Semantic Web Interest Group IRC Chat Logs for 2012-05-17

00:06:06 <gavinc> gsnedders: Well, what do you. It's in fact easy to change respec to make it ISO 2145 :D

00:09:53 <gavinc> http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html# Introduction still needs more work and more N-Triples examples

00:09:54 <dc_swig> A: http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html# from gavinc

00:10:11 <gavinc> dinner time!

00:40:34 <kasei> gavinc: is that document likely to change before the LC? I might send comments, but don't want to spend time documenting typos and such if they'll be fixed before LC.

00:42:38 <gavinc> I will likely read at least once more for typos, otherwise yes please report typos

00:42:54 <dajobe> ohoh

00:43:19 <gavinc> .. other than in missing ('s in the grammar

00:43:39 <gavinc> or at least not if they are just missing in the HTML at not in the raw rule

00:43:46 <gavinc> which is encoded as an attribute on the <tr>

00:43:56 <dajobe> gavinc: could you edit the link under my name to go to www.dajobe.org please?

00:44:32 * drobilla starts SWHATWG, hilarity ensues, etc.

00:44:48 <dajobe> hmm, the tests have not been updated and what's that test 29

00:44:51 <kasei> heh

00:45:21 <dajobe> surely what hixie asks the web gets

00:45:51 <dajobe> s/asks/authors/ would be better

00:46:14 <gavinc> http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html# No more PURL for you

00:46:15 <dc_swig> B: http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html# from gavinc

00:47:12 <dajobe> gavinc: is there a changes section from the turtle note?

00:47:25 <dajobe> oh ther eit is

00:47:49 <gavinc> note, that should be combined with the STOD changes list at the top

00:47:50 <dajobe> zoiks, turtle in html?

00:48:16 * dajobe goes to author an html in turtle

00:49:00 <gavinc> Pish, already have a new rdf:HTML datatype :D

00:49:10 <drobilla> On that, can we just allow both datatypes and languages in the syntax?

00:49:11 <dajobe> looks like I have to make a couple of code changes to raptor then

00:49:15 <drobilla> Screw what the RDF spec says :)

00:49:48 <gavinc> Yeah, the bigger change is to N-Triples

00:49:56 <dajobe> is it?

00:50:12 <dajobe> what?

00:50:14 <gavinc> UTF-8, not single representation of each character

00:50:20 <dajobe> bah

00:50:28 <gavinc> and a mediatype

00:50:29 <dajobe> although that's trivial for me to impl

00:50:41 <dajobe> I totally understand why

00:50:49 <gavinc> I wanted a new name, but oh well

00:50:57 <dajobe> nontriples

00:51:01 <drobilla> Not having a new name for that is.. odd

00:51:12 <gavinc> Well... see N-Triples is just for test cases ;)

00:51:22 <rszeno> ntbf? :)

00:51:30 <dajobe> wtfnt

00:51:32 <kasei> not a single representation of each character? there goes ntriples as the command-line friendly syntax...

00:51:39 <drobilla> Annoying when reality doesn't conform to your excuses, isn't it?

00:51:58 <gavinc> kasei: ah, no it still works for command line stuff

00:52:05 <gavinc> kasei: at least if it doesn't that would be a bug

00:52:26 <dajobe> gavinc: you really should update the test cases, or I will have to do it anyway

00:52:35 <kasei> oh? I can still do a cut and join? without a single representation, I'd expect that that breaks.

00:52:46 <gavinc> kasei: yes.

00:52:55 <gavinc> sigh, I signed on to update to the test cases... no one else signed up to edit

00:53:11 <gavinc> I do in fact care a great deal about the test cases...

00:53:16 <dajobe> great

00:53:45 <dajobe> I just spent today fighting rdfa test case. a rather surprising way of presenting them

00:54:06 <kasei> gavinc: how does join still work if terms can be expressed in multiple different ways? perhaps I've misunderstood what you said.

00:54:20 <gavinc> kasei: CHARCTERS can be

00:55:22 <dajobe> is there a canonical form?

00:55:23 <kasei> ok...?

00:56:31 <gavinc> Spïdermann vs. Sp\u00EFdermann

00:56:45 <gavinc> dajobe: Canonicalization of RDF was not taken on by the RDF WG

00:57:00 <kasei> gavinc: that breaks using cut/join.

00:57:22 <dajobe> I just meant of ntriples encoding. it would be easier if it always generated \u or the utf8, never a mixture. pick one

00:57:30 <kasei> or any other line processing that isn't doing unicode handling.

00:57:55 <gavinc> dajobe: Oracle objects to always UTF-8 (or did) and everyone else objects to always \u

00:58:04 <dajobe> lol

00:58:47 <gavinc> kasei: correct, n-triples now requires unicode safe tools

00:59:17 <gavinc> kasei: It did in the first place anyway, \u00EF would need to be treated as a single character which nothing does

01:00:05 <rszeno> premonition, will be a big mess

01:00:24 <kasei> well, in most situations, that was only necessary if \u escapes were being used for printable ascii chars.

01:00:24 <dajobe> legacy ntriples, you heard it here first

01:00:45 <kasei> if they were only used for chars that couldn't otherwise be encoded, the tools didn't need to understand unicode.

01:01:03 <gavinc> kasei: ... err... N-Triples used them for everything outside of ASCII

01:01:15 <kasei> i know. that's my point.

01:01:53 <dajobe> I don't grok the \u00EF point

01:02:28 <gavinc> I'm confused as to what Spïdermann breaks?

01:03:07 * dajobe wanders off

01:03:17 <gavinc> Or if you prefer what Iñtërnâtiônàlizætiøn in N-Triples breaks, and why that breaks cut and join?

01:03:30 <kasei> it breaks trying to find that literal in several triples, because in each triple it can now be encoded in a different way.

01:03:44 <rszeno> for example md5(first version) != md5(second version)

01:04:13 <gavinc> rszeno: Sure, but N-Triples didn't define order anyway

01:04:37 <kasei> "order"?

01:04:56 <gavinc> the order of triples in the document

01:05:05 <kasei> in the old n-triples, there was one way to represent that triple.

01:05:11 <gavinc> kasei: yes.

01:05:12 <kasei> now there are dozens.

01:05:37 <kasei> so if I try to invoke join on the command line, it won't work anymore.

01:05:39 <gavinc> kasei: yes, but you or your serarlizer can output all in utf-8 and than there is one

01:06:04 <kasei> that's fine, but I started this out asking if this broke cli tools like cut/join, and you said no.

01:06:33 <kasei> it sounds like "yes, but just use a compliant (new) n-triples parser instead"

01:06:34 <gavinc> kasei: okay, I'll change my statment. Mixing escaped and UTF-8 n-triples can

01:07:35 <gavinc> also for the pedantic... that was true in the first place anyway, as N-Triples doesn't require unicode NFC nor does RDF ;)

01:08:05 <gavinc> and we wouldn't really be here if we weren't pedantic would we? ;)

01:08:24 <rszeno> after serialization if is not used same convention for utf-8, the original and the result will never match. How you check if tests are passing or fail?

01:09:12 <gavinc> ... convention for utf-8?

01:09:33 <rszeno> yes, all \u or utf-8 chars and no \u

01:10:17 <gavinc> yeah... N-Triples (2012) is not for test cases N-Triples Test Case Document is

01:10:24 <rszeno> also one could choose to mix this two cases for some reason

01:10:25 <gavinc> I -said- the name sucks :(

01:11:18 <kasei> gavinc: not sure I understand about NFC concerns. that's a problem in the entire stack, no? it's not an n-triples issue.

01:11:52 <gavinc> kasei: Two literals can say the same thing and have diffrent sets of bytes in N-Triples now

01:12:27 <gavinc> can! Not at all likely!

01:12:28 <kasei> yeah, but that's not only an n-triples issue.

01:13:17 <gavinc> Ah, but RDF defines the comparision of things by value (other than IRI) not lexical form

01:13:47 <gavinc> at least under D-entailment

01:14:12 <kasei> which is a specific case

01:14:50 <gavinc> Yes, otherwise 001 != 1 != 1.000 != 00000000001

01:14:52 <kasei> anyway, I'm done arguing this. I'm not particularly happy about the change, but at least I think I understand it now.

01:15:40 <gavinc> kasei: if you think it's worth defining better how to provide N-triples that DOES meet your use case please send an email

01:16:13 <kasei> gavinc: not worth my time. the old n-triples did that job.

03:48:13 <greg-g> greg-g is now known as NinetiesLinuxUse

03:48:24 <NinetiesLinuxUse> NinetiesLinuxUse is now known as greg-g

06:13:45 <kpreid_> kpreid_ is now known as kpreid

07:46:59 <teswtingr> test!!

08:28:35 <myhelp> i am here ...

08:32:34 <Simpson_2> thats your first triple !

08:36:41 <rszeno> in fact are 1 2/3 triple

09:11:10 <awesome1> just a test

09:24:03 <ktkNA> ktkNA is now known as ktk

09:35:51 <awesome1> hi guys

09:36:15 <awesome1> any expert on linux here?

09:39:16 <mischat> what is teh question

09:39:18 <mischat> ?