IRC log of tagmem on 2003-04-28

Timestamps are in UTC.

18:27:34 [RRSAgent]
RRSAgent has joined #tagmem
18:29:08 [Stuart]
Stuart has changed the topic to:
18:38:08 [DanC]
DanC has joined #tagmem
18:39:53 [Stuart]
Hello Dan, how do I make you an operator?
18:40:13 [Stuart]
Have you seen the threads on URI denotation on the URI list?
18:40:41 [DanC]
denotation: some of them
18:42:19 [Stuart]
Have you seen the request from the VoiceBrowser folk for a telcon with the TAG (for this week)? and are you willing/able to participate? That's on the contentTypeOverride issue.
18:44:21 [Norm]
Norm has joined #tagmem
18:45:40 [DanC]
yes, seen the request...
18:48:39 [DanC]
I'd like to think that the right thing will happen without me spending even more time on contentTypeOverride. I sent a review comment, got it endorsed by the TAG, helped Ian write a finding. If they're not convinced, I'm happy for other folks to decide what to do with the inconsistency.
18:50:07 [DanC]
i.e. if I'm the only person who cares, then it's not worth bothering with.
18:51:38 [Stuart]
Ok... but we've been asked to participate in a call about and issue we have open, and think it behooves us to try and find a TAG person or two willing to discuss the issue with them.
18:52:10 [DanC]
I'm willing to discuss it in a TAG telcon. I'm just not inclined to spend extra time on it.
18:52:36 [DanC]
i.e. if it's worth the group's time, then very well. if not, then it's not worth my time either.
18:53:35 [DanC]
or if I could be excused from other TAG duties for a week or two, I could go talk with them.
18:53:49 [Stuart]
Ok, we'll see where we get to today... just trying to feel the ground ahead of the call.
18:53:55 [DanC]
18:57:02 [DanC]
Stuart, have you arranged for a scribe?
18:59:06 [Chris]
Chris has joined #tagmem
18:59:31 [Zakim]
TAG_Weekly()2:30PM has now started
18:59:36 [Zakim]
18:59:43 [Zakim]
18:59:46 [Zakim]
19:00:20 [Zakim]
19:00:29 [DanC]
it rang, but when I answered I heard a dial-tone.
19:00:41 [Zakim]
19:00:45 [Zakim]
19:00:53 [Stuart]
zakim, ??P1 is me
19:00:53 [Zakim]
+Stuart; got it
19:01:17 [TBray]
TBray has joined #tagmem
19:02:23 [Zakim]
19:05:55 [Zakim]
19:06:29 [Zakim]
19:07:28 [DanC]
----- Convene. Williams in the chair. Connolly.
19:07:33 [DanC]
----- Convene. Williams in the chair. Connolly scribe.
19:07:40 [Zakim]
19:07:54 [DanC]
Zakim, who's on the phone?
19:07:54 [Zakim]
On the phone I see Norm, Chris, DanC, Stuart, Tim_Bray, RoyF, PCotton, TimBL
19:08:24 [DanC]
regrets: IanJ.
19:08:27 [DanC]
19:08:38 [DanC]
--- review of 14Apr minutes
19:08:54 [DanC]
PROPOSED to accept as a true record
19:08:57 [DanC]
19:08:59 [tim-lex]
tim-lex has joined #tagmem
19:09:02 [Zakim]
19:09:06 [tim-lex]
I amstain
19:09:19 [DanC]
RESOLVED to accept as a true record, Berners-Lee abstaining
19:09:31 [DanC]
--- review of agenda
19:10:58 [DanC]
-- Next meeting: 5 May?
19:11:02 [DanC]
DanC is available.
19:11:06 [DanC]
regrets Williams
19:11:31 [DanC]
RESOLVED to meet next 5May, Walsh to chair
19:12:00 [TBray]
q+ to mention a liaison thing
19:12:23 [DanC]
-- request to meet with Voice Browser folks re contentTypeOverride-24
19:13:26 [DanC]
Bray: got a liaison request re namespace policies...
19:15:41 [DanC]
DanC: how about monday, 5 May, i.e. our next meeting, for the get-together with Voice Browser folks?
19:16:15 [DanC]
PaulC: yeah, 5May
19:16:27 [DanC]
19:16:59 [DanC]
ACTION Williams to follow up (with Ian, Norm, etc.)
19:17:32 [DanC]
-- 1.1 Meeting planning
19:17:35 [TBray]
Hrmrm my w3c password doesn't work...
19:18:07 [DanC]
StuartW: didn't make much progress on distributed meeting planning [still 2 and 19 June]
19:18:18 [DanC]
PaulC: not much progress on [whatever he was supposed to do here]
19:18:34 [DanC]
DavidO/Chris: no progress here either. but we expect to get on it presently.
19:19:13 [DanC]
PaulC: I gather the team is trying to get AC materials together early.
19:19:34 [DanC]
Action SW 2003/04/14: Propose meeting times, structure to TAG. continues.
19:19:46 [DanC]
Action PC 2003/04/14: Propose meeting times, structure to TAG. continues
19:19:56 [DanC]
--- namespace policy
19:20:08 [Chris]
thats the second report in a week of a password not working on lists but working on the website
19:20:21 [DanC]
PaulC: goes back to ebxml versioning stuff...
19:21:21 [DanC]
PaulC: of TimBray's umpteen theses, how many have we discussed/agreed to? not many, right? that seems to be at the heart of the matter.
19:21:45 [DanC]
PaulC: I'm considering attending an meeting, but I haven't managed to fit it in
19:22:35 [DanC]
Bray: I remain convinced that URNs are almost always a bad idea for XML namespaces.
19:22:39 [DanC]
Chris: because...?
19:22:56 [DanC]
TimBL: because either they're not resolvable or you're reinventing HTTP.
19:23:18 [Stuart]
19:23:26 [Stuart]
ack TBray
19:23:26 [Zakim]
TBray, you wanted to mention a liaison thing
19:23:30 [TBray]
TBray has joined #tagmem
19:24:03 [DanC]
DavidO: seems clear that we need more written justification for this position about URNs
19:24:32 [DanC]
19:24:44 [Stuart]
ack DanC
19:25:03 [tim-lex]
Pointer to or subject line of message on www-tag?
19:25:10 [tim-lex]
19:25:11 [TBray]
Paul's note at
19:25:35 [DanC]
DanC: I tried writing down exactly *why* I feel the way I do about URNs, but I didn't get [...]
19:25:36 [tim-lex]
19:26:30 [DanC]
[?]: ... Masinter on easier to make URNs resolvable than HTTP reliable...
19:26:48 [DanC]
DanC: I answered that in "accessible registries" Internet Draft (connected to issue 9)
19:27:35 [Chris]
q+ to talk about granularity, domains, and namespaces
19:27:38 [DanC]
TimBL: explaining this involves dealing with lots of backgrounds and perspectives...
19:28:11 [DanC]
[... more that I'm not sure how to summarize...]
19:28:18 [tim-lex]
Many of the problems with URNs for namespace names apply to many other prospecive uses.
19:28:35 [Stuart]
ack tim-lex
19:29:04 [Stuart]
ack Chris
19:29:04 [Zakim]
Chris, you wanted to talk about granularity, domains, and namespaces
19:29:15 [TBray]
19:29:26 [DanC]
Chris: if you own the domain, that's one thing, but if you go from geocities to [...?]
19:29:56 [DanC]
Bray: I offer to write about URNs and namespace names for; if the TAG could endorse that, we could send it to them.
19:30:30 [DanC]
PaulC: it should say something about the namespace 1.1 spec that gives URNs as example...
19:30:35 [DanC]
Bray: I'm not inclined to do that.
19:30:48 [DanC]
PaulC: but that's the whole point! we've got a REC that's leading folks to do this...
19:31:37 [DanC]
ACTION Bray: draft a TAG opinion on the use of URNs for namespace names, for review by the TAG
19:31:50 [DanC]
TimBL: there's these xri things too...
19:32:09 [DanC]
... we owe it to prevent this sort of fragmentation and re-invention
19:32:36 [DanC]
... lest the code need in browsers, caches, etc. need doubling
19:33:42 [DanC]
RoyF: I agree... the key is to [? something about syntax?]. Folks assume that because the specs say so, URNs will be persisitent. But persistence is a function of institutional commitment and frequency of use.
19:35:06 [DanC]
... e.g. if we could convince the u.s. library of congress to reserve part of their HTTP namespace, that would be more persistent than [most] URNs
19:35:23 [DanC]
<TBray> Paul's note at
19:35:35 [TBray]
Paul's note at
19:35:42 [DanC]
--- 2. Technical (60min)
19:35:50 [DanC]
---- 2.1 Issues the TAG intends to discuss
19:35:57 [DanC]
----- # IRIEverywhere-27
19:36:19 [DanC]
Bray: also issue 15, uri equivalence
19:36:43 [DanC]
Bray's proposal to close issue 27 and 15:
19:37:23 [DanC]
Bray: namespace 1.1. and XML schema...
19:37:39 [tim-lex]
q+ to express concern about closing the issue unless the identity issues are clear
19:37:47 [Stuart]
ack TBray
19:38:15 [DanC]
19:38:22 [DanC]
19:40:50 [Stuart]
19:40:58 [tim-lex]
ack DanC
19:40:59 [DanC]
ack DanC
19:41:01 [Stuart]
ack DanC
19:41:02 [tim-lex]
ack tim
19:41:02 [Zakim]
tim-lex, you wanted to express concern about closing the issue unless the identity issues are clear
19:41:28 [Chris]
19:42:29 [Stuart]
I'd also like some idea of whether we like what they have done at"> <>
19:42:54 [Chris]
I support what Dan said and say that namespace equivalence does what xml namespaces 1.1 says
19:43:32 [Chris]
19:43:33 [Norm]
19:43:47 [Stuart]
19:43:51 [DanC]
ack chris
19:43:59 [DanC]
TimBL: [something scribe didn't get]
19:45:29 [DanC]
Roy: the namespace spec shouldn't be saying "they're always different" but rather "canonicalization isn't required"
19:45:38 [TBray]
q+ to agree with Roy
19:46:02 [Stuart]
ack Norm
19:46:06 [Chris]
where do you stop?
19:46:28 [Stuart]
19:46:28 [DanC]
Norm: I observe lots and lots of deployed software that's consistent with the namespaces spec. Are you (timbl) proposing to say it's wrong?
19:46:28 [Chris]
does e-acute also compare to e followed by combining acute accent as well?
19:46:46 [Chris]
agree with norm.
19:46:54 [tim-lex]
There may be a lot of softwae which compares the simplictic way, but there is very little data which shows the difference
19:46:56 [tim-lex]
19:46:57 [DanC]
TimBL: yes, because even though the software does the wrong thing, there's very little data that goes there.
19:47:48 [DanC]
StuartW: namespace 1.1 spec also cautions users not to use IRIs until they mature
19:47:55 [DanC]
... in section 9
19:48:03 [DanC]
ack Stuart
19:48:08 [Chris]
we would be doing the xml community a great service by saying, as the TAG, that the 1.1 namespaces CR is correct and will not be held up on architectural grounds when it goes to PR
19:48:08 [DanC]
ack TBray
19:48:08 [Zakim]
TBray, you wanted to agree with Roy
19:48:13 [Stuart]
ack Stuart
19:48:45 [DanC]
Bray: I think section 9 is a worth attempt; I think it's perhaps a bit overspecified, since they'll have to change it.
19:49:27 [Chris]
Section 9 should say that the definition is not normative and defer to the eventual spec in the case of a conflict
19:49:48 [DanC]
Bray: I find some of the examples in 2.3 Comparing IRI References extremely distasteful, but I stipulate that lots of software works this way, so we might as well specify it this way.
19:50:10 [tim-lex]
q+ to argue that the strcpy code for handling namespace names will be replaced by generic URI handling in later versiosn of teh software
19:50:12 [TBray]
19:50:32 [Chris]
xml 1.1 namespaces could usefully update the Unicode 3.2 ref to unicode 4.0, though
19:50:46 [Stuart]
19:50:52 [DanC]
Bray offers write to www-tag saying yes, the IRI specs aren't stable, but yes, [scribe got confused]
19:51:08 [TBray]
write a one-para note saying:
19:51:16 [TBray]
yes, IRIs are the future and you should allow for their use
19:51:23 [TBray]
yes, the specs aren't stable and that's a problem
19:51:40 [TBray]
however, it has been done successfully e.g. XML1.0 and XSchemas
19:51:44 [DaveO]
DaveO has joined #tagmem
19:51:44 [Chris]
and can we say that xml namespaces 1.1 is okay, too?
19:51:48 [Chris]
19:52:14 [Chris]
in that list of 'sucessfully'
19:52:54 [TBray]
don't think we need to in order to resolve this issue
19:52:55 [Chris]
19:53:07 [Chris]
9 is fine with corrections as I noted above
19:53:35 [Zakim]
19:53:36 [Zakim]
19:53:54 [tim-lex]
Zakim, ??P7 s RoyF
19:53:55 [Zakim]
I don't understand '??P7 s RoyF', tim-lex
19:53:59 [tim-lex]
Zakim, ??P7 is RoyF
19:53:59 [Zakim]
+RoyF; got it
19:54:06 [DanC]
DanC: we will either endorse or say it's wrong or we will not have closed the issue to my satisfaction
19:55:54 [Roy]
Roy has joined #tagmem
19:56:18 [Roy]
19:56:35 [Roy]
19:56:37 [DanC]
Bray: I think we can close issue 27, not issue 15, without completely endorsing section 2 of the namespaces CR. [not sure I got that right]
19:56:52 [Stuart]
ack tim-lex
19:56:52 [Zakim]
tim-lex, you wanted to argue that the strcpy code for handling namespace names will be replaced by generic URI handling in later versiosn of teh software
19:57:05 [DanC]
ack DanC
19:58:12 [Stuart]
19:59:09 [DanC]
[... disagreement between TimBL and ChrisL about the correctness or otherwise of section 9 of ...]
19:59:42 [Chris]
well we are both saying its correct but we put different implications on what it says
20:00:01 [Chris]
I say that you convert IRI to URI purely to defereence it
20:00:54 [DanC]
Roy: this %xx stuff doesn't work for non-ascii characters in hostnames.
20:01:03 [Chris]
timbl says that you can convert it to a URI and it has the same identity, which it does not
20:01:13 [Chris]
NOTE** to martin Duerst ;-)
20:01:14 [DanC]
Bray: section 9, last OL is wrong...
20:01:19 [Chris]
escaping all additional characters is wrong
20:01:25 [Chris]
because of IDNA
20:01:27 [DanC]
TimBL: ... conflicts with IDN specs
20:01:34 [tim-lex]
TimBL contends that section 9 is good and resolves issue 27 but point out that it makes section 2 wrong.
20:01:39 [Roy]
ack Roy
20:01:47 [Stuart]
ack Stuart
20:02:15 [Chris]
CL asserts that section 2 is correct and section 9 is mostly cotrrect except as noted, but does not support the meaning that TimBL asserts
20:02:25 [Chris]
for the specific case of xml namespaces
20:02:39 [Chris]
which is, after all, what the spec is *about*
20:03:12 [DanC]
TimBL: I expect strcmp to get replaced in libraries and shared code by general URI parsing/comparison routines
20:06:19 [Chris]
TimBL keeps saying they are inconsistent but they are not
20:06:38 [Chris]
"The IRI references below are also all different for the purposes of identifying namespaces:" is correct
20:06:48 [tim-lex]
- Chris
20:07:22 [Chris]
it does not say "The IRI references below are also all different for the purposes of identifying namespaces even if you randomly decide to convert them to URIs and assume they are still the same:"
20:08:11 [Chris]
it also says, byw "The Candidate Recommendation review period ends at 2359 UTC on 14 February 2003."
20:08:19 [DanC]
Stuart: one option is to not close this, let the IRI spec mature, and not offer guidance meanwhile
20:08:42 [Chris]
so, does the namespaces 1.1 spec wait until IRI matures as well?
20:08:43 [DanC]
they can set a due date for comments, but the CR period ends when the Director decides.
20:08:47 [Chris]
infinite back burner?
20:10:49 [Norm]
Nonne fo these issues are new, however, they already exist. No?
20:10:54 [Norm]
s/Nonne fo/None of/
20:11:17 [DanC]
ACTION TimBL: explain how existing specs are inconsisten [or something?]
20:13:17 [DanC]
Stuart: I remain of the opinion that getting the IRI spec is the right place to get the details right, which involves TAG members commenting on the iri list
20:14:46 [DanC]
[... discussion of postponement of parts or all of the spec...]
20:14:53 [DanC]
[er... of the issue]
20:15:33 [Stuart]
20:26:49 [DanC]
Bray: I don't think you can write code that's conformant to RFC2396bis and namespaces CR1.1 both at the same time.
20:27:29 [DanC]
Stuart: I'm likely to schedule this lower on the agenda in future meetings, so it doesn't put other things further at risk
20:27:34 [Zakim]
20:28:00 [Zakim]
20:28:05 [DanC]
20:28:24 [Zakim]
20:28:26 [Zakim]
20:28:28 [Zakim]
20:28:29 [Zakim]
20:28:29 [Zakim]
20:28:38 [Roy]
Roy has left #tagmem
20:28:39 [Zakim]
20:28:54 [Norm]
Doing anything but strcmp just looks impossibly hard
20:29:51 [DanC]
strcmp is necessary and sufficient and cheap and works and and and... ok, so the fact that %7e differs from %7E is distasteful. Who cares?
20:33:43 [Norm]
Hmm. Tim, apparently, but he's left :-(
20:35:00 [Zakim]
disconnecting the lone participant, Chris, in TAG_Weekly()2:30PM
20:35:01 [Zakim]
TAG_Weekly()2:30PM has ended
20:42:29 [Norm]
Norm has left #tagmem
20:55:37 [DanC]
RRSAgent, bye
20:55:37 [RRSAgent]
I see 2 open action items:
20:55:37 [RRSAgent]
ACTION: Bray to draft a TAG opinion on the use of URNs for namespace names, for review by the TAG [1]
20:55:37 [RRSAgent]
recorded in
20:55:37 [RRSAgent]
ACTION: TimBL to explain how existing specs are inconsisten [or something?] [2]
20:55:37 [RRSAgent]
recorded in