TAG F2F Minutes, Mandelieu

27 Feb 2006, second session

(see previous session, next session)


See also: IRC log



Tim Berners-Lee, Dan Connolly, Roy Fielding, Noah Mendelsohn, David Orchard, Vincent Quint, T. V. Raman, Henry S. Thompson, Ed Rice


Norm Walsh


Vincent Quint


Ed Rice


  1. XMLVersioning-41
  2. xmlFunctions-34

<noah> For the minutes:

<noah> To summarize my comments on the WSA issue. I don't think they way they and we coordinated was a good model for the future.

<noah> When the distinction between reference parameters and reference properties was eliminated, we missed a chance for the TAG to be alerted that there were still plans in some cases to use refParms for identification.

<noah> I think the practical implication was that the TAG eventually realized the true state of play later than was ideal. The good news is that we all eventually converged on a mutually acceptable answer, but I don't think the process that led us there was a model for future interactions.

<DanC_lap> Vincent, did you take seriously Roy's suggestion to take actions that don't all under any other issue and put them in issues.xml under issue42? That seems particularly appropriate for NDW's action on self-describing web.

<DanC_lap> it appeals to me that "self-describing web" would turn into the title of a document that addresses xmlFunctions-34 and some other stuff

<timbl> Me too.

<DanC_lap> (where is dave? when do we expect him back?)

<DanC_lap> (cuz I think xmlVersioning-41 is in the same ball of wax.)

(as is media types)


found at: http://www.w3.org/2001/tag/issues.html#XMLVersioning-41

DO has 4 action items, HT has one relating to this topic

issue started 3 Nov 2003

Text proposed by NW and DO http://lists.w3.org/Archives/Public/www-tag/2003Oct/0137.html

diagram posted at: http://www.w3.org/2001/tag/2005/09/22-diagram1.png

Scribes notes on-going networking issues..L


Henry:Norm and I have not put our heads together yet.
... Following the discussion we had on this topic at the last telecon regarding my email.
... I felt that was a useful discussion and will write that up at more length and bring that up with the group again.
... I'll work with Norm to reconsile what he's been drafting vs what I've been drafting.

Dan:I need this for my presentation..

Henry:You have the message I've sent to date.

V:I guess if there is no progress, we cant really go further today.

Dan:Norm's action continues.

V:yes, but we need a new date.

<DanC_lap> (HT, do you want a separate action?)

V:We have a plan for this issue, Henry will develop his email further and rationalize with Norm.

<DanC_lap> (HT, I guess you're not on the net)

Dan:Is there any current communication?

Noah:I thought so.

V:no, the last time we discussed it at length was our last f2f. and thants when Norm got the issue.

<DanC_lap> (http://www.w3.org/2004/04/webapps-cdf-ws/papers/ ; IBM paper is in .pdf. sigh)

<timbl> http://www.w3.org/2004/04/webapps-cdf-ws/papers/IBM.pdf

<DanC_lap> ACTION: TV to summarize history of DTD/namespace/mimetype version practice, including XHTML, SOAP, and XSLT [recorded in http://www.w3.org/2006/02/27-tagmem-minutes.html#action01]

<DanC_lap> Ed, that action should be read to TV to be sure that's what he had in mind. perhaps not critical, but it would be good

<Zakim> DanC_lap, you wanted to note that the WebOnt WG decided "xml, and get the rest from the instance" wasn't enough

<Zakim> timbl, you wanted to suggest that the mime type needs to (a) hint that it is XML and (b) say what framework is in use out of CDF, RDF, and SOAP, It is NOT useful to try to give an

<Zakim> ht, you wanted to say what Noah was going to say

HT:I guess I've always felt perplexed why we ever introduced the foo+ xml design pattern for mime types because it seems to promise what it hasnet delivered.
... Its not clear to me why 3.. I have a strong temptation to say we should get rid of the 75 or so today and go to zero.


HT:it seems the cleanest way of doing this would be a way to indicated the way of doing the framework, but as Noah said it should be in the document itself.
... then you can look inside the document itself to find out what the framework is.
... Which is differant than what I think your saying.

<Zakim> noah, you wanted to finish what I was saying

Noah:I think whats confusing is in part that media types are serving two parts.

one part is a set of instructions that describe what I can do with this data.

The other thing is things that give us 'early warnings' of what we'll find in the document.

TV the media type should be consistant, or else you wont know who to trust.

<Zakim> DanC_lap, you wanted to note the relevant IETF forum on XML mime types and to say, if Timbl hasn't already, that almost nobody looks inside a document to see whether it's SOAP or

<noah2> Actually, I would prefer to say that media-types serve two "roles".

Dan:almost nobody looks inside a document to see if they should do HTML or soap processing on it.

<noah2> 1) Just because something looks like an XML document doesn't mean it isn't text or something else that accidently looks like XML. So, we use the media type to give you permission to process as XML.

<noah2> 2) The other thing we use media types for is to let us discuss the nature of the document before we look in it. Content negotiation is an example of this.

the majority of the soap implementations are of soap 1.1 which doesnt have a media type.

<Zakim> timbl, you wanted to clarify why you need a framework indicated in the mime type (visibility) and why you need svg+xml.

<noah2> Sometimes, in support of media type usage style #2, you include redundant information in the media type.

<timbl> There are various markets for XML here. There is a market where namespaces mixing is not interesting to people, and so adding the namespace is a pain, and so it could be a default for the MIME type. There is a market for complex expandable systems like the frameworks.

<noah2> Dan: one thing definitely look into soap documents to find out is whether it's a purchase order or someone's job application.

<DanC_lap> it was DO, not DanC, that said "the majority of soap..."

<noah2> That's not in the root element, but it's certainly the kind of thing you might expect in a media type.

Tim:There are differant markets for xml..

1) an invention team who are not really interested in creating name spaces, they're mostly using html tags.

2) 2) Others just want to put some data in a file and they just want to use it as data.

Dave:so one of the ways I think would be really usefull would be to cast people back to Roy's thesus on visibility.

does it go in the xml, does it go in the saop?

<timbl> That was but 1). 2) if folks who do want extensibility and they use a fraemwork.

At what point does a certain amount of software have to know?

and we do have some examples where this is needed and mandated.

VT:in some sense its a protocol riding on a protocol.

Tim:a good thing that some of us can do is to check and make sure we have the consistancy.

HT:Dan, I've been waiting for you to throw in your remark from email that profile may be the right way to do all this stuff.
... to stipulate that there may be some generalized types of content. It seems to me that you suggested that the profile attribute would do much more work in the future then in the past.

Dan:I thought I was writing about fragments..
... are we on 41?

V:no, we're still on self describing web.

Tim:so how does this TAG work?
... so I proposed that we say that its wise to point to which framework we're using.
... I'm proposing a form of a finding is a tree structure to figure out what a document means.
... I wonder if I could get a consensus from the current tag.

Dave:I'm really looking to see why soap+xml is better/worse than .. why did we just pick the two and why is it not working.

<DanC_lap> (is the "it" that timbl is referring to recorded? I wonder if it would be useful to slow down. On the other hand, I'm content to move to a separate agendum, since we have an action on TV to start writing the history/extant-practice of this stuff)

<DanC_lap> TV: one pragmatic reason the [SOAP folks?] might have done inside-xml-doc rather than media type is that to get a media type, you have to go ask somebody else

Noah:We're forgetting that most xml is used between two appliances which both ends know the formats and types.

<DanC_lap> DO: I agree the decentralized aspect of namespaces is a big part of it

Henry:I sparked on VT saying that context negotiation and indexing need to live happily together.

VT:when you pull up the url with a default language page it then pulls up the page with all the other languages also. This is ok, because the content is symantically the same.

HT:That seems slightly sub-optimal.

VT:Yes, but its better than telling the crawlers to crawl with a specific user/role.

HT:You have to do the same work twice.

VT:well, the system as a whole is optimal.
... on the device level, the device to device is not as complex.

Dan:So we were going to move on but then Tim you thought we had consensus but then we didn't have enough consensus.

Tim:The proposal is the mime types is that the mime types if its html its just html, if it's RDF or CDF then the mime type is not used.

Noah:I think it's a plausible position but no more so than the other positions. My feeling is that it's a tough call.

TimBl:if you want to pursue this, but I'm prepared to talk about a hypothetical process where soap is used in an envelope and then you want to be able to stick on this is a .png or an image or something inside. When there is a single thing inside a soap, I can imagine that's a useful thing to open the file because you'll want to retrieve it but I'm worried about trying to combine what people are doing with soap with what people are trying to do with get.

Noah; I heard you say is that we should architecturally try and roll this out. And I just said 'well, ruling that out for all time just didn't seem prudent'

Henry:I just don't see that we have that option, there isn't a way to control this.

Noah:the reason its limited to x is because we say it .

VT:but we have redundancy. There is a finite number of strings someone has come up with that we call mime types.

Noah:The reason that you identify CDF as a framework is ..

Timbl:CDF covers a lot of user interfaces, RDF covers a lot of data.

Noah:I felt like I understood the proposal, but I didn't say I agreed so an email from you describing your position I think that would help.

<scribe> ACTION: TIMBL to To write a short email to make his point so we capture this for future. [recorded in http://www.w3.org/2006/02/27-tagmem-minutes.html#action02]

Noah:I'm hearing Tim say there is a deep differance between frameworks and non-frameworks but now I'm hearing thats not the case and its really a continum.

Dave:a language designer has to specify the level of ignoring the tags you dont use.

Question from ann (audience); The XRI proposal from oasis, what are we going to do? Have you ever heard back from that TC?

Dan:This is not a short question...

<process discussion around how oasis/w3c work together> informative but process issues which are outside of the TAG charter.

<noah> Final representations of "Rule of Least Power" are now world-readable at http://www.w3.org/2001/tag/doc/leastPower.html and http://www.w3.org/2001/tag/doc/leastPower-2006-02-23.html). We are ready for Vincent to link this as an approved TAG finding.

Summary of Action Items

[NEW]ACTION: TIMBL to To write a short email to make his point so we capture this for future. [recorded in http://www.w3.org/2006/02/27-tagmem-minutes.html#action02]
[NEW] ACTION: TV to summarize history of DTD/namespace/mimetype version practice, including XHTML, SOAP, and XSLT [recorded in http://www.w3.org/2006/02/27-tagmem-minutes.html#action01]
[End of minutes]

Next session

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/03/22 14:23:10 $