- From: Dan Connolly <connolly@w3.org>
- Date: Fri, 25 Feb 2000 11:17:02 -0600
- To: xml-dist-app@w3.org
The log of the main channel is available at: http://www.w3.org/2000/02/25-xhtml-irc but we broke out a separate channel for protocols; here's the log of that: Session Start: Fri Feb 25 08:22:42 2000 [08:22] *** Now talking in #xml-dist-app [08:23] *** Ralph has joined #xml-dist-app [08:23] <Ralph> it takes me a few minutes to startup an RRSAgent process... (virc is slow to startup) [08:25] <DanC> I could just work with my local log. Hmm... [08:27] <Ralph> RRSAgent2 doesn't seem to be able to connect to irc.openprojects.net:6665 [08:28] <DanC> I'll just use a local log [08:28] *** DanC changes topic to 'http://lists.w3.org/Archives/Public/xml-dist-app/' [08:29] *** timbl has joined #xml-dist-app [08:29] *** dcleary has joined #xml-dist-app [08:30] *** eric1 has joined #xml-dist-app [08:30] <eric1> hi all [08:30] *** KenM has joined #xml-dist-app [08:30] *** Michael-NM has joined #xml-dist-app [08:30] <timbl> XML-RPC and LDO both describe the attributes and parameters. Thesis for the sake of argument: For every RPC interface you could instead make a namespace with elements for each procedure [08:31] <eric1> the payback would be when people re-use eachothers types and therefor can re-use bits of eachothers engines. [08:31] <timbl> I can see a day when the net is full of <param name="param-name" value="argname"> ... :) [08:32] <KenM> timbl: yes. we have a prototype serializer that does use element names for procedure arguments, object member fields, and dictionary names. [08:32] <timbl> In that case you are using namespaces bigtime [08:33] <timbl> Do you dostub generation (forgive my ignorance)? [08:33] *** Michael-NM has quit IRC (Ping timeout for Michael-NM[206.206.93.62]) [08:33] <KenM> not at the bottom end. the bottom end works a lot like CGI, it expects the server to coerce types as necessary [08:34] <KenM> this avoids the need for IDL or stubs on the clients [08:34] *** dcleary has left #xml-dist-app [08:34] <timbl> Yes. Using namespaces, then the interface definition is the schema. [08:35] <KenM> in the middle, where IDL is already available, the client can still go stubless while the libraries do the coercion [08:36] *** Michael-NM has joined #xml-dist-app [08:36] <timbl> Presumbably the IDL is a seprate language - but can you see it moving into teh schema document so that in the future eth receiver of a message could dereference the namespace pointer and generate stubs on the fly? [08:36] * Michael-NM resurfaces after GTE loses connection, three times in a row. [08:38] <KenM> for object values, easily. I'm not so sure how easily for procedure/method signatures [08:39] <KenM> there would have to be a way to associate a method with an object value [08:40] * timbl surfs http://casbah.org/LDO/xml-serialization.html [08:40] <eric1> i propose that using XML schema to replace the soap types defined in http://www.xmlrpc.com/stories/storyReader$7 would be a "good thing". any comments? [08:40] <timbl> /me finds Example type attribute values are: [08:40] <timbl> perl:scalar [08:40] <timbl> xml-dcd:ui1 [08:40] <timbl> corba:i4 [08:40] <timbl> mime:text/html [08:40] <timbl> ldo:number [08:40] <timbl> seems to be crying out for namespaces [08:41] * timbl not sure what ken means by object value [08:41] <KenM> exactly, though they were originally inspired by URI scheme identifiers [08:42] <Michael-NM> Methods have to be tied to a particular object instance. [08:42] <DanC> Don Box seems to expect to replace parts of the SOAP type system with XML Schemas. [08:42] <Michael-NM> Not just to an object class. [08:42] <timbl> Types should be first class objects <--> have URIs [08:42] <timbl> That way you can define your own and wait till the other XML types are available [08:42] <KenM> timbl: just "objects" [08:42] <DanC> Box on Soap and schemas: http://lists.w3.org/Archives/Public/xml-dist-app/2000Feb/0000.html -> http://www.xml.com/pub/2000/02/09/feature/index.html [08:42] <timbl> and then you can retrospecively document any equivalences [08:46] <timbl> One fo the things which is surface increasingly often is a call for some common work on protocols between all the various xml protocols - what do you think, Ken? [08:48] <KenM> stated simply, I agree. after that, though, it gets complicated. a lot of "XML protocol" is just sending XML instances over TCP or HTTP. A lot of B2B folks want routing, store-n-forward, validation, third-party validation, etc. it's those latter that [08:49] <KenM> "really" need common protocols. [08:51] <timbl> Sounds different layers [08:51] *** iefbr14 has joined #xml-dist-app [08:52] <KenM> yes, payload and header for one. then the payload may be of a general type, but with industry or vendor additions. the header may have provider additions, etc. [08:52] <timbl> Like a simple namespace for associating a request with a reply, a seperate namespace for busness clutter. [08:52] <eric1> i imagine a layer that outside the protocol/request/response layer that allows a client or a server to decide to send the request over SMTP rather than HTTP [08:53] <timbl> I prefer to think of headers as a transparent wrapper. [08:53] <eric1> or serve as a store and forward envelope [08:54] <KenM> I recently posted a draft of an API for LDO that touches on protocol selection <http://casbah.org/pipermail/devel/Week-of-Mon-20000221/000669.html> [08:54] <timbl> Example, make a v simple langauge yu can send by email which means "please take the following and execute an HTTP POST on my behalf with the following conditions:" [08:55] <KenM> I'm thinking of using extended URLs, prefixing a transport URL with an upper-level procotol scheme, like xmlrpc:http://casbah.org/listener [08:56] *** iefbr14 has left #xml-dist-app [08:56] <timbl> Hmmm... I have always had two minds about puttingstructure in URIs [08:57] <eric1> is that better than associating symantic wiht a namespace and having the apps understand the namespace? [08:57] <timbl> like http://(smtp to gateway .com. then x25 to fred and then http to 12.34.45.6)/foo [08:58] <timbl> Ken: Can't you just make the listener a typed object which acceots xmlrpc POST? [08:58] <eric1> the prob with xmlrpc:http%3A//casbah.org/listener is that it's not a general solution. the uri protocol space can become cluttered more quickly than the namespace space. [08:58] <timbl> Afetr all, there are many typed of service on the web. [08:58] <eric1> also, uri protocols need to be centralized [08:58] <KenM> the question becomes: how does the client know? [08:59] <timbl> You can think of XMLRPC as one of teh namespaces whichthe listener supports. [08:59] <eric1> same way either way, it was programmed to look in either the uri or the namespace data [08:59] <timbl> You define am RDF statement (xxx supports nnnn) [09:00] <timbl> You then distribute the RDF statement any way you like. One way is in response to a HEAD request [09:00] <KenM> at the low-end, there is no namespace data (there's no external data outside of verbal/written communication) [09:00] <timbl> It separates the langauges something speaks whith its identity. [09:01] <eric1> if i write something that takes apart xmlrpc: uris, i can just as easily write the same thing to look for xmlns:xrpc="http://foo/xmlrpc" namespaces and look for xrpc:bar parameters [09:01] <timbl> So http://house.blee.org/outside/faucet would have an idenity which would persists as I updates its LDO version [09:02] <KenM> eric1: yes, if the message being sent includes that information in a namespace. [09:02] <eric1> ken: i don't understand your "low-end" statement [09:02] <timbl> Ken, how does a client find out the address of the object in the first place, in a typicla scenario? [09:04] * timbl realised the time [09:05] <KenM> note: I see a difference between remote procedures/methods and XML messaging (and see I've been mixing the two, unfortunately) [09:05] * KenM is OK to continue [09:05] <eric1> i don't see much of a difference [09:05] * eric1 is fine for another hour [09:06] <KenM> XML messaging presupposes that the client starts out with an XML instance. remote procedures/methods deal only with already-defined language objects [09:06] * timbl has to leave [09:06] <eric1> bye tim [09:07] *** timbl is now known as tim-out [09:07] <eric1> what is an XML instance? [09:07] <eric1> (sorry if i'm being dense here) [09:07] <KenM> timbl: one finds the address of the object usually in a written document somewhere, like a "how to use XML-RPC to access the database". [09:09] <Michael-NM> 'instance' = 'document instance', as opposed to 'document type' [09:09] <eric1> roger [09:09] <KenM> in what I'm describing, XML messaging makes XML primary and the client/server language/system secondary (the clients job is to get the XML to the server). In RPC, the client/server language/system is primary and XML secondary [09:11] <KenM> (note, we want to support both in LDO) [09:12] <eric1> is this personel data update a message or a procedure? [09:12] <eric1> <P:person><P:uid DB:type='unique'>123</P:uid><P:status DB:type='update'>fired</P:status></P:person> [09:12] <KenM> on the XML messaging side, you are correct that the message itself can include routing and protocol information. On the RPC side, that has to come from somewhere else, a nameserver, inherently, or from the user. [09:20] * Michael-NM has to drop out now. So long! [09:20] <eric1> bye [09:22] *** Michael-NM has left #xml-dist-app [09:34] <eric1> http://frontier.userland.com/stories/storyReader$1077 [09:35] <KenM> http://www.xmlrpc.com/spec [09:38] <eric1> <param type="i4">41</param> [09:39] <KenM> http://casbah.org/LDO/xml-serialization.html [09:54] *** Ralph has left #xml-dist-app [10:50] *** KenM has quit IRC (will follow-up to xml-dist-app) Dan Connolly wrote: > > I just discovered there are 117 subscribers lurking in this forum. > I hope you'll take advantage of this IRC chat to come out of the > woodwork > and let us know what you're up to. > > -- > Dan Connolly, W3C > http://www.w3.org/People/Connolly/ > > ------------------------------------------------------------------------ > > Subject: W3C Chat before XTech 2000: XHTML - a bridge to the Web of the future > Date: Sun, 20 Feb 2000 21:36:22 -0600 > From: Dan Connolly <connolly@w3.org> > Reply-To: michael@w3.org, ht@w3.org, dv@w3.org, eric@w3.org, www-html@w3.org > Organization: World Wide Web Consortium (http://www.w3.org/) > To: www-html@w3.org > Newsgroups: comp.text.xml,comp.infosystems.www.authoring.html > > XTech 2000 (http://www.gca.org/attend/2000_conferences/xtech_2000/) > is around the corner, and members of W3C team will be there, > including Eric Prud'hommeaux, Michael Sperberg-McQueen, Henry Thompson, > and Daniel Veillard. > > If you're going (or not), you may have questions regarding recent W3C > developments, including > > XHTML 1.0 > http://www.w3.org/TR/xhtml1 > > How W3C works > http://www.w3.org/Consortium/ > > Plus, you might be interested in discussions of XML messaging, > distributed computing, transactions, and protocols: > > http://lists.w3.org/Archives/Public/xml-dist-app > > In this spirit, we'd like to have a pre-conference chat - here are the > details: > > Who: All are welcome > > W3C team members attending the chat include Dan Connolly, > Masayasu Ishikawa, Eric Prud'hommeaux, Michael > Sperberg-McQueen, > Henry Thompson, and Daniel Veillard. > > Also from the HTML Working Group: Steven Pemberton and Shane > McCarron > > When: Friday 25 February, at 1400Z (9am Eastern Time), for about an hour > > (Apologies to the parts of the world where that's > inconvenient.) > The log will go online - our previous chat is at: > http://www.w3.org/1999/12/w3c-irc2409 > > Where: irc://irc.openprojects.net/#w3c > i.e. channel #w3c on irc.openprojects.net > > about this IRC network, see > Open Projects Network - New User? > http://openprojects.nu/about.html > > stay tuned to the XML home page http://www.w3.org/XML/ > for other details. > > What to bring: > > Send your pre-chat ideas, proposals, position papers, and > "here's what I did with XHTML" to > > www-html@w3.org > > Suggested reading: > > XHTML 1.0 > http://www.w3.org/xhtml1 > > HTML Activity > http://www.w3.org/Markup/ > > W3C Extensible Markup Language (XML) Activity > http://www.w3.org/XML/Activity > > www-html mailing list > http://lists.w3.org/Archives/Public/www-html > > xml-dist-app mailing list > http://lists.w3.org/Archives/Public/xml-dist-app > > xml-dev mailing list > http://www.lists.ic.ac.uk/hypermail/xml-dev/ > http://www.egroups.com/list/xml-dev/info.html > > Subject: Call for Participation: W3C Public IRC Chat on XHTML > Sat, 19 Feb 2000 11:33:17 +0100 > http://www.egroups.com/group/xml-dev/18169.html? > > -- > Dan Connolly > http://www.w3.org/People/Connolly/ -- Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Friday, 25 February 2000 12:17:18 UTC