W3C TAG Telcon

3 May 2005

See also: IRC log


TimBL, Vincent, Norm, Ht, DOrchard, Roy, DanC, noah, Ed
Vincent Quint


<scribe> scribe: dorchard

<DanC> (zakim, counts 8. aren't we 9?)

<DanC> (good question... can binary wait?)

DaveO: what's up with reviewing binary?

noah: no concrete plan on chartering

daveo: we should do something wrt chartering

Vincent: add to agenda?

<DanC> (yes, pls add binaryXML-30 after 14 on today's agenda)

daveo: I'd like the TAG to do something..

noah: a) figure out what to do with doc; b) have discussion about whether tag should participate in charter

<DanC> (for reference, http://www.w3.org/2001/tag/2005/03/action-summary.html is at 2005/05/02 16:44:29 )

<Vincent> Scribes: DO (3 May), NDW, HT, DC, ER, RF, NM

<noah> Actually, what I suggested was a bit less than what was scribed: a) [..got that right...] b) decide what we need to schedule to make sure that we later are prepared for the discussion of whether to be active in charter considerations for Binary XML

sorry noah..

<DanC> . http://www.w3.org/2001/tag/2005/04/26-minutes $Date: 2005/05/03 23:25:50 $

<noah> * Noah notes that he thought Dan just wanted to be sure we had stable links early

<noah> * Relying on the RRSAgent link makes it impossible to do cleanup and editing, which I often find to be worth the trouble.

HT: wanted minutes right next to agenda..
... minutes in cvs, use grep, very nice for searching

vincent: to email on topic of minutes.

<DanC> (btw... this week I'm happy because the minutes settled down by the time the agenda came out)

vincent: please review the TAG update for the AC meeting, team expects within a few days.

<DanC> Subject: DRAFT Summary for the AC meeting

binary xml

<ht> Last XBC mail: http://lists.w3.org/Archives/Public/www-tag/2005Apr/0085.html

<ht> AC meeting summary: http://lists.w3.org/Archives/Member/tag/2005May/0000.html


<DanC> I think this is Ed's reply to dave http://lists.w3.org/Archives/Public/www-tag/2005Apr/0085.html

noah: let's talk about what we want to do.
... where to send? xbc mailing list?
... we know there is a charter, as TAG should we do anything formally or informally?

<Ed> Should we review the message compared to the charter and give the TAG's feedback on the proposed charter?

<noah> Suggest there are two things we might consider doing (either or both):

<noah> 1) Take something resembling note 0085 and send it somewhere as formal comments from the TAG on the work of the XBC group

<noah> and/or

daveo: strawman: be formally involved and send review comments to xbc list + w3ct and/or AC

<noah> 2) Decide that we want to play some formal role in shaping/encouraging/discouraging a possible charter for a new Binary working group (we've been told an initial draft charter is being circulated)

timbl: no decision to make a finding

<noah> +1

timbl: part of problem is difficult to come to conclusion when anonymized

dan: point by point through 85
... not inclined to endorse point #1

<noah> This is not entirely a process question: to some degree, I hear Dave saying: "Having W3C do a binary XML is a technical decision with architectural implications". It so happens that's coming up in the form of a charter.

<DanC> DanC is leading a discussion of the 12 points (well, at least point 1) in http://lists.w3.org/Archives/Public/www-tag/2005Apr/0085.html

<noah> I do agree that the note emphasizes process and workgroup formation a bit more than I would.

ht: tag did not believe the case for value prop for xml binary has been made. Somebody from the audience will say, what is tag suggesting, and an individual tag response

<noah> I think Henry just said the same thing as what I typed above. So, we agree.

ht: could be something like what #1 suggests

Danc: can endorse sentence "The Working Group did not provide benchmarks ..."

<DanC> DanC: maybe change "provide" to publish; maybe they provided it internally

<Roy> It matches the comments I sent

<Roy> ... to say that metrics are needed to convince

<DanC> "quantitative studies are essential to a correct understanding of the trade-offs" -- http://www.w3.org/TR/webarch/#binary

Norm: I agree with point #2

<DanC> (the goal here is a short statement that we can agree to)

daveo: goes through point #3

noah: could do oversimplification if we focus on "speed"
... lots of variabilities, like utf8 to utf16 conversion.
... missing piece: 20 something use cases, pick one like web services, then could get 3 or 4 test environments
... then can look at speed changes..

<DanC> (so... I think we got one keeper and 2 that maybe merit more discussion but aren't there yet. maybe that's enough for this week and we can play the game again next week?)

<DanC> (I yield the floor back to the chair; you can hand it back to me to go thru more points, or we can put this aside for a week, Vincent )

noah: maybe we are saying is that we are less convinced, perhaps narrow to a small # of use cases, and then do rigorous benchmarks

vincent: let's close and move on
... add to next weeks charter

<noah> Actually, I'm not saying that we are or not convinced of merits of binary. What I said was: I'm uncomfortable talking about speed gains across a huge range of use cases. So, this may be another reason to first narrow to a handful of use cases, then consider whether rigorous speed benchmarks are possible for those.

vincent: heartbeat requirements are for WGs
... tag should provide summary every 3 months

timbl: heartbeat is to prevent wgs from going into a corner for a long time without community feedback
... tag is decidely not in a corner. minutes, reviews, etc. are all public

<timbl> timbl: TAG tends to be under a lot of scrutiny and to be engaged in debate in public fora so this is less of a concern than for some WGs.

<Roy> http://lists.w3.org/Archives/Public/www-tag/2005May/0001.html

danc: was there a specific scenario that opened this up?

<DanC> (http://www.w3.org/2001/tag/doc/mime-respect gives me a wierd response)

noah: vincent, your path assumes people need more time to review, does it really need more time?

<DanC> (http://www.w3.org/2001/tag/doc/mime-respect.html works better... but it's not that long... mixing might be handy)

<Roy> (bummer... the XML version is served by default)

<Roy> (can we give it a stylesheet?)

ht: if I did another experiment like noah's, thing put with media type app/java, then the agreement is that the server would run as java and return text/html

<DanC> (the static/dynamic terminology doesn't appeal to me. I liked the way Roy put it ..."HTTP is often, but not always used for filesystem access")

ht: is this ok?

dan: static/dynamic is misleading

noah: roy wrote 3 is a nifty feature when user is making an informed request
... not necessarily static, could be file system based, but let's not act apologetic when it's not file system based. Lots of things are "dynamic"

roy: if we write something down, it would be a new finding, wanted to get it on mailing list

timbl: want to be careful about addressing apache, lots of servers don't use file system conventions

<DanC> (Zope is such a system, fyi, timbl)

timbl: careful not to get into the trap of doing architecture based upon some implementations.

<DanC> (such a system, i.e. one where the filenames have nothing (necessarily) to do with media types)

noah: general case: web isn't about files at all. some are dynamic like clocks, or some are dynamic like java, then tremendously common special case of kind of like a file systems
... and timbl is saying even file system is even a special idiom
... roy's advice can be applied at various layers

danc: I could see another version that links to 17 server documents that say how you do "it" on their server

timbl: could do where we say "and apache does it this way"

roy: apache does it on the uri, file could be .cgi

vincent: back to list

issue 37

ht: relates to 2 concerns of mine: 1) schema component designators in last call.
... 2) most of energy in schema has gone into the "rhs", stuff after the hash
... concern about lhs
... and this relates to versioning

<noah> Schema component designators is a last call draft and is at: http://www.w3.org/TR/2005/WD-xmlschema-ref-20050329/

<DanC> (http://lists.w3.org/Archives/Public/public-webont-comments/2005Mar/0004 "simple barenames for schema component designators" )

ht: got into range-14 problems
... I don't understand what awww says about secondary resources
... status report complete, don't need to talk about today

<DanC> (the "secondary resources" text is not beautiful, according to anybody I know)

<DanC> "The Last Call comment period is expected to end 26 April 2005"

<ht> dorchard: there's a problem with using fragids to refer to abstract components and/but you put an RDDL doc't at the namespace URI

<noah> Helping Dave Scribe: Dave says that he believe's there is a problem when RDDL is used

<ht> timbl: Using RDDL is BAD when you have a fragid

noah: rddl tends to be about namespaces. schema isn't
... schema model is: there is something called a schema, which is separate from the schema declarations

<ht> noah: Schemas are not one-to-one with namespaces, so schema isn't looking at using namespace as URI which you attach the fragid to to ref a component

danc: namespace spec uses term namespace name

ht: schema wg has not proposed that component designators are generated from namespace name + sep + frag-id.

timbl: getting confused between namespaces in general, on lhs, vs schemas

<DanC> (vincent, this is the discussion I think we should have at our next ftf. I earlier requested that we look at the XPath namespace doc, but clearly there are XML Schema component designator specifics and WSDL specifics to look at too)

<Zakim> DanC, you wanted to note that schema component designator thingy is in last call, and doesn't mix well with barenames

<timbl> t-15

danc: something architectural about having lhs be both namespace and schema

<DanC> ACTION: HT to prepare abstractComponentRefs materials for ftf discussion (with help from DanC) [recorded in http://www.w3.org/2005/05/03-tagmem-minutes.html#action01]

ht: struggling to find common ground just within the TAG and find the different positions
... roy and I disagree about even whether it's a fact that the same uri can be used for different resources

<DanC> (I made no comment on foo vs foo# )

ht: dan and I disagree about whether bare hash x.com/blah# can be used to distinguish between info resource and not
... if I want to talk pates, I should use blah#, and then p.hayes home page should use blah

<timbl> timbl: No, that is nonsense.

ht: continues to talk about his attempt to discribe current situation as he perceives it

noah: foo and foo# are different uris, no more the same than foo and bar

<DanC> http://lists.w3.org/Archives/Public/www-tag/2005Mar/0101.html

<DanC> 0101 is SWBPD WG Resolution Regarding httpRange-14

<noah> Noah notes from webarch: http://www.w3.org/TR/webarch/#pr-uri-collision

<DanC> how did we phrase that in webarch? ah... " Assign distinct URIs to distinct resources."

<noah> Right, that's the link pasted above. Note that it's a constraint, not a good practice note.

<DanC> claim t1: { ?X log:uri [ str:startsWith "http:"; str:notContains "#" ] } => { ?X a webarch:InformationResource }

<DanC> claim t2: InformationResource owl:disjointFrom rdfs:Property.

<DanC> claim dc1: dc:title a rdfs:Property

<DanC> claims t1, t2, and dc1 are mutually inconsistent. capice, ht?

<ht> t1 in English -- all http URIs w/o # denote info. resources?

<DanC> yes

<Norm> If the web page is abstract, how is it different from the abstract title?

<DanC> using the same uri for both the property and the page says they're not different, Norm. That's one way of looking at the contradiction.

<ht> So yes, capio, and that was what I was trying to say wrt ambiguity

<ht> You and Tim (but not SWBPG) are saying that vanilla http: URIs should not be used for non-info resources

<DanC> the semantic web best practices WG is asking the TAG to endorse dc1

<DanC> yes, ht

<timbl> introduce HTTP-Range-14 => runoutoftime

<ht> And I _think_ that Roy disagrees with you and Tim

<ht> Sorry I wasn't clear that I did understand that bit, it was _why_ roy disagreed that I didn't get

<DanC> yes... an older example is: rf:robot1 a PhysicalThing, where rf:robot1 starts with http: and has no #

<ht> David, do you know what to do next wrt minutes?



Summary of Action Items

[NEW] ACTION: HT to prepare abstractComponentRefs materials for ftf discussion (with help from DanC) [recorded in http://www.w3.org/2005/05/03-tagmem-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.122 (CVS log)
$Date: 2005/03/31 04:43:41 $