12:01:38 RRSAgent has joined #owl 12:01:38 logging to http://www.w3.org/2008/07/29-owl-irc 12:01:57 zakim, this will be owl 12:01:57 ok, IanH; I see SW_OWL()8:00AM scheduled to start now 12:02:32 IanH has changed the topic to: http://www.w3.org/2007/OWL/wiki/F2F3_Agenda#Day_2 12:02:49 Zakim, this will be owlwg 12:02:49 ok, IanH; I see SW_OWL()8:00AM scheduled to start 2 minutes ago 12:03:02 RRSAgent, make records public 12:27:38 IanH has joined #owl 12:34:46 good morning. :) 12:59:58 SW_OWL()8:00AM has now started 13:00:04 +??P2 13:00:13 zakim, ??P2 is me 13:00:13 +uli; got it 13:00:24 and good morning to you! 13:02:11 zakim, mute me 13:02:11 sorry, uli, muting is not permitted when only one person is present 13:02:24 pha has joined #owl 13:02:26 zakim, mute me 13:02:26 sorry, uli, muting is not permitted when only one person is present 13:03:00 Zhe has joined #owl 13:03:55 Hey, I should use this opportunity for advertisement! 13:04:53 Please remember to submit to OWLED 2008 -- deadline for abstracts is next week, more details can be found at http://www.webont.org/owled/2008/ 13:05:26 we are looking forward to receiving your submissions. 13:06:03 Also, around ISWC and OWLED, there will be interesting tutorials -- please have a look at http://iswc2008.semanticweb.org/program/tutorials/ 13:06:15 zakim, mute me 13:06:15 sorry, uli, muting is not permitted when only one person is present 13:06:37 +Meeting_Room 13:06:41 zakim, mute me 13:06:41 uli should now be muted 13:10:03 +??P1 13:10:16 zakim, ??p1 is me 13:10:16 +bparsia; got it 13:10:19 zakim, mute me 13:10:19 bparsia should now be muted 13:11:17 msmith has joined #owl 13:11:31 zakim, who is here? 13:11:31 On the phone I see uli (muted), Meeting_Room, bparsia (muted) 13:11:32 On IRC I see msmith, Zhe, pha, IanH, RRSAgent, Zakim, bparsia, uli, ewallace, jar, Carsten, sandro, trackbot 13:11:42 hiho 13:11:48 hullo 13:11:56 Achille has joined #owl 13:11:58 alanr has joined #owl 13:11:58 pfps has joined #owl 13:12:18 sandro could you please make me the scribe? I forgot the magic command 13:12:38 Mirek has joined #owl 13:12:59 ekw has joined #owl 13:13:10 scribenick Achille 13:13:15 + +00493514633aaaa 13:13:20 bmotik has joined #owl 13:13:24 zakim, aaaa is me 13:13:24 +Carsten; got it 13:13:33 zakim, mute me 13:13:33 Carsten should now be muted 13:13:47 topic: datatype roundup 13:15:02 working from http://lists.w3.org/Archives/Public/public-owl-wg/2008Jul/0306.html 13:15:07 subtopic: Issue 126 : list of normative datatype 13:15:38 ianh: as per the previous email, we decided to have xsd:float discret as in XML Schema 13:16:17 ianh: we need to decide about rational 13:16:32 q+ 13:16:38 q- 13:16:43 q? 13:16:48 mike: we decide to postpone a decision on rational for the time of discussing N-ary datatype 13:17:11 mike = msmith 13:17:12 m_schnei has joined #owl 13:17:32 ianh: owl:numberPlus contains -0 what happen to facet? 13:17:41 s/ianh/boris 13:17:58 alanr: XSD should answer this question 13:18:09 msmith : XSD says that 0=-0 13:18:44 for +/-0 see http://www.w3.org/TR/xmlschema11-2/#float 13:19:10 alanr: my worry is that float and double aren't not the same 13:19:23 s/aren't not/ aren't 13:20:51 msmith: xsd NaN when used for range defined an empty range 13:21:14 It's definitely not maxfloat 13:22:03 alanr: adding 1 to maxfloat gets you to infinity 13:22:20 alanr: same thing for double 13:23:16 boris: any range that contains 0, also contains -0? 13:23:28 boris: XSD does different thing than us 13:23:52 boris: the pb is that we also have true number (integer decimal, etc) 13:24:00 alanr: what is the consequence? 13:24:16 q? 13:24:55 m_schnei: -0 also denotes 0 13:25:12 q? 13:25:20 boris: it should be fine since integer does not contain -0 13:26:13 dlm has joined #owl 13:26:26 I like owl:real 13:26:29 ianh: owl:number is renamed owl:real 13:26:33 q? 13:26:46 ianh: owl:rational depends on the N-ary discussion 13:26:49 m_schnei: if you ask for "{-0} subset [-1,+1]", then the reasoner should say yes, because the float *literal* "-0" denotes the *value* ZERO 13:26:51 q? 13:26:55 ... nothing about integer? 13:27:14 alanr: we should add a min conformance of 64 bit integer 13:27:46 ian: what about "owl:number" vs. "owl:real"? 13:27:59 alanr: what happen when there is a constant above the 64 bit restriction? 13:28:07 q+ 13:28:10 ianh: we should discuss that later 13:28:30 q? 13:28:47 several people: "owl:real" is better 13:28:48 zakim, unmute me 13:28:48 bparsia should no longer be muted 13:28:53 q? 13:28:56 alanr: a datatype range greater that the max integer is still statisfiable 13:28:57 ack bparsia 13:29:20 zakim, mute me 13:29:20 bparsia should now be muted 13:29:47 alanr: let make the datatype document easily searchable for the key word conformance 13:30:19 ianh: xsd:float and xsd:double value space are now discrete 13:30:39 q? 13:30:55 alanr: how do we decide between XSD & IEEE? 13:31:07 ... for float and double 13:31:33 I prefer IEEE in general 13:32:01 m_schnei: NaN is not comparable to anything else 13:32:10 There are a lot of NaN. And for any NaN and any other float/double NaN != that number, including itself 13:32:47 msmith: xsd:float has a single lexical representation 13:33:09 q? 13:33:14 m_schnei: NaN is not comparable with itself! 13:33:39 alanr: you should also consider the need for counting float elements 13:33:55 msmith: can reasoning introduce NaN? 13:34:32 boris: let's go with XSD 13:34:58 ianh: everybody seems to agree with going with XSD 13:35:10 q+ 13:35:26 alanr: is it useful to have xsd:decimal as a type? 13:35:46 q- 13:35:46 msmith: it gives you arbitrary precision decimal number 13:35:55 What's the question? 13:36:07 1/10 is a decimal but not a binary number (e.g., float) 13:36:10 alanr: how do i test that something is a decimal number 13:36:19 I can't understand alan rut 13:36:31 What's the question? 13:36:39 http://www.w3.org/TR/xmlschema11-2/#decimal 13:36:40 pfps: XSD has a well answer to that question. please read the spec? 13:36:46 I can't udnerstand alan at all 13:36:48 s/?/ 13:36:52 q? 13:36:53 zakim, unmute me 13:36:54 bparsia should no longer be muted 13:36:59 q? 13:36:59 q+ 13:37:33 alanr: i'll read xsd spec to get the answer to my question 13:37:59 alanr: my pb is bt real and decimal 13:38:41 pfps: decimal= integer * 10^integer 13:39:02 I still didn't catch that from alan 13:39:17 It would help if he slowed down, becuase thphone seems to cut out on soft bits 13:39:19 Alan: Is it required that every datatype for serialization is also a datatype for range restrictions? I don't think so. 13:39:58 m_schnei: I can approximate as precisely as wish sqrt(2) using decimal but not with float or double 13:40:13 q? 13:40:20 But the way, I wasn't intending those searchs as determinative 13:40:29 I was just gathering some evidence 13:40:31 boris: we should allow it as a legacy datatype 13:40:41 +1 to boris 13:40:49 q+ 13:40:53 alanr: that's agood argument. thanks! 13:40:57 zakim, unmute me 13:40:59 bparsia was not muted, bparsia 13:41:00 ack bijan 13:41:06 ack bparsia 13:42:23 bijan I did not get your point, could you type it? 13:43:09 alanr: there is not oneon one mapping between decimal and real 13:43:33 I'm worried about things like (not sure that is a correct case) 0.1^^xsd:float != 0.1^^xsd:decimal 13:43:53 q? 13:44:04 alanr: decimal as a datatype is a synonym for real 13:44:29 -1 to decimal as a synonym for real 13:44:39 zhe: do we need to specify a minimal conformance for decimal 13:44:47 decimal isn't even a synonym for rational :) 13:44:48 boris: there is such restriction in XSD 13:45:25 Integer, rational, decimal etc. 13:45:33 Algebraic reals are denumeral 13:45:37 Only the transcendental 13:45:47 +1 evan 13:45:51 q? 13:45:55 Alan Rector was really glad about decimal 13:45:57 ekw: decimal is the most useful datetype in XSD 13:46:34 alanr: I'm ok with keeping deciimal in the spec 13:46:44 zhe and others interested in conformance, new page itemizing what needs to be addressed at http://www.w3.org/2007/OWL/wiki/ConformanceIssues 13:47:19 m_schnei: I do not understand why xsd:decimal is owl:real 13:47:33 zakim, mute me 13:47:33 bparsia should now be muted 13:47:36 ianh: we have discussed it and decided to leave them separate 13:48:01 q? 13:49:04 alanr: for base64Binary, should will allow length, minLength, maxLength? 13:49:40 m_schnei: why these datatypes (base64Bin, hexBin)? 13:49:44 boris: why not? 13:49:54 Wouldn't it make sense to make a lot of this exotic stuff optional? 13:49:58 q? 13:49:59 m_schnei: why not everything from xsd? 13:50:21 q+ 13:50:28 alanr: we are trying to get an well agreed list of acceptable datatype 13:50:28 zakim, unmute me 13:50:28 bparsia should no longer be muted 13:50:29 q? 13:50:30 Seems bad to *force* implementors to have datatypes like base64bin 13:50:44 q? 13:50:49 ack bparsia 13:51:09 bijan: the most important thing is to make sure that we say something meaningful about each datatype 13:51:21 ... about its usage 13:51:42 zakim, mute me 13:51:42 bparsia should now be muted 13:51:50 ... in swoogle I found a couple of these datatypes, but I need to investigate further 13:52:08 +1 to Carsten (I think...) 13:52:33 13:52:33 13:52:33 13:52:34 13:52:34 boris: i think it is not difficult to implement these types. All you need is to find the number of element in a given range 13:52:34 13:52:34 13:52:36 13:52:38 13:52:43 q? 13:52:49 boris: it is almost identical to string 13:52:58 ...this finiteness could still cause performance problems, but i guess having a warning about this could suffice? 13:52:59 I'm ok in forcing implementmation if it's relatively trivial 13:53:07 q? 13:53:32 alanr: what concern from Carsten, but everybody else seems to agree to support them 13:53:48 Carsten: Seems bad to *force* implementors to have datatypes like base64bin 13:53:49 +1 to boris 13:54:00 My proposal was to make the exotic stuff optional 13:54:25 boris: the implementation overhead is almost identical for string. So if we have string, we can also have binary datatype 13:54:25 q? 13:54:31 optional = incompatible 13:55:02 q+ 13:55:10 zakim, unmute me 13:55:10 Carsten should no longer be muted 13:55:12 q? 13:55:14 ianh: we have decided to avoid optional datatype 13:55:14 carsten, this is easy stuff 13:55:34 carsten: I do not care so much about whether this is optional 13:55:49 ... it seems to me that these types are of limited use 13:56:20 ... i am not sure that the only important point is the ability to count elements in a range 13:56:42 zakim, mute me 13:56:42 Carsten should now be muted 13:56:46 Carsten, I think this is a moot discussion 13:56:55 ... for completeness we may have to also check that a range does not contain a given element 13:57:13 q? 13:57:20 q- 13:57:20 boris: you can easily map all the float to integer 13:57:21 ack Carsten 13:57:27 zakim, mute me 13:57:27 Carsten should now be muted 13:57:49 ok, thanks for the explanation, I have to admit that it is not easy to follow the discussion remotely 13:57:53 boris: the complexity is not worse than for integer 13:57:54 q? 13:58:00 an email summing up would be nice, indeed 13:58:13 ... i can send u code to get the next float and the number of float in a range 13:58:14 See: http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm 13:58:26 ... I have redrew my objection 13:58:34 In particular: http://www.cygnus-software.com/papers/comparingfloats/Comparing%20floating%20point%20numbers.htm#_Toc135149455 13:58:37 ianh: we are don't with numeric 13:59:54 There's a standard one for iSO 14:00:00 ianh: yesterday , we agree to support xsd:dateTime with timezone 14:00:17 As a point of interest, ISO 8601 fixes a reference calendar date to the Gregorian calendar of 1875-05-20 as the date the Convention du Mètre was signed in Paris 14:00:31 pfps: datetime is dense so all the problem wih counting 14:02:13 evan: do we support leap seconds? 14:02:25 q? 14:02:37 pfps: if we don't do arithmetics, it doesn't matter, since the ordering isn't hurt by leaps 14:03:35 ianh: xsd:dateTime facet: length, minLength,maxLength 14:04:31 q? 14:05:23 q? 14:05:36 alanr: why not just create a new datatype owl:dateTime? 14:05:46 ... which will require timezone 14:06:21 +1 to xsd:datetime 14:06:37 +178 to xsd:datetime 14:06:46 I would call it a conformance thing 14:07:38 ...I very much like Ian's suggestion about the 'may repair al gusto'! 14:07:48 ianh: we have already decided to let application do they own repair 14:07:56 q? 14:08:42 q? 14:09:03 boris: xsd makes an error here by saying that it is isomorphic to the timeline 14:09:08 -1000 to tuple model 14:09:32 ianh: let's not reconsider our decision made yesterday about mandating timezone 14:09:54 q? 14:09:57 Btw, both generating and consuming tools can offer corrections and repairs 14:10:16 q+ 14:10:24 q- 14:10:39 m_schnei: in practice, it is not a pb because any application that accept non-timezone has to come up with some strategy to deal with them 14:10:45 q? 14:11:16 ianh: that is precisely the point we cannot mandate any behavior to all apps 14:11:37 sandro: we should talk to the XSD 1.1 WG what we think is wrong with non-timezoned dateTimeS 14:11:42 sandro: we should make a formal comment to XSD 1.1 working group 14:12:07 pfps: I'll send them a forml comment 14:12:14 ACTION: pfps to draft a comment on XML Schema Datatypes 1.1 draft 14:12:14 Sorry, couldn't find user - pfps 14:12:27 ACTION: peter to draft a comment on XML Schema Datatypes 1.1 draft 14:12:27 Sorry, amibiguous username (more than one match) - peter 14:12:27 Try using a different identifier, such as family name or username (eg. ppatelsc, phaase) 14:12:39 ACTION: ppatelsc to draft a comment on XML Schema Datatypes 1.1 draft 14:12:39 Created ACTION-176 - Draft a comment on XML Schema Datatypes 1.1 draft [on Peter Patel-Schneider - due 2008-08-05]. 14:12:41 ianh: internally we require timezone, but let applications do the right repair 14:13:42 I would also suggest that we document different repairs. Whether on the owl wg wiki or the owled one 14:13:53 q? 14:14:51 q? 14:15:12 Achille has joined #owl 14:15:24 Zhe has joined #owl 14:15:48 straw poll: should we use XSD dateTime, or our own dateTime datatype 14:16:01 baojie has joined #owl 14:16:27 ...too general/complex isn't bad? 14:16:49 q? 14:17:02 pfps: the comparison bt datetime with and w/o timezone is not feasable in our specification 14:17:03 straw poll: should we use XSD dateTime with a requirement for timezone component in text of spec 14:17:06 STRAWPOLL: Should we use owl:dateTime instead of xsd:dateTime (+1 = owl:dateTime, -1 = xsd:dateTime) 14:17:19 +1 14:17:23 +1 14:17:26 -1 14:17:27 0 14:17:34 0 14:17:37 -1 to xsd:dateTime 14:17:40 er 14:17:44 -1 to owl:dateTime 14:17:47 +1 14:17:52 Alan: to be clear, we'll still accept xsd:dateTime on input 14:17:57 using boris' poll 14:17:59 -1 14:18:03 -0 14:18:33 zakim, unmute me 14:18:33 bparsia should no longer be muted 14:18:37 +1 14:18:38 0 (+/- 1) 14:18:44 bijan:yes I want xsd:dateTime 14:19:06 ianh: we almost have a split decision 14:20:13 q? 14:20:24 Not just doable, but acceptible 14:20:28 alanr: we have agreed on xsd:dateTime with required timeZone 14:20:35 I don't find, e.g., an interval to be a user-sensible default 14:20:53 -1 14:21:32 alanr: in other case we tried to keep xsd semantics 14:21:43 s/other case/ other cases/ 14:21:47 will owl:dateTime require a timezone? 14:21:55 yes 14:22:26 q? 14:22:57 ekw: I am not comfortable by XSD 1.1 new semantics on dateTime 14:23:33 I would prefer that we calledit xsd:dateTime and said that we add additional restrictions (the way implemetnation might only support 64 bit integers) 14:23:36 ianh: we defeer the issue of xsd:dateTime vs owl:dateTime for the next meeting 14:25:23 all the details on the decision made w.r.t datatype are posted at http://lists.w3.org/Archives/Public/public-owl-wg/2008Jul/0433.html 14:25:40 s/decision/decisions 14:25:50 PROPOSED: We'll handle the datatypes named in http://lists.w3.org/Archives/Public/public-owl-wg/2008Jul/0433.html as described in that message. This closes ISSUE-126. 14:25:57 +1 Oxford 14:26:02 q? 14:26:04 +1 Science Commons 14:26:10 +1 W3C 14:26:12 +1 IBM 14:26:13 +1 (FZI) 14:26:15 +1 RPI 14:26:17 +1 C&P 14:26:18 +1 NIST 14:26:29 +1 ORACLE 14:26:32 +1 manchester 14:26:43 RESOLVED: We'll handle the datatypes named in http://lists.w3.org/Archives/Public/public-owl-wg/2008Jul/0433.html as described in that message. This closes ISSUE-126. 14:27:48 ianh: we agreed yesterday to make issue 132 an editorial issue 14:27:59 s/ianh/boris 14:28:17 topic: N-ary datatype 14:28:30 ianh: a proposal is on the wiki 14:28:55 q+ 14:29:00 q- 14:29:03 ianh: there is compromise to make it optional, 14:29:23 q? 14:29:27 ... if you decide to support it we specify exactly how it should be implemented 14:29:37 q+ 14:29:55 alanr: there are interop issues here 14:30:07 ? 14:30:33 Hey! 14:30:37 zakim, unmute me 14:30:37 bparsia was not muted, bparsia 14:30:55 q? 14:30:58 alanr: when you use that aspect of the spec it is unlikely that you will be portable 14:31:00 ack bparsia 14:31:03 maybe 14:31:08 not clear 14:31:48 q? 14:32:08 +1 to bijan. There can be OWL DL conformance and OWL Nary datatype conformance as separate things 14:32:08 I never use racer because of nominals 14:32:15 bijan: i don't see why this is a pb. it is hard to me to imagine that someone using an equation would not be aware that he is doing something hard for implementation 14:32:23 consider it not to support OWL 14:32:29 zakim, mute me 14:32:29 bparsia should now be muted 14:32:41 ... this is just one of many things that tools will not support 14:32:53 That'll be ture 14:33:06 And it won't include linear 14:33:07 alanr: I'm still concern about interoperability 14:33:13 q+ 14:33:25 zakim, unmute me 14:33:25 bparsia should no longer be muted 14:33:54 q? 14:33:55 msmith: this goes to the conformance discussion 14:34:10 m_schnei: thinks that typical name for this is "a standard extension" 14:34:12 ... we will have similar conformance isssue with profile. 14:34:13 ack bparsia 14:34:23 that was a separate q! 14:34:25 q+ 14:34:29 ... it is just another facet of this problem of conformance 14:34:37 ack bparisa 14:34:41 q? 14:34:42 jar has joined #owl 14:34:46 ack bparsia 14:35:01 bijan: i really don't understand why putting it in a different document that solve the problem 14:35:48 bijan: species validation will help you understand whether you stand w.r.t. portability 14:36:01 q? 14:36:01 expressive checking in general 14:36:13 If you have nominals, you can't use KAON2 14:36:51 q? 14:36:53 What's the question? 14:37:15 ianh: will there be a specific proposal about N-ary ? 14:37:36 yes, I do 14:37:49 bijan: it will be an extensibility point 14:38:05 ... we will have a specific proposal 14:38:15 ... in a separate spec 14:38:33