[02:26:22] Dürst leaves the room [02:27:39] Dürst joins the room [03:35:51] Dürst leaves the room [03:38:44] Dürst joins the room [05:28:48] cabala joins the room [05:29:01] cabala leaves the room [06:52:58] yone joins the room [06:54:35] Chris Waigl joins the room [06:56:12] Chris Waigl leaves the room [07:01:10] cary joins the room [07:04:12] Dave Thaler joins the room [07:09:16] tony.l.hansen joins the room [07:09:58] I'm just waiting for Ted or Marc to connect again. [07:10:28] resnick joins the room [07:10:31] ywang830 joins the room [07:11:47] Is anyone jabber scribing? I'm in another WG meeting right now, but hope to be over in iri later [07:12:18] Still working on audio problems. (connecting Skype to local audio system) [07:13:49] psa joins the room [07:18:07] Martin, have you been able to hear us on Skype? [07:18:30] earlier, yes. Recently, no. [07:18:44] Can you hear us now? [07:18:54] Now Ted has me on hold. [07:19:09] No longer on hold, but line seems to be dead. [07:19:25] Julian joins the room [07:19:30] We can hear you. [07:19:36] You obviously can't hear us. [07:19:55] Ah, now that's new. That's the reverse of what we had before. [07:20:07] Marc is now speaking. [07:20:21] So I could actually give my presentation? It would be difficult to answer questions, though. [07:20:27] Not hearing anybody at all. [07:20:33] :-( [07:20:37] But you might hear my typing. [07:20:39] joseph.yee joins the room [07:20:41] We do. [07:20:59] Please ask Joseph how he did it yesterday with John Klensin. [07:21:19] Are you sure your speaker is on? [07:21:20] that worked fine, even with both me and John skyped in. [07:21:41] Yes, my speaker is on. [07:21:48] Didn't change anything there. [07:22:22] Start your presentation. We'll type to you. [07:22:51] Dürst: can you listen to the audio stream [07:22:52] You can listen on the stream, but there will be a delay. [07:23:27] right [07:23:31] 2 hands or so. [07:23:39] of how many? [07:23:42] 5 or 6 [07:23:50] Out of 20 or so. [07:24:03] Dave Thaler leaves the room [07:24:18] Dave Thaler joins the room [07:26:52] We'll do them in the chat room. [07:26:56] Any questions? [07:27:00] No questions. [07:27:02] go [07:27:51] Thomas Roessler joins the room [07:28:58] reviews requested for http://trac.tools.ietf.org/wg/iri/trac/ticket/35 and http://trac.tools.ietf.org/wg/iri/trac/ticket/18 [07:29:46] questions? [07:29:48] Hold.. [07:29:57] Larry: [07:30:09] Was there any question on the list about the closed issues? [07:30:35] Was there any discussion? [07:31:56] Larry back to mic [07:32:14] A lot of these look easy and it's fine. Were there any where you had any doubts about these? [07:32:34] Jacky Yao (Health Yao) joins the room [07:32:46] Laugh. :-) [07:33:12] No other questions. [07:33:14] go [07:33:18] hardie@jabber.psg.com joins the room [07:36:12] we see two experts [07:36:17] :) [07:36:24] I can review for XMPP [07:36:24] ADs [07:36:33] Alexey for IMAP [07:36:39] OK, we'll be checking. [07:36:48] PSA Question: [07:36:52] peter is at the microphone [07:37:17] Is this only about the inclusion of these IRIs in docs, or does this apply in other contexts? [07:37:59] (Larry at the mic when you are done) [07:38:07] Larry speaks: [07:38:38] If there is no know doc encoding, then there is no ambiguity. So the only time there is special processing is when it is contained in something with document encoding. [07:38:44] continue [07:38:49] Larry. [07:38:52] More. [07:38:57] hold martin. [07:39:11] HOLD MARTIN. [07:39:33] Thanks. [07:39:34] alexey.melnikov joins the room [07:39:38] Julian at mic [07:40:03] had discussion about XHR, it hardcodes to UTF-8 [07:40:37] @martin. Go. [07:41:34] (I believe this is http://www.w3.org/TR/XMLHttpRequest/, Section 3.6.1, item 6) [07:41:40] The query part of IMAP URIs contains an IMAP SEARCH criteria, which includes the charset parameter. So this doesn't have to be UTF-8 :-( [07:41:41] in chatroom only: other possible contexts include instant messages (XMPP is all UTF-8, but non-XMPP IM technologies are not so sophisticated), email messages, etc. [07:42:19] no feedback in room at this time [07:42:20] no feedback [07:42:24] hold on [07:42:27] Larry to the mic [07:42:45] Marc says: We need resolution to move forward. [07:42:47] Marc as chair: we need to move forward, so we need resolution [07:42:50] :) [07:42:54] Larry speaks [07:43:10] psa assumes that Martin can de-dup chatroom messages [07:43:28] de-dup??? [07:43:39] counterproposal to build API [07:43:41] Goal was to allow HTML to ref IRI doc. But HTML WG did not accept. [07:43:47] Ref: http://www.w3.org/html/wg/tracker/issues/56 [07:43:50] http://lists.w3.org/Archives/Public/public-html/2010Apr/0147.html [07:43:59] Counter proposal for API to parse into parts. [07:44:06] [07:44:20] also: http://lists.w3.org/Archives/Public/public-html/2011Mar/0124.html [07:44:30] But maybe we can get rid of the reference without impeding progress. [07:44:30] (de-dup: if two people type the same thing in different words, you can figure out that the meaning of what we're trying to convey) [07:44:53] (Martin: Say that you want to speak in the chat before starting again) [07:44:55] yes, I can de-dup [07:45:05] Larry notes that the motivation in this document may be gone, since other ways of approaching the problem seems to be what will happen. [07:45:17] So taking it out may be harmless and would help avoid delay [07:45:19] My hand is up [07:45:38] OK, go [07:45:39] go [07:47:12] (Larry goes to mic. When you are done.) [07:47:22] Larry speaks: [07:47:50] There was no feedback that the current draft did not address their needs. It was just that the *schedule* did not address their needs. [07:47:58] go martin. [07:49:02] Hold for Thomas [07:49:44] Marc says: Must reword text, because the initial assumption is gone. [07:50:02] We'll discuss next steps at the end. [07:50:08] Martin. Go to #11. [07:50:28] we do need to discuss next steps, so let's make sure we have time for that discussion (we started 20 minutes late) [07:50:54] slide 12 [07:51:03] (bidi) [07:53:00] No. go. [07:53:00] no comments [07:53:08] slide 13 [07:53:43] no. go. [07:54:30] questions? [07:54:33] psa: [07:54:37] St. Peter at the mic [07:55:02] Martin's slide "except trivial"--is anything not mentioned here trivial? [07:55:37] (there are more issues in the tracker -- are they trivial if you did *not* include them in your slides?) [07:55:55] Marc: how resolve? [07:56:06] Martin? [07:56:14] Answer Marc. [07:56:58] (people smile) [07:57:14] So, let me suggest a base question: do we currently see any truly blocking issues at this point? Such that worrying about the low hanging fruit is not the right notion? [07:57:32] Martin: Answer. [07:58:24] (for the notes, Martin says that one issue is the bidi handling, the other thing is what to do with HTML compatibility) [07:58:41] Marc asks: THe only blocking is bidi? [07:58:55] Larry at the mic [07:59:27] Larry says that he doesn't think there is anything block, but some difficult ones. [07:59:37] Jacky Yao (Health Yao) leaves the room [07:59:38] The first of these is the lack of connection with implementations [07:59:54] Speculation replaces knowledge--easy to get shot down [08:00:00] @marc asks how we can resovle that? [08:00:11] Larry: we get them to participate or we give up? [08:00:37] Larry has two other things: handling of domain names in the IDNA version transition [08:00:46] And how IRIs sort out their reference in a consistent way [08:00:55] (difficult wordsmithing issue, more than something else) [08:01:01] "IRIs sort out their reference"? [08:01:47] BTW, I received an email from Adil Allawi regarding bidi, he offered to help with that, perhaps by writing a separate "bidi in IRI" Internet-Draft [08:02:13] When was that (the mail from Adil)? [08:02:38] A few days ago, privately. [08:03:22] who has the mic? [08:03:36] Marc has the mic. [08:03:47] (Raise hand if wanted. I'll say go.) [08:03:56] ok [08:04:03] Marc asks who knows the implementors? [08:04:20] PSA asks who are the customers for the work? How do we track them down? [08:04:26] Folks who write URI parsers? [08:04:37] I know Anne (Opera) and Shawn (Microsoft). They are not imlementors, but they are the contacts to implementers. [08:04:43] Larry says: Browsers (e.g. IE), [08:04:50] Firefox [08:04:53] Chrome [08:05:00] We also know Adam, but he is very busy with security. [08:05:24] Thomas at the mic [08:05:30] My guess is that it's not that the implementers are not interested, it's just that they are holding back to let somebody else do the work. [08:05:52] It seems IRIs/URIs don't have that high a priority for them. [08:06:06] This discussion leads to scheduling. [08:06:18] THere's tons of very fine details in HTML5, but for URIs/IRIs, it's just very general. [08:06:21] HTML5 said "remove ref to IRI" because it is too slow. [08:06:23] Thomas is summarizing HTML5 [08:06:33] I agree we should pick up the discussion again for scheduling. [08:06:34] HTML5 is being stern about schedule [08:06:41] What does that mean for a "living doc"like HTML5? [08:06:43] the issue that was open is now closed [08:06:53] hardie@jabber.psg.com: only some higher power knows [08:07:35] If you're herding cats, being stern about schedule is one way to do it. [08:07:43] Overall window is about 6 months. [08:08:13] We need to figure out way forward this meeting or very shortly after that has buy-in from browser folks and takes into account HTML5. [08:08:18] They are going to last call in May, according to their schedule. [08:08:35] They might be innundated with last call comments. [08:08:50] context about Adam Barth's spec: http://lists.w3.org/Archives/Public/public-iri/2010Sep/0000.html [08:08:56] Marc speaking [08:09:05] Roy at the mic line, Julian behind him [08:09:25] Marc says we need to get feedback from that community. [08:09:30] Roy speaks: [08:09:30] Marc wonders if we can get the browser guys involved [08:09:34] Roy at the mic [08:09:54] There isn't an IRI parser in browsers. There's a location bar parser, etc. [08:10:08] On the IETF side, we approach as "interoperability results". [08:10:15] So there's a disconnect. [08:10:32] Would prefer if W3C or HTML5 deals with the browser issues. [08:10:36] Re Adam's draft: it was sent out last September; there was feedback, but never any followup. (so essentially Adam walked away) [08:10:43] Prefer IETF to deal with what's on the wire. [08:11:08] mnot joins the room [08:11:52] Way to move forward is to remove the parsing references bits from the spec and work with the HTML5 community to do that, and try to use a term other than "references". [08:11:59] PSA to mic: [08:12:10] Pete: Please tell me when meeting is moving to next draft [08:12:32] FYI, Adam Barth on audio but having problems with references. [08:12:50] "Roy, what needs to be in the HTML5 document? What kind of reference do they need?" [08:12:51] (trying to listen to audio feed via separate computer) [08:12:52] gah [08:12:54] s/references/jabber/ [08:13:35] Roy: Browsers do not use IRIs. The deal in URIs. They use IRI mechanisms to provide display format for location and status bar. [08:13:56] Don't know if they follow IRI spec or they're just going to UTF-8. [08:14:19] Julian: They closed the issue as a pre-last call issue, but could be opened after last call if there are new issues. [08:14:42] ...new information..., such as new specs. [08:15:11] (some else please jabber that) [08:15:36] PSA: perhaps an IRI is at the presentation layer, but when its ove the wire it's a URI. [08:15:51] (That was the description of the IRI in the previous doc) [08:15:59] Larry, Julian, Thomas at the mic now. [08:16:05] There are quite a few IRIs in HTML documents, and those go over the wire. [08:16:23] Julian: these are leaking into hxr [08:16:28] needs the same handling [08:16:38] Larry: wishes it were true, but it's not [08:16:52] What does Larry wish it were true? [08:16:55] In lots of different instances, this is not true. The handling of IDNA [08:17:03] (that IRI was only a presentation element) [08:17:15] Above was an answer to you, martin. [08:17:44] Larry is describing how the IRIs get parsed and interpreted independently. [08:18:18] I can now listen to Larry on a separate computer. But I don't know how long the delay of the audio feed is. [08:18:31] It's usually a few seconds [08:18:44] Adam says: HTML5 needs parse and address algorithm (breaking out the different syntactic pieces of an IRI) and then resolving a relative reference [08:19:14] RFC 3986 staes how to break down the components: http://greenbytes.de/tech/webdav/rfc3986.html#rfc.section.B [08:19:22] resnick at the mic [08:20:16] resnick says, sounds like the current state of affairs is broken, sometimes IRIs are used as protocol elements and sometimes parsed into pieces [08:20:58] many people at the mic [08:21:01] Roy at mic [08:21:26] Roy says URI parsing ref algo and IRI parsing ref algo are essentially identical [08:21:47] URI spec assumes that you will pass to resolver [08:22:12] how get from hostname to name you will resolve via DNS? [08:22:19] pass domain name to resolver [08:22:19] Larry speaking [08:22:55] "I was starting to say 'presentation of iri' vs 'iri'" [08:23:11] abarth joins the room [08:23:23] my hand is up [08:23:36] i succeed at jabber! [08:23:42] Dürst: ok, just a sec [08:23:45] hi adam [08:23:47] Martin will come after Ted. [08:23:50] hi [08:23:55] John C Klensin joins the room [08:23:57] abarth: probably a gmail issue [08:24:05] ok [08:24:07] Ted at mic [08:24:12] then Martin [08:26:04] Martin, go [08:26:15] Martin go. [08:26:26] my hand is down [08:26:34] OK, back to Roy at Mic. [08:26:35] we didn't hear anything from you, Martin [08:26:39] OK, go martin. [08:26:47] OK, got it [08:26:50] Larry speaking: [08:26:51] bkihara.l joins the room [08:27:11] The delay is something like 10 seconds or so. [08:27:29] Jacky Yao (Health Yao) joins the room [08:28:23] So you are hearing through the stream Martin? [08:28:50] yes, it's better than not hearing anything [08:29:11] resnick to the mic [08:33:04] Dave Thaler leaves the room [08:33:10] Dave Thaler joins the room [08:40:46] Ted to the mic [08:41:40] are you going to deal with RFC4395bis? [08:43:12] to Ted: the text you just read is from last evening. It's back from RFC 3987, it was different in -04, but there was strong support at the last IETF to change it towards what we have now. [08:44:52] I think there is also some disconnect in that the HTML5 people (-Adam) were expecting 'us' to do it, and we were hoping to bring them in. [08:46:20] mnot leaves the room [08:48:03] Action item Thomas to set up a teleconference within a few weeks to have HTML5 and IRI WG exponents talk together. [08:50:01] Will this be a WG conference call, or just a 'design team' conference call (for which we wouldn't need the two weeks, as far as I understand) [08:50:44] @martin you should be able to talk now. [08:51:01] I'm waiting for the WG to get to RFC 4395bis [08:51:16] try to say something just to confirm the mic [08:52:01] hardie@jabber.psg.com thinks the interim has to be face-to-face to drive this. [08:52:30] @martin any other comments? [08:52:34] cary leaves the room [08:52:35] psa thinks hardie might be right but that's a challenge on short notice [08:52:56] interim face-to-face ain't gonna work [08:53:02] Thomas Roessler: I don't think so either [08:53:17] at least not the short-term one -- perhaps one a bit farther out [08:53:44] Hold one sec [08:54:04] OK, go. [08:56:31] http://trac.tools.ietf.org/wg/iri/trac/ticket/63 [08:57:28] question from Marc [08:57:31] (at the mic) [09:00:48] Thomas Roessler leaves the room [09:01:53] I'm fine with Dave's suggestion, and Larry said he was as well. Option 1 on slide 4 sounds good. [09:03:29] I didn't follow that Martin. [09:03:41] Julian prefers choice 3 [09:04:13] Dave, the syntax might allow something like %00, because you want to transport that as data, but wouldn't allow this as actual raw data. [09:04:29] hold on [09:04:42] wait martin [09:05:45] Listen to room. Dave will restate question. [09:06:36] And listen to Julian too. [09:07:13] There's some grammatical error in (3). There's no 'or' that fits the 'either'. [09:07:42] Now Larry. [09:08:08] Larry refers to http://trac.tools.ietf.org/wg/iri/trac/ticket/48 [09:08:27] To quite some extent, scheme definitions are paper exercises. There are not too many software components that actually check scheme-specific syntax. [09:08:44] in your example, I think that should be disallowed (which is option 1), which is to say that there's no IRI without a URI representation and vice versa [09:08:54] And back to Julian [09:09:02] that = @Martin's jabber note about %00 [09:10:16] to Dave: yes, we want to make sure there's no IRI without a corresponding URI, but the example is the other way round. [09:10:27] thats why I said "and vice versa" :) [09:11:15] But it isn't described if one doesn't recognize the scheme or a provisional registration is so fuzzy as to be equivalent to that. So either Dave is right or this is terminally wedged. [09:11:50] John, did you want that at the mic? [09:11:51] @Ted: +1 (or several) [09:12:27] Marc is thinking. [09:12:29] @Ted: If you think it would be helpful, yes. I'm feeling pretty depressed about the whole direction of this conversation and the WG. [09:12:37] some conversation about #3, it should read: (3) Require in 4395bis that any scheme definition that defines the scheme on the IRI level either include the necessary percent-encoding (explicitly) or by a general provision such as in (1) above. [09:12:39] to Dave: you also need %-encoding for cases such as a literal '%' in a mail address that goes into a mailto: URI/IRI, and so on. [09:12:47] yep [09:12:51] But that's an issue separate from IRIs. [09:13:01] People in favor of 1 humming. [09:13:06] Hi John, I think we're going to move past it. [09:13:08] If you want me to continue, just say so here. [09:13:16] People in favor of 3 humming. [09:13:20] consensus for #1 [09:13:22] @Ted: wmf [09:13:28] Martin, go. [09:13:29] or wfm [09:15:44] Adam: your opportunity to propose "URL" [09:16:31] I think the authors all agree with you. [09:16:34] Nodding in agreement. [09:16:41] Go ahead. [09:17:54] fragments are scheme-specific [09:18:07] sm joins the room [09:18:14] e.g., data IRIs and javascript IRIs handle fragments differently than other schemes [09:18:15] @abarth they are media type specific [09:19:01] Wait. [09:19:06] Roy to microphone. [09:19:29] Remove #55 as WG item. [09:19:42] or representation specific, to use the terms of 3986. [09:19:45] Agreement in room. [09:19:48] Larry to mic. [09:20:11] Adam, that's not correct for data URIs [09:20:28] Adam, if a UA does that it's a bug (I'm aware of one in Mozilla) [09:20:42] Admo, dunno about javascript [09:20:50] i think that depends on your definition of a bug :) [09:20:58] it's not going to change anytime soon [09:21:03] => not bug [09:21:46] Let's see how that goes :-) [09:21:58] @abarth, do you agree that the registration doc is not the place to hash out what frag identifiers mean? [09:22:13] yes [09:22:15] Martin. go [09:22:42] Mic: But the problem here is that, if the scheme is unknown to a particular implementation of something and it provides special limits (length or otherwise) that aren't known (because the scheme is unknown), the user is guaranteed a completely incomprehensible error situation. [09:23:18] Ted to mic. [09:23:41] I was just repeating John's comment to the mic [09:23:42] He reads John's comment from jabber. [09:23:51] Larry to the mic. [09:24:07] Users do not type URIs. They click on them. [09:25:01] As an example, the reserved set in URNs is larger than the base URI set. [09:25:04] Roy to mic. [09:25:10] the address bar is editiable, last time i checked [09:25:15] I think in general scheme-specific length limits just don't make sense. There may be some very specific schemes that by definition don't allow more than a certain length. E.g. a hypothetical 'dns' scheme may have a 255 length limit (if the only thing it contains is the domain name). [09:25:33] the dns scheme is not hypothetical [09:25:54] Roy, could you send that to the IRI WG list? [09:26:13] I don't see it as really being much different from saying what characters are legal [09:26:24] to Dave: I agree. [09:26:34] nothing stops a user from typing in (or clicking on) a malformed (by a scheme's definition) URI [09:27:27] i have roy's doc [09:27:32] Anything else to the mic, Martin? [09:27:42] no, not from my side. [09:27:44] if the license is fine, i can use the info [09:27:46] ok [09:27:48] OK, done. [09:27:49] Jacky Yao (Health Yao) leaves the room [09:27:50] we are adjourned [09:27:51] Larry: so I type in an IRI, and it gets translated (dumb mapping) to a URI that is filled with %-encodings and possibly in a different order. Then the user gets back a message in terms of that URI that says something rude about something the user neither typed, clicked on, or has any potential of seeing or even reliably converting back into whatever was typed. Put differently I don't think "that is what happens" with the implication of "users better just suck it up" as an appropriate way to design UIs. If one wants well-designed UIs, then whatever generates the error messages has to have clear access to whatever the user typed/ or saw. (no need to take that to the mic) [09:28:20] yone leaves the room [09:28:30] abarth leaves the room [09:28:32] thanks everybody! [09:28:48] John C Klensin leaves the room [09:28:59] resnick leaves the room [09:29:06] to John: Actually, some things like that happen with IDNs in browsers. [09:29:28] ywang830 leaves the room [09:29:38] sm leaves the room [09:29:40] bkihara.l leaves the room [09:30:50] hardie@jabber.psg.com leaves the room [09:31:11] Julian leaves the room [09:32:04] alexey.melnikov leaves the room [09:34:04] psa leaves the room [09:35:46] Jacky Yao (Health Yao) joins the room [09:37:27] tony.l.hansen leaves the room [09:39:00] joseph.yee leaves the room [09:42:44] Jacky Yao (Health Yao) leaves the room [09:43:01] tony.l.hansen joins the room [09:47:34] Dave Thaler leaves the room [09:54:08] tony.l.hansen leaves the room [09:55:34] Dürst leaves the room [09:56:32] Dave Thaler joins the room [09:56:38] Dave Thaler leaves the room [10:10:48] bkihara.l joins the room [10:11:40] bkihara.l leaves the room [10:53:13] Julian joins the room [10:54:02] Julian leaves the room [10:54:45] psa joins the room [10:59:06] mnot joins the room [10:59:14] alexey.melnikov joins the room [11:08:05] alexey.melnikov leaves the room [11:08:05] alexey.melnikov joins the room [11:09:36] Thomas Roessler joins the room [11:34:11] mnot leaves the room [12:08:44] Thomas Roessler leaves the room [13:03:23] alexey.melnikov leaves the room [13:19:15] psa leaves the room