, but it awaits my return

22:36:31 <DanC> gnokii groks 8290 calendar?

22:36:31 <DanC> From: Dan Connolly (connolly@w3.org)

22:36:31 <DanC> Date: Thu Oct 04 2001 - 22:45:02 BST

22:37:03 <edd> my Big Dream is write a sync framework for it like the Palm conduit stuff in GNOME

22:37:11 <edd> maybe over the holidays (ha ha)

22:38:07 <edd> i think there are a lot of twisty special cases with each cellphone device. i found assumptions in the OBEX code that weren't true for my phone.

22:40:21 * edd notes his adventures from http://www.pasta.cs.uit.no/pipermail/obex/2001-September/000191.html

22:40:45 <edd> perhaps we should communicate more DanC, we seem to have been doing the OBEX/cellphone dance at around the same time

22:42:08 <DanC> hmm... I set up a wiki to collect some of my thoughts on this sort of stuff... http://dm93.org/z2001/PerfectOffice but I haven't made it publicly writeable yet.

22:43:15 * DanC noodles on a bot that would keep dm93.org in sync with #rdfig somehow...

22:43:51 <deltab> I once ran PerfectOffice, and it didn't have any of those :-)

22:44:32 * DanC goes back to debugging P3P rules...

22:47:48 <DanC> phpht. found the bug: <http://www.w3.org/2000/12/P3Pv1#>. vs. <http://www.w3.org/2001/09/P3Pv#1> .

22:50:52 <DanC> found another bug...

23:00:03 <timbl> em, you pointed my at this phone ... it has a calendar ... but it doesn't ahve any sync s/w. It points to Truesync(g) from starfish

23:00:17 <timbl> (g) Goodlemark

23:00:20 <timbl> (g) Googlemark

23:01:15 <DanC> aha! got 2nd P3P rule working.

23:01:21 <DanC> check out the antecedent, tim:

23:01:27 <DanC> {

23:01:27 <DanC> :p a p3pr:Policy;

23:01:27 <DanC> log:racine [ log:semantics [ log:includes {

23:01:27 <DanC> :p p3pr:statement [

23:01:27 <DanC> p3pr:data [ p3pr:ref :x ]

23:01:28 <DanC> ].

23:01:30 <DanC> } ] ].

23:01:32 <DanC> :o p3pr:categories [ a _:InterestingCategory ].

23:01:34 <DanC> }

23:01:36 <DanC> log:implies { :p a _:Rule2Target }.

23:02:07 <timbl> Ok, so I need to workon n3 readability? ;)

23:02:51 <bijan> log:racine?

23:02:52 <DanC> you can't parse it? don't try to read much into the p3pr:foo terms.

23:02:56 * edd boggles too

23:03:06 <DanC> <a#b> log:racine <a>.

23:03:19 <timbl> I can parse all the log:stuff

23:03:27 <timbl> I wondred about a "which"

23:03:30 <DanC> log:racine takes the #foo off something, I think. french for root, I gather from tim.

23:03:53 <bijan> Hmm. I predict that the indentation/brace placement wars for N3 are giong to be *vicious.

23:04:07 <edd> not to mention the interlingual ones :)

23:04:29 <bijan> Hm...But why is that a log:?

23:04:31 * timbl nurdles a p3pr:statement [ p:data [ p:ref :x ]]

23:04:43 <DanC> yeah, me too, tim.

23:05:06 <DanC> as I said earlier:

23:05:06 <DanC> <DanC> well... it's easy enough to map XML syntax into an RDF graph, I guess.

23:05:06 <DanC> <DanC> it's much tougher to tell whether the resulting relationships are supposed to be 1-1, 1-many... whether lists are supposed to have the "and that's all!" feature, and so on

23:05:07 <sbp> nurdle?

23:05:08 <timbl> It is interesting herethat you have a long chain (althe close brackets at the end). I wonder how common that will be.

23:05:39 <DanC> danc:noodle = timbl:nurdle. it got garbled over the phone, I think.

23:05:51 <DanC> family time for me. lemme check in real quick...

23:06:14 <timbl> local:nurdle = english:cogitatesALittleAbout"

23:06:27 <DanC> the victory is: I'm taking P3P from the wild (i.e. from AT&T's web site), using grokPolicy.xsl to turn it into RDF, then using cwm rules on it.

23:06:49 <sbp> Thanks. I'd always wondered about nurdle.n3.py, and all I could find were links to Tiddlywinks...

23:06:59 <timbl> In what langauge are your personal langauges expressed?

23:07:05 <DanC> p3pAArules.n3,v <-- p3pAArules.n3

23:07:05 <DanC> new revision: 1.2; previous revision: 1.1

23:07:15 * DanC is away: family time

23:07:22 * bijan adds "DanC victory" next to "pyrrhic victory" in his lexicon.

23:07:52 <DanC> that is http://www.w3.org/2000/10/swap/p3p/p3pAArules.n3

23:09:10 <deltab> is that meant to be public?

23:09:17 * timbl assuems he can give it world access

23:09:51 * sbp finds nothing in http://dev.w3.org/cvsweb/~checkout~/2000/10/swap/p3p/ but knows is behind anyway

23:09:56 <sbp> s/is/it's/

23:10:37 * timbl smiles at """ .... (very long """ a :TODO .

23:11:28 <timbl> /me notes appel:crtdby s:subPropertyOf dc:creator.

23:11:58 <timbl> DC score over appl in making up names.

23:12:21 <timbl> nick tim-fmly

23:12:24 <timbl> timbl is now known as tim-fmly

23:21:24 <dajobe>http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/

23:21:24 <dc_rdfig> D: http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/ from dajobe

23:21:36 <dajobe> D:|RDF/XML Syntax work-in-progress

23:21:37 <dc_rdfig> titled item D

23:21:44 <dajobe> D:what I've mostly been working on today

23:21:45 <dc_rdfig> commented item D

23:22:15 <dajobe> D:new stuff: [http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/#section-RELAXNG-Schema|RELAX NG Schema] (Non-norm)

23:22:16 <dc_rdfig> commented item D

23:22:41 <dajobe> D:plus serialisation notes, more issue updates

23:22:41 <dc_rdfig> commented item D

23:23:10 <dajobe> D:more work tomorrow; current latest version is CVS revision $1.109$

23:23:11 <dc_rdfig> commented item D

23:24:03 <sbp> neat stuff, dajobe

23:24:23 <dajobe> just want to get it more mostly-done; rather than mostly-to-do :)

23:24:31 <sbp> heh, heh

23:25:40 <sbp> It's really well edited, and nicely laid out... Difficult to achieve when you have a WG to answer to

23:25:51 <dajobe> thanks

23:26:19 <dajobe> some issues with doc publishing style yet to be sorted out; even with new style guide

23:26:36 <bijan> Yay! New XML.com with a <taglines/>

23:32:22 * bwm gets p3p data schema generation code running

23:36:04 * sbp wonders what was meant by "In what langauge are your personal langauges expressed?"

23:36:37 <sbp> ah, joke, got it...

23:36:52 * sbp slow

23:39:00 * sbp goes through the P3P stuff

23:57:26 * dajobe waves g'nite

23:59:50 <DanC> bwm, check out grokDataSchema.xsl