08:07 *** Log started on 2000-02-25 UTC-0500 *** 08:08 [halindrom] This is Shane 08:09 [Ralph] *** Ralph (swick@seahorse.w3.org) has joined channel #w3c 08:09 [Ralph] Hi, Daniel and Shane. Session starts in 51 minutes. 08:09 [DV_] ok 08:09 [DV_] actually I have this channel always open 08:10 [Ralph] I don't have enough [mental] bandwidth to keep channels perpetually open :-) 08:11 [Ralph] XHTML session announcement: http://lists.w3.org/Archives/Public/www-html/2000Feb/0203.html 08:11 [Ralph] (for which this is the irc log) 08:12 [Ralph] ( this == http://www.w3.org/2000/02/25-xhtml-irc ) 08:26 [Yves] *** Yves (ylafon@tarantula.inria.fr) has joined channel #w3c 08:27 [Yves] \adduser ua *@*.w3.org -1 #w3c 08:27 [Yves] \adduser ua DV_ -1 #w3c 08:27 [Yves] \adduser ua Ralph -1 #w3c 08:27 [Yves] \saveall 08:27 [Yves] \whois ralph 08:28 [Yves] ralph: leave/join :) 08:28 [Ralph] *** Ralph (swick@seahorse.w3.org) has left channel #w3c 08:28 [Ralph] *** Ralph (swick@seahorse.w3.org) has joined channel #w3c 08:29 [DV_] *** DV_ (veillard@tux.inrialpes.fr) has joined channel #w3c 08:29 [Yves] \me 08:30 [Yves] \whois dv_ 08:30 [Yves] \adduser ua dv_ -1 #w3c 08:30 [Yves] \whois dv_ 08:30 [Yves] \saveall 08:31 [Yves] \op * 08:35 [Ralph] previous irc session: http://www.w3.org/1999/12/w3c-irc2409 08:48 [StevenPemberton] *** StevenPemberton (steven.pem@pembo.cwi.nl) has joined channel #w3c 08:48 [StevenPemberton] # Appears as LANCE. 08:48 [StevenPemberton] Salve 08:51 [eric] *** eric (eric@quake.w3.org) has joined channel #w3c 08:53 [Shane] Morning Steven 08:53 [trj] *** trj (tjaffey@sourceforge.net) has joined channel #w3c 08:53 [StevenPemberton] Good afternoon 08:53 [trj] ullo 08:56 [Ian] *** Ian (ian@adsl-151-202-33-119.bellatlantic.net) has joined channel #w3c 08:57 [Loggy_] *** Loggy_ (loggy@koala.inria.fr) has joined channel #w3c 08:58 [eric] hi loggy, welcome 08:58 [DV_] hi all 08:58 [Ian] Hello from NYC. 08:58 [Loggy_] hi eric ;) 08:58 [StevenPemberton] There are the chimes for the hour 08:59 [Shane] How pleasant - chimes. 08:59 [Shane] I have screaming children. 08:59 [DanC] *** DanC (connolly@cs2736-248.austin.rr.com) has joined channel #w3c 08:59 [StevenPemberton] Aha, do they scream on the hour? 08:59 [Shane] statistically, yes. 09:00 [StevenPemberton] Hi Dan 09:00 [DanC] Hi. This is Dan Connolly, W3C, writing from Austin, Texas. 09:00 [DanC] Everyone is invited to introduce themselves likewise. 09:00 [Shane] Hi. This is Shane McCarron, ApTest, writing from Minneapolis, Minnesota. 09:01 [DV_] Daniel Veillard, w3C from Grenoble, France 09:01 [eric] Eric Prud'hommeaux, W3C, writing from Cambridge, MA 09:01 [StevenPemberton] This is Steven Pemberton, Chair of the W3C HTML Working Group, in Amsterdam, The Netherlands 09:01 [Ralph] Ralph Swick, W3C, Cambridge MA 09:01 [Ralph] [[[ /who #w3c 09:01 [Ralph] #w3c DanC H connolly@cs2736-248.austin.rr.com :3 Dan Connolly 09:01 [Ralph] #w3c Loggy_ H@ loggy@koala.inria.fr :2 Loggin' service 09:01 [Ralph] #w3c Ian H ian@adsl-151-202-33-119.bellatlantic.net :0 ian 09:01 [Ralph] #w3c eric H eric@quake.w3.org :3 Eric Prud'hommeaux 09:01 [Ralph] #w3c StevenPemberton H steven.pem@pembo.cwi.nl :2 Steven Pemberton 09:01 [Ralph] #w3c DV_ H@ veillard@tux.inrialpes.fr :2 veillard 09:01 [Ralph] #w3c Ralph H@ swick@seahorse.w3.org :0 Ralph R. Swick 09:01 [Ralph] #w3c Yves H@ ylafon@tarantula.inria.fr :1 The Fool on The Hill 09:01 [Ralph] #w3c RRSAgent H@ swick@seahorse.w3.org :2 Ralph R. Swick 09:02 [Ralph] #w3c Shane H@ ahby@spm.themacs.com :0 shane@themacs.com 09:02 [Ralph] #w3c End of /WHO list. 09:02 [Ralph] ]]] 09:02 [DV_] Yep basically W3C and WG members, right ? 09:03 [DanC] um, Ralph, take care with that /who thing... folks haven't volunteered to put their email addresses in the log. 09:03 [SusanL] *** SusanL (lesch@24-25-222-186.san.rr.com) has joined channel #w3c 09:03 [Ralph] oh, sorry folks 09:03 [timbl] *** timbl (timbl@18.23.10.166) has joined channel #w3c 09:03 [StevenPemberton] Welcome Susan. Do you want to introduce yourself? 09:03 [mimasa] *** mimasa (mimasa@133.27.195.235) has joined channel #w3c 09:04 [eric] the email address does show up in the join action message 09:04 [SusanL] Hello. Susan Lesch, San Diego, USA 09:04 [DV_] Ralph I understand that anybody can get those informations 09:04 [DV_] Hi Susan 09:04 [Ralph] DV, anyone that's here now. Not in the future, however. 09:04 [Shane] Do we have an agenda for this morning? 09:04 [StevenPemberton] Anyway, it's not a true email address. 09:05 [DanC] I've been noodling about XHTML and mixed-namespace documents. I had some fun last night: I used Amaya to save a mixed HTML/MathML document, and then tested the experimental support for such documents that we've added to the validator: http://validator.w3.org/check?uri=http%3A%2F%2Fthinker.ods.org%2Fxht-math.html 09:05 [kmacleod] *** kmacleod (ken@c941885-a.west1.ia.home.com) has joined channel #w3c 09:05 [DanC] Agenda: XHTML stuff, XML and protocols idea, general questions about how W3C works. Or something like that. 09:06 [DanC] I see a few new folks. I encourage you to introduce yourself: name, affiliation, city from which you're writing. 09:06 [Shane] DanC, which version of the XHTML DTDs were you using in your experiments? 09:06 [DanC] email address is optional. 09:06 [Michael-NM] *** Michael-NM (cmsmcq@206.206.93.62) has joined channel #w3c 09:06 [mimasa] Hello, I'm Masayasu Ishikawa, W3C 09:06 [DanC] no DTD in the experiment. Using DTDs with mixed-namespace documents is awkward. I expect to use XML Schemas for such things, mostly. 09:07 [Michael-NM] Good morning, I'm Michael Sperberg-McQueen, W3C. Greetings from northern New Mexico. 09:07 [DanC] You can use DTDs and namespaces, but one party has to assemble the DTD for all the parts. 09:07 [kmacleod] Ken MacLeod, independent, Des Moines, Iowa, USA; author of several Perl XML modules, the LDO messaging library, and the Perl implementation of XML-RPC 09:07 [StevenPemberton] So what does 'validate' mean w.r.t your test last night? 09:08 [DanC] test validate: just check for xml well-formedness and the XHTML namespace name, and list other namespaces. No XML schema support yet. 09:08 [Ian] Hello, I'm Ian Jacobs, W3C Editor and Communications Guy, in New York (USA). 09:09 [Shane] I think that we would be wise to avoid the validation rat hole this morning. 09:09 [DanC] replay details of experiment/test for folks that joined since: [08:05] <DanC> I've been noodling about XHTML and mixed-namespace documents. I had some fun last night: I used Amaya to save a mixed HTML/MathML document, and then tested the experimental support for such documents that we've added to the validator: http://validator.w3.org/check?uri=http%3A%2F%2Fthinker.ods.org%2Fxht-math.html 09:09 [DV_] Shane: too late ;-) 09:10 [DanC] Ken, in XML-RPC and such, is there much use of datatyping? ints, dates, floats, that sort of thing? 09:10 [DV_] Ok let's ask the question then ... how many opensource schema implemetations are on the way ? 09:11 [kmacleod] XML-RPC has a fixed handful of basic datatypes 09:11 [hugo] *** hugo (hugo@voodoo.w3.org) has joined channel #w3c 09:11 [DanC] pointer to that list, Ken? 09:11 [kmacleod] <http://www.xmlrpc.com/spec> 09:11 [DV_] DanC: you're not coming from w3.org 09:12 [eric] DanC: there are some types defined midway through http://www.xmlrpc.com/stories/storyReader$7 09:12 [hugo] (hi) 09:13 [DV_] Michael-NM: except Henry stuff any info on Schema implementations ? 09:14 [DanC] I'm pretty sure the IBM alphaworks folks are working on XML Schema support. I don't recall the details of the license. 09:14 [DV_] Okay 09:14 [Michael-NM] IBM's work is part of AlphaWorks, i.e. open source. 09:15 [Shane] AlphaWorks is not necessarily open source 09:15 [Michael-NM] I believe parts of it are showing up in their contributions to Apache, but I didn't catch all the details. 09:15 [DanC] As you mentioned, DV, Henry's group (http://www.ltg.ed.ac.uk/) is working on one (which should show up in http://dev.w3.org/ eventually) 09:15 [Shane] Most projects have restricted use licenses, and many are not available in source form. 09:15 [DV_] DanC: http://dev.w3.org/cvsweb/xmlschema/ more precisely 09:16 [kmacleod] LDO includes a spec that is similar to XML-RPC but takes a different approach. We still use a fixed set of elements (<dictionary>, <list>, <atom>) but each element has a `type' attribute. The `type' attribute can use any of several schemes for def. types 09:16 [kmacleod] <http://casbah.org/LDO/> 09:16 [Michael-NM] I believe IBM's work is on an open-source schema processor, with non-open-source adaptations to their products. 09:16 [DanC] background: Chat Topic: LDO - an XML transport library Ken MacLeod (Sat, Feb 19 2000) http://lists.w3.org/Archives/Public/www-html/2000Feb/0193.html 09:19 [timbl] (Tim Berners-Lee, W3C, Eastern Mass) Interesting connection between layers and namespaces occured to me yesterday looking at various XML protocols. 09:19 [DanC] Michael, have you looked at http://www.xmlrpc.com/spec? The scalar types map pretty straightforwardly to XML Schema primitive datatypes, but I wonder about structs and arrays. 09:19 [eric] It seems like the layering and multiplexing described in http://lists.w3.org/Archives/Public/www-html/2000Feb/0193.html could be accomplished with namespaces 09:20 [timbl] Just like an ethernet packet has bits at eth ends for each layer, so many layers can be in XML 09:20 [timbl] The advantage of it is you can use a "layer" at any stage more easily if you use XML 09:20 [eric] The main thing that would settled is what you do if you encounter a anmespace you recognize inside one you don't recognize 09:21 [Michael-NM] The RPC scalars map readily, as you say. (Though 'ASCII string' is kind of a shock to me.) 09:22 [eric] I guess XMLprotocol is germain here only where one encapsulates XML-RPC or similar data inside XHTML 09:22 [eric] (and thereby blows up the DTD) 09:22 [dcleary] *** dcleary (dcleary@psc.progress.com) has joined channel #w3c 09:22 [Michael-NM] Arrays look very straightforward, though: they work just as the XML Schema group keeps telling people they should work. 09:22 [kmacleod] Namespaces are good when you have "middleware", agents in the middle that perform routing, store-and-forward, etc. 09:22 [Shane] Namespaces are also useful for data transformation ala IBM's WBI 09:22 [Michael-NM] Structs, on the other hand, will not map easily or directly, because they require (or appear to require) context-sensitive information about the position of hte member. 09:22 [DanC] WBI? pointer? 09:23 [Shane] Working... 09:23 [kmacleod] Most of our work to date in LDO has been point-to-point and we haven't touched namespaces yet. SOAP uses namespaces. 09:23 [eric] i beleive a lot of the modern protocol serialization language use names parameters 09:23 [eric] this obviates the need for structure ordering 09:24 [eric] s/names/named/ 09:24 [Michael-NM] Eric - Right, it blows up the DTD to consider the XML content of a wrapper as a black box. 09:24 [Shane] WBI: WBI Development kit for Java: http://www.alphaworks.ibm.com 09:24 [DanC] I discovered a bunch of traffic on html-tidy (http://lists.w3.org/Archives/Public/html-tidy/) the other day... including discussion of a port of tidy to Java (http://www3.sympatico.ca/ac.quick/jtidy.html). 09:24 [Michael-NM] But that's where XML Schema comes in; you can define a schema for the wrapper layer which says, effectively, "Payload goes here, and don't check anything except that it's WF XML." 09:25 [eric] ken: the SOAP debugger doesn't use namespaces, are there engines that do? 09:25 [kmacleod] I don't know, I got that info from the spec 09:25 [timbl] "Payload goes here, and I define what goes her as a FOO and you can declare your widget to be a subclass of FOO" 09:26 [Michael-NM] That is also possible: the payload can be a white box, not a black box. 09:26 [Michael-NM] Schema-designer's choice. 09:27 [Michael-NM] eric, ken - what does 'using namespaces' mean in a debugger? 09:28 [eric] the debugger (which i have in some window here) is a little CGI script where you submit a SOAP request and get back an answer 09:28 [eric] the example query and the replies don't use namespaces. 09:28 [DanC] I actually do have some XHTML stuff that I'd like to know about, but xml-and-protcols seems to be using up all the bandwidth here... 09:28 [DanC] So I've created #xml-dist-app ; please continue the protocols discussion there. 09:28 [eric] roger 09:28 [Michael-NM] XML and protocols will expand to fill all available bandwidth, and brainwidth. 09:28 [Shane] Well - I am happy to discuss XHTML - let's get to ut. 09:28 [Shane] s/ut/it/ 09:29 [mimasa] Ditto. 09:29 [DanC] Steven, I had some questions about mixing DTD bits using XHTML modularization, and you said you'd send details. I don't recall getting them, but perhaps you could explain now? 09:29 [StevenPemberton] Could you elaborate your question? 09:30 [DanC] Steven, see http://www.w3.org/1999/11/chairs298#ac2307 [other folks: sorry, member-confidential] 09:31 [DanC] my question back in Nov was: [[ DanC: does your DTD-based modularization support decentralization? 09:31 [DanC] ... can someone else add a module? 09:31 [DanC] ]] 09:31 [Shane] Absolutely. 09:31 [Shane] Anyone can add modules and create new, hybrid document types. 09:31 [DanC] could we walk thru an example, please? 09:31 [Shane] That is the whole point. 09:31 [Shane] There are detailed examples in the public draft - let's try one. 09:32 [Shane] Can we ignore the namespace magic for the sake of simplicity? 09:32 [DanC] I'd rather not ignore the namespace magic. 09:32 [Shane] Okay - that makes the example harder to understand, but... 09:32 [DanC] but if you'll cover it in a 2nd example, I could live without it for the 1st example. 09:32 [StevenPemberton] In brief: all along we have tested adding other modules. Murray in particular seems to have dozens of DTDs with different modules. 09:33 [Shane] Fair enough. Let's do a simple example. 09:33 [Shane] You create a module that has two new elements in it: 09:33 [DanC] But Murray wrote the XHTML DTDs. That's not decentralization ;-) 09:33 [Shane] <myelement> 09:33 [DanC] Shane, pls point to the part of the spec that explains this, while you're at it. 09:33 [StevenPemberton] Yes, but Math WG did it too. 09:33 [Shane] and <myotherelement> 09:33 [StevenPemberton] And I18N 09:33 [DanC] in stead of <myelement>, please let's be concrete: let's use catalog markup: <item>, <description>, <price> 09:34 [dcleary] *** dcleary (dcleary@psc.progress.com) has left channel #w3c 09:34 [mimasa] Yeah, I wrote Ruby DTD. 09:34 [DanC] Steven, pointer to Math WG's module? 09:34 [DanC] Mimasa, pointer? 09:34 [mimasa] See the latest draft. 09:34 [DanC] "latest draft"? please assume I know nothing about W3C, XHTML, etc. 09:35 [mimasa] For Math: http://www.w3.org/TR/2000/WD-MathML2-20000211 09:35 [Shane] Let's not quibble. The part of the public spec that defines this is at http://www.w3.org/TR/xhtml-building/developing.html 09:35 [Michael-NM] *** Michael-NM (cmsmcq@206.206.93.62) has joined channel #w3c 09:35 [DanC] I'm not trying to quibble. I'm trying to make the archive of this discussion useful to other folks. 09:35 [Shane] Please note that this has now been folded back into the modularization draft, but for purposes of discussion. 09:35 [Shane] Okay - concrete example: 09:36 [Shane] An <item> has a content model with two elements, <description> and <price> 09:36 [Shane] These are defined in a DTD module. 09:36 [Shane] The author must create a new "DTD driver" to bring in this module. 09:36 [DanC] mimasa, could you point to the specific part of the MathML spec where the module is given? 09:36 [Shane] The MathML spec that is modularized is not yet public, I don't think. 09:37 [mimasa] http://www.w3.org/TR/2000/WD-MathML2-20000211/appendixa.html#parsing:dtd 09:37 [DanC] sigh: "The full DTD is available as a zip archive." 09:37 [mimasa] But you can use MathML itself as a module. 09:38 [Shane] In this example, the <item> element needs to work anywhere in the DTD. 09:38 [DanC] yes, Shane, I'm following the example too. pls continue. 09:39 [Michael-NM] I'm following, too. Pls continue. 09:39 [DanC] Are the author (i.e. the DTD driver writer) and the DTD writer the same party? or may they be different parties? 09:39 [DanC] s/DTD writer/DTD module writer/ 09:39 [Shane] So in the DTD driver, you define a PE that inserts the <item> element into the Flow content model. 09:39 [Shane] Good question - doesn't matter. 09:40 [Shane] We envision both uses 09:40 [DV_] Probably stupid question: what's the default encoding of XHTmL document ? UTF-8, right ? 09:40 [DanC] "doesn't matter" -- so they may be different parties. Good. 09:40 [Shane] Yes, it can be UTF-8. 09:40 [Shane] sorry - default. Default is UTF-8. 09:40 [DanC] encoding: I'm pretty sure XHTML completely defers to XML 1.0 on encoding details. 09:40 [DV_] Shane: ok, thanks 09:41 [Shane] Next, you instantiate your new module. 09:41 [Lucio] *** Lucio (picci@picci.spbo.unibo.it) has joined channel #w3c 09:41 [Shane] Finally, you hand off processing to an existing modular DTD driver (like XHTML Basic). 09:41 [Shane] These 4 steps will get you a new document type that includes your new element and its content model. 09:42 [Shane] Of course, this simple example ignores the namespace issue completely. 09:42 [Michael-NM] The unpleasant part, using DTDs and parameter entities, is that the module writer and the caller both have to do a lot of manual name wrangling. 09:43 [Shane] What do you mean by "manual name wrangling"? 09:43 [DanC] manual name wrangling: quite. e.g. nobody else can use the parameter entity name mathml-prefix 09:43 [Michael-NM] To allow complete independence of modules, you really want the set of parameter entities in each module to be disjoint from the set in every other module. 09:43 [Shane] exactly. In the document we require that modules use unique prefixes for their PEs 09:44 [Michael-NM] Good - I obviously need to read the current version; I'm behind! 09:44 [Shane] So the MyML module set would use MYML.element.qname, for example. 09:44 [Shane] Do you want to talk about namespace magic now? 09:44 [DanC] that is: if the Open Catalog Consortium defines %catalog-prefix and the Open Puchase Order Consortium uses the name %catalog-prefix, I can't use both of their modules together, right? 09:44 [DanC] s/uses/defines/ 09:44 [Shane] That's correct. It would fail. 09:45 [Shane] Of course, you would know it failed pretty much right away ;-) 09:45 [Michael-NM] Right. So perhaps %occ.org.catalog.prefix 09:45 [DanC] er... so how is this decentralized? 09:45 [Michael-NM] and %opoc.org.catalog.prefix .. I.e. use your domain name. 09:45 [Ralph] %org.occ.catalog.prefix 09:45 [Shane] What do you mean? Because there is the potential for collisions? 09:45 [Michael-NM] Dan, think of it as an optimistic algorithm. 09:46 [Shane] It is as decentralized as creating your own C language library and linking it with someone else's. 09:46 [DanC] yes! because there's the potential for collisions. If it's decentralized, any two parties should be able to independently develop modules, and a third party should be able to use them together, no? 09:46 [Ralph] optimism doesn't scale to Web-sized names :-) 09:46 [DanC] analogy with C: exactly! 09:46 [StevenPemberton] You're only creating the DTD once; it isn't like with XML documents that they are being created on the fly. 09:46 [Michael-NM] Even namespaces don't do that, because they don't actually say that you are only allowed to define namespace URIs in domains you own. 09:46 [Shane] Absolutely, Dan. You can combine them arbitrarily. If there are name collisions, some one has to resolve them. 09:47 [DanC] yes, each of the Open Catalog Consortium and the Open Purchase Order consortium makes up their DTD once. But they didn't see each other's work. So they didn't keep their names distinct. So I loose if I try to use their work together. 09:47 [Shane] We could require that you use domain names - that's not a bad suggestion, actually. 09:47 [Shane] All we say is that it must be unique. 09:47 [Shane] Remember that humans are creating these hybrid DTDs, and humans can work to resolve conflicts. 09:47 [eric] i think that saying it must be unique is stront enough 09:47 [Shane] thanks, erc. 09:48 [eric] folks could use some reg service if they don't have thir onw domain 09:48 [DanC] how can I, a human, resolve the conflict between the catalog consortium's DTD and the purchase order consortium's DTD? 09:48 [Ralph] I'm a human creating a hybrid document that wants to use modules written by other humans, neither of whom anticipated my need 09:48 [Shane] By grabbing the DTDs and massaging them. 09:48 [Shane] We had to do this with the MathML DTD early on, because there was a conflict. 09:48 [Shane] Now the Math Working Group is fixing this. 09:49 [timbl] Michael: Namepsaces don't say you cant dfeine URIs in someone else's space but the URI spec (eg for HTTP) does! 09:49 [timbl] Michael: The fact that there is delegation of authority in DNS is a property of the URI not the spec using it. 09:49 [Ralph] if I "massage" someone else's DTD then how can I declare to the world that I'm reallytruly using their definitions just with naming changes? sounds too optimistic to me! 09:49 [StevenPemberton] Let's make this clear; Modularisation isn't there to replace Namespaces! It's there to allow us to build a small number of DTDs that are in the works, guearanteeing that they are consistent with each other 09:49 [eric] are we talking about namespace uniqueness or FPIs in DTDs? 09:49 [DanC] This (resolving conflicts in MathML) is HTML tag wars all over again, no? It's clearly not decentralized. 09:50 [Shane] Not true Dan. 09:50 [Michael-NM] So it's an error if I 'define' a URI in, say, stanford.edu. 09:50 [Michael-NM] How is this error detected? 09:51 [StevenPemberton] The Hypertext Coordination Group was of the opinion that we need something like this for DTDs. Once Schemas are up and running, Modularisation will fade away. 09:51 [Shane] Tag wars are alleviated by XML Namespaces and qualified names. 09:51 [Shane] The issue here is one of DTD parameter entity collisions. These can be avoided by sensible people doing sensible things. 09:51 [DanC] Shane, I've given a definition of decentralized (ability of party C to combine work of parties A and B even though A and B didn't see each other's work), and you agreed XHTML modularization doesn't meet it; you've said party C has to manually re-engineer A and B's work. 09:51 [Michael-NM] Yes, indeed. And the distinct-prefix-in-PEs rule is pretty much all you really need to make relatively static DTDs that allow combinations of modules. 09:52 [StevenPemberton] And modularisation allows you to use namespaces; REQUIRES you to use namespaces if you define a new module. 09:52 [Shane] No, I did not say that. 09:52 [Shane] I said that if the party A and party B did not follow the rules, party C might need to re-engineer it. 09:52 [Shane] That's why we have rules. 09:52 [DanC] then I misunderstood, Shane. What do I do about the use of the same PE name by catalog.org and purchaseorder.org? I don't have the right to create derivitave works from their DTDs, by the way. 09:53 [Shane] XHTML Modularization only defines mechanisms in the presence of conforming modules. 09:53 [Shane] Modules that do not have unique PE naming are, by definition, non-conforming. 09:53 [Shane] I can't help you. 09:53 [DanC] what rules did catalog.org and purchaseorder.org break when they each chose %catalog-prefix as a PE name? 09:54 [Shane] catalog-prefix does not have an organization- unique prefix 09:54 [Shane] I think the output from this meeting is that such a prefix should be a requirement. 09:54 [DanC] that would perhaps be progress. 09:55 [DanC] it would be awkward, but such is life, I guess. 09:55 [StevenPemberton] I repeat: we need modularisation for our (HTML WG) own ends. If others want to use it, we specify the conditions under which it works. The important part of the Modularisation document are the modules. 09:55 [mimasa] Note that DTD is *an* implementation method of modularization, not the only one. 09:55 [Shane] Steven, I must disagree. 09:55 [StevenPemberton] The modularisation bit is just existing SGML technology. Nothing new. 09:56 [Michael-NM] The new bit is saying the conditions under which it works. 09:56 [Shane] Modularization is a generally useful mechanism for creating new document types. 09:56 [DanC] "for our own ends" --- that's tantamount to saying it's *not* decentralized. That says "you have to talk to us if you want to do another module." I could accept that answer, but I'm not getting that answer consistently. 09:56 [Shane] The DTD implementation has some clever bits in it, but modularization transcends this low level crap. 09:56 [Shane] That is not an answer you will get from the HTML Working Group. 09:56 [StevenPemberton] Sure, it's really useful, and HTML Writers Guild is using it for Proj. Gutenberg as I understand it. But it is a means to our ends, not the end in itself. 09:57 [Shane] Who is "our" in that context? 09:57 [StevenPemberton] our=html wg 09:57 [Michael-NM] OK, I have learned that I need to read your current draft. Along with ten other current drafts; will someone stop the calendar for a couple weeks for me, please? 09:58 [Shane] heh. 09:58 [DanC] Welcome to W3C, Michael! 1/2 ;-) 09:58 [Shane] Note that when we say "current" draft, we really mean the one that is member only. 09:58 [DV_] Michael-NM: 6 month for me after Xlink REC ! 09:58 [Shane] at http://www.w3.org/MarkUp/Group/2000/WD-xhtml-modularization-20000214/ 09:58 [DanC] current draft: do you expect to do another last call? Or can we just review the current public draft? 09:59 [Shane] We do not expect to do another last call. We have addressed all last call comments, and expect to proceed to PR shortly. 09:59 [Shane] I got the final DTDs last night from Murray. 09:59 [StevenPemberton] Modularisation is not new technology; it works on all existing XML parsers; it allows you to join bits together to make a DTD. 09:59 [Shane] Steven, exactly. The value is the rules that permit the ready combination of arbitrary modules. 09:59 [StevenPemberton] CUrrent draft is as modified following Last call comments 10:00 [DanC] er... if you're saying that you've changed the spec enough that reading the current public draft doesn't accomplish roughly the same thing as reading the current internal draft, then another last call is in order, no? 10:00 [Michael-NM] Is there a convenient way to get the whole draft onto my HD, for plane reading? 10:00 [Shane] Not according to process. 10:00 [StevenPemberton] Yes, there is a single file version, see top of doc 10:00 [Shane] Yes - it is in PDF, PS, zip, and tarball - help yourself. 10:00 [Michael-NM] Thanks. 10:00 [DanC] yes, according to process, your spec exits last call when it's reviewed by a large community and no substantial changes result. 10:01 [Ian] http://www.w3.org/Guide/CandidateRec 10:01 [Ian] (Member-only, sorry) 10:01 [Shane] Who defines substantial? 10:02 [Shane] Seriously, the reason I point to the internal draft is that we editorially combined two documents into one and improved the examples. 10:02 [Shane] It is easier to comprehend. 10:02 [Shane] We have made no substantive changes to the normative requirements, except that we clarified how namespace integration works. 10:02 [Michael-NM] i think the WG must define it, but it must be subject to appeals ... 10:03 [DanC] defines substantial: the Director, in response to the request for CR/PR. The WG chair can decide another last call is necessary on his own, of course. 10:03 [Shane] Steven, do you think we need another last call? 10:03 [Shane] puts Steven on the spot 10:03 [DanC] If you're telling me that reading the public last call spec is not enough, you're telling me you need to do last call again. 10:03 [Shane] Read two public specs - that's enough. 10:04 [DanC] very well. 10:04 [StevenPemberton] I don't see the need yet, depending on the waves that the style attribute are going to cause. 10:04 [Shane] Oh, that. 10:04 [Ian] Can you summarize the style attribute issue here? 10:04 [Shane] sure 10:04 [StevenPemberton] We dropped it. 10:04 [Ian] That's a pretty good summary. 10:04 [Ian] Why? 10:04 [Shane] The style attribute is in the legacy module. We didn't drop it. 10:04 [StevenPemberton] We dropped it long ago, but CSS have just noticed. 10:05 [Shane] The legacy module is not included in XHTML 1.1 DTD. 10:05 [StevenPemberton] We dropped it from XHTML 1.1 10:05 [mimasa] And from XHTML Basic. 10:05 [Ian] I guess I should reread the threads. 10:05 [Shane] So to use styles, you must use embedded style sheets with the <style> element, or link them with the <link> element. 10:06 [Ian] I don't think there's an accessibility issue with the style attribute, so I suppose there must be other reasons. 10:06 [Shane] The rationale is that the style attribute has no binding to a style grammar, and we need to accomodate any style grammar. 10:06 [Shane] There are other reasons, but that is my personal favorite. 10:06 [DanC] do you know if tidy supports changing <p style="xyz">...</p> to <style>i12: xyz</style> <p id=i12>...</p>? 10:06 [Shane] Another good one is that we should not require XHTML family user agents to have multiple parsers. 10:06 [StevenPemberton] Ian: the 'why' is long. 1) Styling is not part of XHTML. 2) If there is a need for a style attribute, it is needed in all XML, not just XHTML, and so should be in an XML or Style namespace; 3) it is not language independent (you cant have one for CSS, one for XSL, etc), 4).... 10:07 [mimasa] FYI, ISO-HTML also dropped style attribute. 10:07 [Ian] Thank you SP. 10:07 [DanC] style attribute in all of XML: whee! back to the stylesheet linking PI issue! 10:07 [mimasa] Exactly. 10:07 [Shane] The best proposal I have heard lately is that a CSS module be created that has a style attribute in its own namespace. 10:08 [Shane] Sortyof <p css:style="foo" xsl:style="otherfoo"> of you want. 10:08 [DanC] discussion of removal of style attr: e.g. Re: inline CSS - Argument for fantasai (Thu, Feb 24 2000) http://lists.w3.org/Archives/Public/www-html/2000Feb/0354.html 10:08 [Shane] s/of/if/ 10:08 [StevenPemberton] That sounds like the best route to me too (css;style=....) 10:08 [DanC] Steven, do you now if anybody's gone on record objecting to the removeal of style=? i.e. have they sent a change request to www-html-editor? 10:08 [StevenPemberton] But there is still an ongoing discussion of xml-like style attributes that are language independent. I like that too. 10:09 [Shane] I do not envision the HTML Working Group making any substantive changes as a result of the CSS Working Group's request. 10:09 [Shane] The CSS working group sent their comments to that list (I think). 10:09 [StevenPemberton] Dan: the official comment is rather weak, but I have heard that there are strong feelings. 10:09 [Ian] Some stuff from the CSS WG: 10:09 [Ian] http://lists.w3.org/Archives/Member/w3c-html-wg/2000JanMar/att-0291/02-NOTE-CSSXHTML.htm 10:09 [Shane] In the CSS working group, or elsewhere. 10:09 [Shane] There are lots of strong feelings about the style attribute, but not in the greater community as far as I can tell. 10:10 [DanC] Sorry, I have a teleconference now. You're welcome to continue, and I'll try to pay some attention, but note that I might be distracted. 10:10 [StevenPemberton] Tim: the proposal is XML props exactly like what you just types. 10:10 [StevenPemberton] Works for both CSL as CSS 10:10 [Shane] Does anyone care how the namespace magic works? 10:10 [Shane] Or should we save that for another chat? 10:11 [Ralph] I care about namespaces, but I also have to depart 10:11 [DanC] I care how the namespac magic works. 10:11 [StevenPemberton] s/CSL/XSL/ 10:11 [Shane] Basically, the document instance can indicate what prefixes it uses for what namespaces (if any), and the DTD... 10:12 [Shane] magically transforms itself to use the appropriate prefixes 10:12 [Shane] It also propogates all of the prefixed attributes to all of the elements in the DTD. 10:12 [Shane] This means that, for example: 10:13 [Shane] <p xmlns:mathml="..." xmlns:xlink="..." > etc validates 10:13 [Shane] The reason for this is that XSLT permits the propogation of namespace attributes throughout the document instance as a result of transofrmation. 10:13 [StevenPemberton] The only pain is that if you care about DTD validation, and you use a prefix, you have to use prefixes on all elements enclosed, since scoping doesn't work with DTDs. 10:13 [Shane] On all elements for which you have declared a prefix, that is. 10:13 [StevenPemberton] Yes. 10:13 [Shane] Also, you cannot change the prefix in the middle of a document instance. 10:14 [Shane] We clarify these limitations in the latest draft. 10:14 [Shane] They were sort of implicit before. 10:14 [Shane] We also require that modules define a default prefix. This means that you can turn on prefixing without specifying a prefix, and then use the default. 10:15 [Shane] It makes document instances a little less complex. 10:17 [Shane] The details of this magic are somewhat complicated, and are more readily understood by looking at the draft I think. 10:18 [jojo] *** jojo (mojo@adsl-209-76-108-112.dsl.snfc21.pacbell.net) has joined channel #w3c 10:21 [Shane] works for me 10:22 [Michael-NM] Ciao. Thanks for the conversation. 10:22 [Ralph] thanks very much, Steven and Shane. I learned something. 10:22 [StevenPemberton] Byee 10:22 [Michael-NM] *** Michael-NM (cmsmcq@206.206.93.62) has left channel #w3c 10:34 [Shane] later all 10:34 [Shane] *** Shane (ahby@spm.themacs.com) has left channel #w3c 10:34 [Ian] *** Ian (ian@adsl-151-202-33-119.bellatlantic.net) has left channel #w3c 10:36 [mimasa] Bye. 10:39 [Lucio] *** Lucio (picci@picci.spbo.unibo.it) has left channel #w3c 10:52 *** Log ended on 2000-02-25 ***