IRC log of tagmem on 2002-06-03

Timestamps are in UTC.

18:51:03 [RRSAgent]
RRSAgent has joined #tagmem
18:51:11 [timmit]
rrsagent, stop
19:00:29 [TBray]
TBray has joined #tagmem
19:01:37 [Zakim]
+Tim.Bray
19:01:44 [TBray]
Howdy
19:02:28 [timmit]
Chris said he woudl be late
19:02:38 [timmit]
PaulC, Ian, Stu sent regrets
19:02:59 [timmit]
Waiting for DO, DanCon ,...
19:03:05 [timmit]
Zakim, who is here?
19:03:06 [Zakim]
I see TimBL, Norm, Roy, Tim.Bray
19:03:42 [Norm]
DanCon: http://www.w3.org/2002/06/03-tag
19:04:20 [Zakim]
+DanC
19:05:00 [DanCon]
0. scribe
19:05:03 [DanCon]
DanC volunteers.
19:05:22 [DanCon]
tbl partial regrets; norm to chair if/when timbl leaves
19:05:55 [timmit]
http://www.w3.org/2002/05/27-tag-summary
19:05:57 [Zakim]
+DOrchard
19:05:59 [DanCon]
proposed: http://www.w3.org/2002/05/27-tag-summary is a true record of 27May telcon
19:06:17 [DanCon]
RF notes pointer from agenda busted.
19:06:38 [Dave]
Dave has joined #tagmem
19:06:39 [DanCon]
danc will fix.
19:07:12 [DanCon]
RESOLVED. http://www.w3.org/2002/05/27-tag-summary is a true record of 27May telcon
19:07:30 [DanCon]
bray 2nds the prosal. (which was just resolved)
19:07:41 [DanCon]
-- agenda review...
19:07:48 [DanCon]
bray: some stuff from 27May seems missing...
19:08:17 [DanCon]
... never mind.
19:08:26 [DanCon]
PROPOSED: to meet 10jun
19:08:33 [DanCon]
potential regrets: DaveO
19:09:09 [DanCon]
DanCon: chair?
19:09:23 [DanCon]
timbl has a telcon just before
19:09:53 [DanCon]
NOTE WELL: Stu/Ian are expected to organize the agenda.
19:10:04 [DanCon]
RESOLVED: to meet again 10Jun.
19:10:11 [DanCon]
s/Stu/Suart/
19:10:30 [DanCon]
=== 1.1 New issues?
19:11:37 [DanCon]
DanC proposes to accept the issue raised by robla about error stuff
19:12:02 [DanCon]
to wit error recovery practices
19:12:27 [DanCon]
RESOLVED: to accept issue errorRecoveryPractices-NN
19:12:35 [DanCon]
RESOLVED: to accept issue errorRecoveryPractices-20
19:12:44 [DanCon]
RESOLVED: to accept issue errorHandling-20
19:12:55 [DanCon]
==== 2. Technical (75min)
19:13:17 [DanCon]
continued: ACTION IJ 2002/03/18: Integrate/combine one-page summaries (Revised)
19:13:19 [Zakim]
+Chris
19:13:34 [Zakim]
-Chris
19:13:48 [DanCon]
continued: (in progress): ACTION TBL 2002/05/05: Negotiate more of IJ time for arch doc
19:14:44 [DanCon]
bray: you guys (Ian/TimBL/Fielding) came up with a compressed intro, and in a recent telcon, there was positive feedback... any news?
19:15:10 [DanCon]
is this it? http://www.w3.org/2001/tag/2002/0508-intro
19:15:57 [timmit]
http://www.w3.org/2002/05/24-architecture
19:16:25 [timmit]
Not currently published
19:16:32 [DanCon]
bray: yeah; http://www.w3.org/2002/05/24-architecture; that one's good.
19:16:55 [DanCon]
Fielding: I took a look; it's rough notes from a discussion...
19:17:32 [DanCon]
Bray: I prefer this to the sort of tutorial style in 0508-intro
19:17:50 [DanCon]
... and I think there was support for this style last week.
19:18:32 [DanCon]
timbl: presumably, there would be some text between the points?
19:18:56 [DanCon]
Bray: just a little bit
19:19:13 [DanCon]
DanCon: I hope the ultimate document has stories/examples.
19:19:23 [DanCon]
Bray: yes, examples are likely the best way to elaborate
19:19:40 [ChrisL]
ChrisL has joined #tagmem
19:19:47 [DanCon]
TimBL: hmm... different colors for different stuff...
19:20:11 [DanCon]
[ push: ACTION CL 2002/05/05: Add concern regarding non-western characters to the POST scenario (issue whenToUseGet-7) ]
19:20:35 [DanCon]
re DO/TB/CL 2002/05/05: Polish up DO's .1-level draft and find out what's going on with XForms ...
19:20:42 [DanCon]
Bray reports some progress...
19:20:55 [DanCon]
Orchard: as we discussed last week...
19:21:27 [DanCon]
... looking at XForms was on the basis of one technical approach; that doesn't seem like the most likely approach at this point.
19:21:51 [ChrisL]
because?
19:21:52 [DanCon]
Bray: the XMLP WG decided to address this issue before last call...
19:22:43 [DanCon]
pending XMLP response: ACTION DO/TB/CL 2002/05/05: Polish up DO's .1-level draft and find out what's going on with XForms
19:22:58 [timmit]
agenda+ TimB concern of xmlp understanding
19:23:30 [DanCon]
continued: ACTION NW 2002/05/20: Draft a finding for formattingProperties-19; find out source of issue from CSS WG.
19:24:22 [DanCon]
continued: ACTION IJ 2002/05/20: Revise and publish whenToUseGet-7 finding
19:25:42 [DanCon]
bray proposes to postpone 1. Confirm completed status of Internet Media Type registration, consistency of use until PaulC is available.
19:27:34 [DanCon]
It was discussed in www-tag that Microsoft Internet Explorer is inconsistent with this finding as drafted...
19:28:12 [DanCon]
Tantek's objection (http://lists.w3.org/Archives/Public/www-tag/2002May/0110) is noted.
19:28:50 [ChrisL]
MSIE for windows. Not clear that IE for Mac does the same thing
19:28:57 [timmit]
http://www.w3.org/2001/04/roadmap/xml-charset.svg
19:29:09 [timmit]
DanC: Wasn't this diagram broken?
19:30:28 [ChrisL]
yes its broken in the case of text/css with no explicit charset
19:32:05 [timmit]
For Roy: http://www.w3.org/2001/04/roadmap/xml-charset.n3 ;-)
19:32:08 [DanCon]
DanCon: how about citing the known inconsistency
19:32:40 [DanCon]
NW offers to make a PNG version of the diagram
19:32:44 [Zakim]
+Chris
19:33:37 [DanCon]
ACTION DanC: research the bug in the svg diagram.
19:33:51 [ChrisL]
the format of the diagram is not going to affect its content
19:34:27 [Dave]
"I see no inconsistency between my view and the text"? heh
19:34:50 [ChrisL]
mt objection did not take account of the titkle - scope is xml only, so objection was wrong
19:35:09 [ChrisL]
(typing one handed, sory)
19:35:38 [DanCon]
continued: ACTION CL 2002/05/05: Add concern regarding non-western characters to the POST scenario (issue whenToUseGet-7)
19:36:43 [DanCon]
timbl: I haven't checked this "Do all "shoulds" of RFC 3023 section 7.1 apply?"; anybody else
19:36:44 [DanCon]
?
19:37:39 [DanCon]
Bray: I think the rFC3023 editors would agree that text/xml should be deprecated in the general case. (because of the charset bugs)
19:37:53 [timmit]
http://www.ietf.org/rfc/rfc3023 7.1 referencing
19:38:46 [DanCon]
yikes! application/xml has a charset parameter? phpht.
19:39:35 [DanCon]
ChrisL: I objected to that, but the editor wasn't convinced.
19:40:03 [DanCon]
bray proposes an issue: RFC3023-21
19:40:51 [DanCon]
... in partcular: the charset bit under application
19:40:56 [DanCon]
... and deprecate text/xml
19:41:10 [DanCon]
... due to charset fallback conflict.
19:41:36 [DanCon]
... re application/xml, defaults are broken. [Chris said it right; scribe missed it]
19:42:59 [DanCon]
RESOLVED: to accept RFC3023-charset-21
19:43:29 [DanCon]
ACTION Chris: to write up the issue in the next day or so.
19:43:34 [DanCon]
... to www-tag
19:44:27 [DanCon]
=== skipping 2.1.2 # QNames as Identifiers (issue qnamdAsId-18). (~5min); preempted by charmod...
19:44:48 [DanCon]
=== 2.2.1. Finalize TAG position on charmodReview-17. (~20min)
19:45:12 [DanCon]
Bray: large parts of it (bytes/chars/glyphs) should go to REC right away, but early uniform normalization needs CR.
19:45:23 [DanCon]
(Bray is summarizing something...)
19:45:42 [DanCon]
ChrisL: collation/sorting stuff needs more work.
19:45:55 [DanCon]
Chris' msg: http://lists.w3.org/Archives/Public/www-tag/2002May/0164.html
19:46:01 [DanCon]
moved, 2nded...
19:46:05 [Roy]
+1
19:46:49 [DanCon]
NW is willing to re-state his comments as a diff from chris's
19:46:50 [Roy]
Z doesn't grok Apache votes
19:47:47 [timmit]
Yes we are resolved
19:47:48 [ChrisL]
not
19:47:56 [ChrisL]
;-)
19:47:57 [DanCon]
;)
19:48:14 [DanCon]
Bray: re "you should select one encoding"...
19:48:46 [timmit]
TimBL: agree ith tb
19:48:49 [DanCon]
... for humans, I suggest MUST not restrict any further than what XML already allows.
19:49:06 [DanCon]
e.g. party A writing in 8859-1, party B writing in shift-jis.
19:49:53 [DanCon]
Bray: for SOAP, I could agree with limiting to a single encoding. But not for stuff you might be editing in emacs; e.g. SVG, XHTML.
19:50:05 [DanCon]
ChrisL: well, if you want to be sure, you have to use one of the standard encodings.
19:50:25 [DanCon]
... parties that choose some other encoding take a risk in doing so.
19:51:29 [DanCon]
RESOLVED: agreed to http://lists.w3.org/Archives/Public/www-tag/2002May/0164.html , subject to ammendments by norm.
19:51:36 [DanCon]
ACTION Norm: tell I18N WG.
19:52:04 [DanCon]
2nded: for humans, I suggest MUST not restrict any further than what XML already allows.
19:52:51 [DanCon]
ChrisL: an XML-based thingy that says "UTF-16 isn't allowed" doesn't seem cost-effective.
19:53:14 [DanCon]
... there's extra work to flag the error, and no benefit.
19:54:07 [DanCon]
RESOLVED: if it is in XML, you MUST not restrict the encodings to anything less than UTF-8 and UTF-16.
19:54:31 [DanCon]
PROPOSED: for those specification that are designed for authoring by humans, ...
19:55:02 [DanCon]
... you must not restrct the encoding beyond what the XML spec says. [DanC thinks this misses the point about shift-jis etc.]
19:55:27 [ChrisL]
(I noticed an error in the diagram, case of utf-8 with bom (now legal) is not really adressed
19:57:15 [DanCon]
[some example cases... one with EU and iso8859-15... scribe isn't following]
19:58:21 [DanCon]
Zakim, who's here?
19:58:22 [Zakim]
I see TimBL, Norm, Roy, Tim.Bray, DanC, DOrchard, Chris
19:58:30 [DanCon]
regrets: Ian, PaulC, Stuart
19:59:11 [DanCon]
[another example: suppose MathML banned 8859-1?]
20:00:12 [DanCon]
PROPSED: specs should not impose char encoding rules beyond what's in XML...
20:00:23 [Zakim]
-TimBL
20:00:24 [DanCon]
... i.e. consumers must grok UTF-16/UTF-8, may grok others.
20:02:50 [DanCon]
[Norm takes the chair]
20:03:30 [DanCon]
RF: the key is [something about consistency of parts of messages?]
20:04:55 [DanCon]
PROPOSED: specifications SHOULD NOT add character encoding rules beyond those in the XML spec.
20:05:05 [Norm]
SHOULD NOT in specifications add rules for character encoding beyond what is provide in XML. And you MUST NOT restrict character sets beyond what XML allows (you MUST support UTF8 and UTF16)
20:05:24 [DanCon]
... in particular we disagree with "you should select one encoding"
20:06:51 [Norm]
Examples:
20:06:59 [Norm]
Limit to UTF8 for machine-to-machine is OK.
20:07:03 [DanCon]
Bray: e.g. for some routing protocol that uses XML, MUST NOT use other than utf-8/16 is acceptable, but not for something human-editable like MathML
20:07:37 [ChrisL]
disagree with norms utf-8 onlt thing
20:07:49 [ChrisL]
no, it was goot in general
20:08:41 [DanCon]
PROPSED: SHOULD NOT in specifications add rules for character encoding beyond what is provide in XML. And you MUST NOT restrict character sets beyond what XML allows (you MUST support UTF8 and UTF16) . e.g. for some routing protocol that uses XML, MUST NOT use other than utf-8/16 is acceptable, but not for something human-editable like MathML
20:08:42 [DanCon]
* Norm decides to stop interfering with DanCon's minute taking
20:09:09 [DanCon]
2nded
20:10:24 [Roy]
My "key" point was that, if XML is used within another protocol that depends on the value of the charset parameter, the the charset MUST only be given if the entire document is within that character encoding.
20:10:25 [DanCon]
so RESOLVED.
20:10:25 [DanCon]
ACTION Norm: tell I18N WG while he's at it.
20:10:26 [DanCon]
agenda?
20:10:46 [DanCon]
== XMLP WG and whenToUseGet-7
20:11:09 [DanCon]
DaveO: XMLP has decided to slip SOAP 1.2 last call for >= 2 weeks while a GET task force works on this...
20:11:48 [DanCon]
DaveO: the lead proposal doesn't say how to encode a SOAP request as a URI.
20:12:22 [DanCon]
... lets publishers issue URIs that folks can GET, but doesn't specify how to compte the URI from a SOAP request.
20:13:02 [DanCon]
DaveO: also, the task force asked for, and were granted, permission to communicate directly with www-tag
20:13:46 [DanCon]
DaveO: a new message exchange pattern, "one way poll" is part of the lead proposal...
20:13:53 [DanCon]
... there's an HTTP binding under consideration.
20:13:56 [Norm]
s/poll/pull, i think
20:15:07 [DanCon]
DanCon: bummer; I hope to know the address of all these SOAP gettable things in advance.
20:15:28 [DanCon]
RF: I prefer not to; I prefer the names are meaningful. yes, I accept the cost of the extra round-trip.
20:16:07 [DanCon]
Bray: I have a concern: making providers think about the names is extra work, and folks will do the least work.
20:17:20 [DanCon]
[scribe joins discussion... ]
20:19:45 [DanCon]
ChrisL: [an example of a multi-step submission...]
20:21:04 [DanCon]
Bray clarifies: the objective isn't to model *all* SOAP transactions into URI space... just the trade-some-params-for-some-info cases.
20:23:30 [DanCon]
[... discussion of architectural points, deployment, ...]
20:24:07 [DanCon]
RF: I don't like gateways [scribe missed something]. They tend to bypass administrative review.
20:24:19 [ChrisL]
roy said he was against named systems that create themselves because it bypasses administrative review of what is available
20:25:25 [DanCon]
[scribe is now pretty much lost]
20:27:45 [DanCon]
PROPOSED: if soap content turns out to be useful, having URIs where you can GET it is good.
20:28:49 [DanCon]
this is orthogonal to whether there's a well-known for computing such a URI from the parameters.
20:29:04 [DanCon]
well-known algorithm.
20:29:07 [Norm]
The flip side of Roy's proposal is that if there exists a SOA
20:29:28 [TBray]
SOA?
20:29:29 [Norm]
...if there exists a SOAP service that provides useful information, it should be addressable via a URI.
20:30:18 [DanCon]
Roy: who's in the overlap between TAG and this task force? DaveO: me and Stuart
20:30:33 [Roy]
Yes
20:30:40 [DanCon]
DaveO: and PaulC, sometimes.
20:32:15 [Norm]
ACTION: DaveO will report to the XMLP task force that for operations where GET would be appropriate SOAP nodes should provide the ability to use GET
20:32:17 [DanCon]
RESOLVED: [something]. ACTION DaveO: tell the XMLP task force.
20:32:27 [Zakim]
-Chris
20:32:30 [Zakim]
-Tim.Bray
20:32:30 [Zakim]
-Norm
20:32:32 [Zakim]
-DOrchard
20:32:35 [Zakim]
-DanC
20:32:36 [Zakim]
-Roy
20:32:36 [Zakim]
TAG_Weekly()2:30PM has ended
20:33:02 [Norm]
DanCon: does the last action I provided satisfy your question about the minutes?
20:33:28 [DanCon]
hmm... well, it doesn't say "trade a few parmeters for some info" like I wanted it to, but maybe that wasn't what we decided.
20:33:45 [DanCon]
I suppose it's academic. What matters is what DaveO sends.
20:34:09 [DanCon]
er... so in short, yes.
20:34:12 [Norm]
Agreed. At this point, DaveO will say what he says :-)
22:23:00 [DanCon]
Zakim, bye
22:23:01 [Zakim]
Zakim has left #tagmem
22:23:05 [DanCon]
RRSAgent, bye