[01:00:26] YATSUGATAKE joins the room [01:03:15] YATSUGATAKE leaves the room [01:14:18] YATSUGATAKE joins the room [02:39:24] YATSUGATAKE leaves the room [04:20:53] YATSUGATAKE joins the room [04:23:42] YATSUGATAKE leaves the room [04:24:08] YATSUGATAKE joins the room [04:24:55] YATSUGATAKE is now known as MartinJDürst [06:11:21] Yoshiro Yoneya joins the room [06:13:00] fujiwara24 joins the room [06:13:50] resnick joins the room [06:14:00] MartinJDürst leaves the room [06:14:41] I will be jabbering. [06:15:53] YATSUGATAKE joins the room [06:18:49] Julian joins the room [06:20:59] Ted "Notes well" [06:21:09] Barry Leiba joins the room [06:21:27] Intro [06:21:49] Agenda [06:21:55] ROOM: audio doesn't appear to work [06:22:11] Dave Thaler joins the room [06:22:17] YATSUGATAKE is Martin J. Dürst [06:22:24] Julian, do you hear Pete? [06:22:30] Larry Masinter is on the agenda, but you are asked to participate. [06:22:34] I hear lots of mumbling [06:22:44] =JeffH joins the room [06:22:44] "bluesheet" [06:22:54] You are in the wrong room I think. [06:23:04] This is Cattleyea EAST. Are you in the right one? [06:23:18] Yoshiro Yoneya leaves the room [06:23:25] frodek joins the room [06:23:30] Yoshiro Yoneya joins the room [06:23:31] kariem joins the room [06:23:40] it appears I clicked the wrong link. sorry. [06:23:56] Randall Gellens joins the room [06:24:13] John C Klensin joins the room [06:25:24] martin.thomson joins the room [06:25:48] Technical difficulties, please stand by. [06:27:01] Takehito Akagiri joins the room [06:27:47] Ted: Intro: Attempting to restore interoperability [06:29:12] Slides here: IRI Slides [06:29:33] Ted: Trying to bring together communities. [06:29:49] Coordination is the goal. [06:30:13] Klensin: Counter-theory - IRI's themselves, as i18n tool, is broken. [06:30:21] doug.mtview joins the room [06:31:06] For end users, mixing URI syntax with i18n stuff turns out bad. [06:31:41] History slide: [06:31:44] Larry: [06:32:14] to John: (for ASCII syntax characters in IRIs) not on keyboard, yes in some cases. Not displayable, essentially never! [06:32:55] Long history since 1993. [06:33:11] also to John: Do you have anything better? [06:33:31] Document review slide [06:34:21] Proposal: 3 documents iri-bis, mailto-bis, and update to 4395 [06:35:15] Everything else is proposed to be out of scope. [06:35:26] 3 issues: [06:35:43] IRI as protocol element vs. IRI->URI [06:35:58] tonyhansen joins the room [06:36:10] tonyhansen leaves the room [06:38:52] YATSUGATAKE... I agree with "essentually never" in, e.g., east Asia. The problem is much worse in, e.g., some areas using Arabic script. Wrt your question, I think we should be looking at content identifiers that are fully localizable, rather than trying to stuff some "international" characters into the user-unfriendly and syntactically over-sensitive URL/URI format. People do have solutions to the problem for which IRIs (or even raw domain names) will work, but those solutions tend to assume that there is only one protocol identifier ("http") and no need for delimiters, a tail, or other syntax bits. That is very bad for a multiprotocol Internet and, if IRI are examined in that context, they just aren't needed at all (the domain names are sufficient). [06:39:21] Joe Hildebrand joins the room [06:39:27] Did you intend to put that into the room, jck? [06:39:40] Yep [06:39:44] k [06:40:06] Second issue: IDNA reference. [06:40:24] Third issue: "liberal processing" has been going on for some time. [06:40:51] tonyhansen joins the room [06:41:34] tonyhansen leaves the room [06:42:04] John: are you saying that if you have IDNs you don't need non-ASCII in trailing parts (e.g., path, user name)? [06:42:04] tonyhansen joins the room [06:42:14] Review of doc & committee slide [06:42:14] marcos@de joins the room [06:42:55] Randall: Good question! [06:43:04] tonyhansen leaves the room [06:44:41] (Remote participants: If you want to say something to the microphone, please put "mic:" in your comment.) [06:46:25] longer list of issues: http://lists.w3.org/Archives/Public/public-iri/2009Oct/0009.html [06:46:44] Opening the mic. [06:47:14] Joe Hildebrand leaves the room: Disconnected. [06:47:18] Martin Dürst: Still need to make sure we have a well defined conversion between URI and IRI. [06:48:27] tonyhansen joins the room [06:48:34] Melnikov: Is there a difference on the wire? [06:48:42] Randy: certainly not. I think we need something that can identify content, not just hosts, which implies one needs the effective functionality of URIs. And I don't think i18n NAPTR records pointing to ASCII URIs are the answer, either. But, as I've wandered around the world in the last few years, I keep running into the mixed-ASCII-and-local problem that IRIs cause and local populations who say "well, we have avoided that problem by avoiding the use of anything but domain names". I don't think that scales, I think it locks out other protocols, and I think it ultimately regionalizes the Internet. [06:49:50] Masinter: in the previous documents, host names would have been UTF-8 hex encoded passed on as a URI. [06:49:53] gkanai@gmail.com joins the room [06:50:06] It was optional that one could make it IDN. [06:50:31] This could be bad. [06:52:40] Dürst: Need additional input on what to do with domain names. [06:53:01] John: If we can't accept IDN-with-the-rest-ASCII, and given the problems with IRIs, what do we do? [06:53:08] Klensin: DIscussion during plenary Thursday. Please come. [06:54:24] Ted: Followup: How would we get rid of the "second encoding". [06:54:52] Klensin: Stop digging the hole, or go dig some other hole. [06:55:27] Hildebrand: There should be a null encoding. [06:56:05] Chair: What's that? [06:56:17] Hildebrand: UTF-8? [06:56:27] Masinter, that's going to be in there. [06:57:13] Cheshire: Definitely a plenary topic. No such thing as a null encoding. [06:57:46] Masinter: Null encoding is only theoretical. [06:58:02] stpeter joins the room [06:58:21] Roach: What's the interaction with IDNA? [06:58:27] Masinter: Coordinate with IDNA. [06:58:57] Martin Dürst at the mic [06:59:00] Joe Hildebrand joins the room [06:59:13] Dürst: Agree that two representations is bad. However, domain appears in multiple places in IRIs [06:59:43] hmm, the NAZGUL WG? [06:59:44] Hardie: Current state of affairs is messy. [07:00:09] What I was trying to say is that XML Infoset does more or less define an encoding-free view over unicode, does it not? [07:00:30] If we did something new, it doesn't help everything else out there. [07:01:45] Leaving it alone is not a choice. [07:01:56] Michael Smith at the mic [07:01:59] Michael Smith: [07:02:05] (jinx) [07:02:08] Normalized Addresses for Zones Graded in Unicode Languages [07:02:12] if you double the u, cause we're in Japan, NAZGUUL: never assume zee grunts'll use uri legitimately [07:02:45] Smith: HTML5 includes text on "URL" processing. [07:03:58] Want something that can be normatively reference in HTML5 that documents browser behavior. [07:04:38] Ted: Discussion limited to http? [07:04:43] Smith: Yup. [07:05:47] Stuff from the goals page should be in the charter. [07:06:14] TPLUNKE joins the room [07:06:24] TPLUNKE leaves the room [07:06:28] Klensin: Agree with Ted, but because URI is icky for users, we're going to have to move away from the base. [07:07:07] Because of the multiple encoding problem and moving away from base, an OK scope would be to move away from ASCII-only URIs. [07:07:25] Chair: Are folks interested in taking this on. [07:07:46] Klensin: Clarification: Could do it protocol by protocol. [07:08:41] Could also do it all. [07:09:06] Dürst: Willing to put in a bit more work, but I do have a day job. [07:09:39] marc.blanchet.qc joins the room [07:13:00] Dürst: we don't need i18n for things like IP addresses (if there were an ip: scheme) [07:13:02] presnick joins the room [07:13:16] Back. [07:13:19] presnick: WB [07:13:31] Ted: Bringing all of the URI schemes up to date burned out people. This is a bigger job. [07:13:40] (Missed what Martin said.) [07:14:18] where is Lower Slabovia? [07:14:21] Klensin: If we had a URI scheme which identified IP addresses and go down this path, we'll end up with internationalized digits. [07:14:47] Leiba: Why isn't this just a presentation issue? [07:15:00] Ted: Failed theory. It's actually deployed as a protocol element. [07:15:46] Martin said: don't need i18n for URI/IRI for e.g. IP addresses. But need for mailto: (not currently in RFC, but proposed in draft), e.g. for Subject: [07:16:01] resnick leaves the room [07:16:10] Roach: Doing this for SIP would be really bad. [07:16:43] marcos@de leaves the room [07:16:45] marcos@de joins the room [07:17:01] Not clear that you can include % encoding. [07:17:20] For registry, it's always a Zipf law: Some are very important, very well kept, others are rarely used, not very well registered,... [07:17:51] Masinter: Make a framework for new IRIs. [07:17:57] erikvanderpoel joins the room [07:18:34] Ted: Would love to force people to layer presentation, but can't. [07:19:05] Also, attempt at human-friendly names did not go well. [07:19:28] Protocol elements bleed into human side of things. But we can't stay so liberal. [07:20:26] Masinter: Trying to keep bad things from happening to humans who use this unwittingly should be a priority. [07:23:17] Masinter: we don't need wholesale changes, only some fixes near the edges [07:23:31] Resnick: time check [07:23:32] More than halfway through! [07:23:39] Session ends 17:00! [07:23:42] Resnick at the mic (not in the chair) [07:24:26] Resnick: the presentation layer leaks into protocol elements, but IRIs currently are not encoded [07:24:45] Resnick: IRIs provide i18n, but so do percent-encoding [07:25:17] Resnick: we shouldn't conflate encoding with internationalization [07:26:05] Resnick: we need to be careful about terminology [07:26:23] Resnick: do you want this to be represented in any encoding? [07:28:08] [sorry, missed what Leiba said] [07:28:45] Hardie at chair: we trying to get to a point where the number of translations is minimized, not that there is only one encoding or representation [07:29:13] Leiba said that occasional appearances of IRIs in HTML docs aren't protocol elements [07:29:40] They turn into protocol elements once a e.g. browser parses them and handle them over to oder layers [07:30:33] Masinter: distinction between protocol and protocol elements [07:32:11] ian joins the room [07:32:32] Masinter: can we focus on changes to IRI, mailto, and BCP for i18n? [07:33:06] presnick leaves the room [07:33:06] Klensin: I think we can't limit it to just those three, scope might grow based on analysis [07:33:39] Klensin: I'm uncomfortable with working on mailto here, especially given the existence of EAI [07:34:07] presnick joins the room [07:34:09] presnick leaves the room [07:34:26] Klensin: requires knowledge of email [07:34:47] Dürst: I am not an expert on email and what can go into email addresses [07:35:15] Randall Gellens leaves the room [07:35:24] Randall Gellens joins the room [07:35:44] frodek leaves the room [07:35:56] Randall Gellens leaves the room [07:35:59] Melnikov: I'm conflicted ... (1) happy about reducing scope (2) the mailto / EAI interaction needs to be addressed (3) if this WG is chartered only for http and not other schemes the solutions might not be generic [07:37:04] stpeter is now known as πßå [07:37:22] <πßå> might as well use some non-ASCII characters :) [07:37:49] presnick joins the room [07:38:03] <πßå> Klensin: unwise to use only one scheme as an example, 2 or 3 would be good [07:38:24] <πßå> Klensin: mailto might be one of the worst examples [07:38:33] <πßå> Hardie: maybe look at URNs [07:39:25] <πßå> Dürst: RFC 3987 has references to pop: and imap: and data: etc. [07:39:40] <πßå> Dürst: agree that it's extremely important to look at other schemes [07:40:32] marcos@de leaves the room [07:42:52] <πßå> Resnick: 5 more minutes on general issues [07:43:35] <πßå> Hardie: Question 1, are we all convinced that there is a problem here for the IETF to solve, with the alternative unstated [07:43:45] <πßå> Hardie: yes? [07:43:57] <πßå> hums [07:43:58] <πßå> no? [07:44:01] <πßå> no hums [07:44:06] <πßå> not enough info? [07:44:08] <πßå> no hums [07:44:11] πßå is now known as stpeter [07:44:27] ian raises his hand [07:45:41] [hum formation in process] [07:46:08] Hardie had asked whether people would join a mailing list and review documents [07:47:12] ian hums [07:47:24] (that was humming to yes we could start over) [07:49:35] should the charter include an explicit list of schemes to review? [07:49:56] larry hums close to the mike :-) [07:50:27] [would this include a list that would be at a minimum the following] [07:50:30] marcos@de joins the room [07:50:36] Lisa at the mic [07:52:39] re-hum [07:54:04] Barry Leiba leaves the room [07:54:13] Joe Hildebrand leaves the room: Disconnected. [07:54:22] doug.mtview leaves the room [07:54:25] tonyhansen leaves the room [07:54:33] Yoshiro Yoneya leaves the room [07:54:36] ian leaves the room [07:54:36] kariem leaves the room [07:54:40] presnick leaves the room [07:54:47] Julian leaves the room [07:55:01] martin.thomson leaves the room [07:56:29] gkanai@gmail.com leaves the room [07:56:50] fujiwara24 leaves the room [07:58:45] =JeffH leaves the room [08:01:28] marc.blanchet.qc leaves the room [08:04:11] erikvanderpoel leaves the room [08:10:51] stpeter leaves the room [08:12:43] Takehito Akagiri leaves the room [08:24:09] marcos@de leaves the room [08:28:51] John C Klensin leaves the room [08:28:57] Dave Thaler leaves the room [08:42:40] YATSUGATAKE leaves the room [21:45:15] Joe Hildebrand joins the room [22:12:43] Joe Hildebrand leaves the room