Combined minutes of IDNABIS WG meetings IETF 73 Minneapolis MN scribe: Andrew Sullivan Day 1, 2008-11-17 The meeting was called to order at 15:20. The scribe was appointed for both minutes and Jabber. Adjustments to the agenda were entertained. The Chair suggested that there was the potential for confusion between lookup time and registration time, and noted that the idea is to trade some processing at time of registration in order to make lookups cheaper. The Chair declared that there was prior consensus to split the normative definitions out from the rationale document. The editor has completed this. The Chair stated that he believed there to be consensus on the five-document structure, and urged that the EG focus on content rather than form henceforth. The Chair understood responses on the issue of assignment permanence to indicate rough conensus in that area. The previous "Tranche 4" issue consensus call uncovered disagreement. The Chair , and noted as well the somewhat troubling small number of responses to the requests for input. The Chair urged participants to participate more actively. On "Tranche 5", the Chair suggests that for now we leave alone the principle that any change requires IETF action. The document editor thinks that the issues raised in "Tranche 6" discussion have been addressed. Please check; if the Chair hears no additional remarks, he will declare consensus. On "Tranche 7" the Chair detected rough consensus. On "Tranche 8", the Chair interpreted the responses as rough consensus on the specific exceptions for Eszett and Greek Final Sigma, as well as the rule prohibiting JAMO characters. Some discussion of the Eszett issue arose, but then was deferred until the consideration of the tables document. The Chair opened discussion of issues on the definmitions document. There was none. Chair opened discussion of issues in the protocol document. There was none. The Chair opened discussion of the bidi document. Paul Hoffman, Pete Resnick, John Klensin, Harald Alvestrand, Ted Hardie, Eric Brunner-Williams, and Vint Cerf speaking as an individual engaged in a discussion of whether the rationale material in bidi ought to be moved to the rationale document. On the one hand, it's nice to have all the rationale in one place. On the other hand, bi-directional issues are sufficiently different from other cases that perhaps they should be collected all in one document. The Chair formed a sense that the existing structure is correct, but asked people who find it confusing to send text for the document. The Chair opened discussion of the tables document. This resulted in quite a bit of discussion. The first issue was raised by Paul Hoffman, who asked whether the IANA registry will be authoritative. It is not: the rules should be. The second issue was also raised by Paul Hoffman, who asked why a registry ought to be created at all given that it isn't authoritative. A discussion among Paul Hoffman, Patrik Fältström, Harald Alvestrand, the Chair, John Klensin, and Eric Brunner-Williams about the dangers, utility, and likely effects of having a non-normative registry at IANA. The Chair suggested that the IANA-listed tables should have a warning about them being non-normative, and also that an IANA registry could be created for the contextual rules. The third issue was a return to the discussion of Eszett, raised by Pete Resnick. A discussion ensued among him, Barry Leiba, Marcos Sanz, Paul Hoffman, the Chair, John Klensin, and Eric Brunner-Williams, attempting to clarify what the mapping between ss and Eszett would be, and what the effects would be. The participants appeared to agree that it should be up to registry policy, and that there are some dangers here that need warnings. Ted Hardie then suggested that the categories in section 2 needed to be more explicit, because they are not mutually exclusive as currently written. Yoshiro Yoneya asked about checking the contextual rules in Appendix A. Patrik Fältström, speaking as document editor, said the appendix would be significantly fixed in the -04 version, so please check it then. The Chair then turned the discussion to the topic of the Arabic-Indic digits. The issue is that there are non-extended and extended Arabic-Indic digits that interact in strange ways with one another and with European digits. Harald Alvestrand proposed a rule intended to solve the problem. Discussion among Harald, John Klensin, Eric Brunner-Williams, the Chair, Patrik Fältström, Paul Hoffman, and Ted Hardie about the rules, whether the proposed rule is the right one, what the effects might be, and whether the WG is in a position to decide. The topic became homework for WG participants to ponder over night. The Chair opened discussion of the rationale document. Harald Alvestrand objected, saying the meeting time had been exceeded. There was some confusion over the timetable, and then it emerged that Harald was wrong, so the Chair told him off. Much laughter. Discussion ensued involving John Klensin, Ed Lewis, Marcos Sanz, and Yohiro Yoneya. John asked repeatedly that participants send text patches with any report of an issue. The Chair adjourned the meeting at 17:17 local time. Day 2, 2008-11-18 The Chair called the meeting to order at 09:00. He detected consensus on idnabis-defs, notwithstanding some non-substantive changes that were on the way. The Chair supported a suggestion in the protocol document to move section 4.1 to 4.3. The Chair raised the issue of outstanding issues with the bidi document. In particular, he wanted to confirm the status of inter-label checking. Harald Alvestrand clarified that whereas he had previously been arguing for inter-label checking, he has removed any suggestion of it from the latest document version. The Chair then returned to the Arabic-Indic digits issue. Patrik Fältström reported that, as the editor of the tables document, he would add 20 exceptions and two contextual rules unless he heard a compelling argument to do other wise. Discussion involving Eric Brunner-Williams, Harald Alvestrand, John Klensin, the Chair, Pete Resnick, Patrik Fältström, and Barry Leiba about what the rules should be. The alternatives for non-extended Arabic-Indic are as follows: If an indic digit exists in a label, 1. do not allow any other digits 2. do not allow an ASCII digit or an extended Arabic-Indic digit 3. do not allow extended Arabc-Indic digits Everyone seemed to agree that (1) is too restrictive. Eric Brunner-Williams and John Klensin were to consult on different experiences that might lead to different choices between (2) and (3). Andrew Sullivan asked why the issue is not just a problem that should be left with registry policy. John Klensin replied that, if the user agent transformations that are apparently in the field work the way they are thought to, the registry can't tell what the user is trying to type, so restricting the options from which registries can choose is important. The Chair finaly turned to section 7.2 of the rationale document. He noted a problematic use of the word "sunrise". With no further agenda items, the Chair entertained the participants with a singing robot device, and outlined his plans to replace himself with such a device. He adjourned the meeting at 09:40 local time.