Re: jabber log from Monday's IEA BOF?

[Tony Hansen <tony@xxxxxxx>]

Thanks to Yoshiro YONEYA and Ludovic Poitou, a complete jabber log of 
the iea meeting is attached. They both had parts of the jabber log -- 
the combination is attached.

	Tony Hansen
	tony@xxxxxxx

Tony Hansen wrote:

> If you had jabber turned on during Monday's IEA BOF, would you please 
> check to see if you happen to have had logging turned on in your jabber 
> client? "Due to circumstances beyond our control" (tm), jabber logging 
> was not turned on at the server, and we're looking for a copy.

New Conversation at: 1/3/2004 15:24:35
[15:24]ludomp has entered the room.
[15:24]iea
[15:24]yone has entered the room.
[15:24]leslie has entered the room.
[15:24]randy_g has entered the room.
[15:24]tonyhansen has entered the room.
[15:24]hardie has entered the room.
[15:24]SAH has entered the room.
[15:24]resnick has entered the room.
[15:24]amarine has entered the room.
[15:24]agaton has entered the room.
[15:24]alexeymelnikov has entered the room.
[15:25]* SAH has set the topic to: Internationalizing Email Adress BOF
[15:31]<tonyhansen> this is the IEA bof, part 2, pete resnick presiding
[15:32]<SAH> Agenda can be found at http://www.ietf.org/ietf/04mar/iea.txt
[15:32]<tonyhansen> pete: what are our next steps?
[15:32]<tonyhansen> paul hoffman starting off
[15:33]<tonyhansen> paul: overview of proposals so far
[15:33]<tonyhansen> mailing list: ietf-imaa@xxxxxxx
[15:34]<tonyhansen> www.imc.org/ietf-imaa
[15:34]<tonyhansen> history: started w/ a proposal, then discussion on proposal, rather than starting with requirements
[15:35]<tonyhansen> more history: then counter proposal, with arguments against 1st proposal, this led to better requirements
[15:35]<tonyhansen> proposal #1: ACE on the LHS only; #2 - UTF-8 in address, protected by ESMTP
[15:35]<tonyhansen> #3: UTF-8 in all headers, proteted by esmtp, #4: address mapping in the MUA
[15:37]<tonyhansen> now discussing +/-'s of each
[15:38]<tonyhansen> #1: + only MUAs need update; - exposure of ugliness to non-updated clients
[15:38]<tonyhansen> #2: + fairly tightly scoped; - must update SMTP
[15:40]<tonyhansen> (SMTP update has many drawbacks -- critical mass needed -- how to downgrade or bounce -- ...)
[15:40]<amarine> april sings kumbaya
[15:40]<tonyhansen> #3: + all headers treated the same; - must update SMTP again
[15:42]<tonyhansen> #4: + trivial to update; - doesn't gain much
[15:44]<tonyhansen> jon klensin: there is a combined (#5?) proposal that combines that combines #1 and #3
[15:44]<tonyhansen> the mailing list had lots of comments, but mostly by potential users and not by implementers
[15:46]<tonyhansen> this could potentially lead to a good idea not being implemented
[15:47]<tonyhansen> idna's adoption in the past year has not met expectations
[15:48]<tonyhansen> usoft's internet explorer lack of implementation has dampened things
[15:49]<tonyhansen> are people just tired, after working so much on idna?
[15:49]<tonyhansen> paul is handing mike off to jon
[15:50]<tonyhansen> jon: where are we? we need interest from email implementers! until we do, we can keep thinking deep thoughts on it
[15:51]<tonyhansen> we have agreement that if we go utf8, we will need transport machinery to go with it
[15:52]<tonyhansen> transport layer will take a long time -- example of mime
[15:52]<tonyhansen> what is it important to optimize in email? what are tradeoffs in importance?
[15:53]<tonyhansen> if jon were to receive an arabic email address, he doesn't think he could do much  with it
[15:53]<tonyhansen> however, it's more rational to consider the problem within a community of users (chinese, klingon, ...)
[15:54]<randy_g> John: community of users of common language much larger than community using divergent languages
[15:54]<randy_g> John: easier to get community using common language to upgrade their mail infrastructure
[15:54]<tonyhansen> so maybe a transport upgrade is okay, because a community will upgrade more qiuckly than full internet community
[15:55]<tonyhansen> but it still falls back to needing mail implementors, because otherwise the point is moot
[15:56]<tonyhansen> paul: much agreement between he and jon, but there is another factor
[15:57]<tonyhansen> paul: maybe we should do nothing, and let mailng start over and do it right from scratch
[15:57]<randy_g> Paul: large numbers of people want to do a new mail infrastructure to fix the spam problem, or to fix the foo problem
[15:58]<tonyhansen> jon: he thinks odds of mail-ng is <0
[15:58]<randy_g> Paul: Many of these people start sentences with 'If you only' or 'All you need to do is'
[16:04]<tonyhansen> (randy, thanks for filling in the blanks)
[16:09]<randy_g> John: mentions how much better mail infrastructure is now then 15 years ago.  Percent-hacks, anyone?
[16:09]<tonyhansen> jon: we have a system that works, mostly, and people take it for granted
[16:09]<hardie> !@:: anyone ?
[16:09]<tonyhansen> paul: it's a serious issue of have incompatible systems
[16:09]<tonyhansen> jon: those incompatible systems are all now gone
[16:09]<randy_g> John: mentions history of proprietary mail systems, which often couldn't communicate with each other except by translating to and from Internet mail format
[16:09]<tonyhansen> jon: switching to mail-ng will bring back proprietary systems
[16:09]<randy_g> John: but the existing mail system will never go away
[16:09]<randy_g> John: we'll end up with multiple systems
[16:09]<tonyhansen> ted hardie: remembers past too, but
[16:09]<randy_g> Ted: IM shows that new msging infrastructure can be deployed
[16:09]<tonyhansen> ted: new messaging infractures can and do appear, eg im
[16:10]<tonyhansen> ted: a new infrastructure that won't be smtp-ng, but something fundamentally different
[16:17]<tonyhansen> ted: mall will continue as any-to-any network, but new infrastructure will provide additional features that may gateway into smtp
[16:17]<tonyhansen> ted: body of users will be all users who use all languages
[16:17]<tonyhansen> ted: can we live  with this messaging infrastructure for its useful lifetime
[16:17]<tonyhansen> jon: oddds of new infrastructure are 10-20 years, not 5 years
[16:17]<randy_g> Gred V: SMS was a very successful new msging service that was deployed recently
[16:17]<tonyhansen> greg v: one infrastructure that was sucessfully and quickly deployed was short text to sms
[16:17]<tonyhansen> greg: we need to fix what we got cause that's what we got
[16:17]claudioallocchio has entered the room.
[16:17]<randy_g> John: history of Internet is that new applications, which fill a vacuum, deploy quickly
[16:17]<randy_g> John: new applications that replace an existing one deploy very slowly
[16:17]<tonyhansen>  pete: even back in the days of email gateways, we could still say 'send it through that gateway'
[16:17]<tonyhansen> someone: other systems weren't designed to be compatible, nature of design procss may make that point moot.
[16:17]<tonyhansen> someone: an impetus is fear of english as being international language
[16:17]<tonyhansen> someone: => nathaniel borenstein
[16:17]leslie has left the room.
[16:17]leslie has entered the room.
[16:17]leslie has left the room.
[16:17]<tonyhansen> nat: get over it
[16:17]<tonyhansen> nat: learn ascii
[16:18]leslie has entered the room.
[16:19]<tonyhansen> yoshiro yoneya: demand is high, keep iternet stable. hope we will move to developing an encoding
[16:19]<tonyhansen> pete: again, who's going to implement
[16:20]<tonyhansen> jon: predictable that soon, the culture ministry of some country (lower slobovia) will decide that email WILL use character set XYZ
[16:20]<hardie> Funny story:  travelling in Xinjiang (Turkic-speaking area of western China), I ended up staying at a guest house in Hami with a group of Japanese tourists and Arabic-speaking business folk.  It turned out that their common language wasn't English or even putonghua, it was kis-swahili, because the Arabic-speakers traded in Africa and the Japanese had done a Peace Corps-equivalent stint there.
[16:21]<tonyhansen> jon: and ietf will respond with 'if you're going to do this, this is how you're going to do it', or #2 (???) , or #3 'no standard'
[16:21]<hardie> Moral:  you may think English is the lingua franca of the world because you speak it.  If you spoke putonghau, kiswahili, or Spanish, you might see those are regional equivalents just as important to their communities.
[16:22]<tonyhansen> jon: saying 'learn an iea-form!' won't get us anywhere
[16:22]claudioallocchio has left the room.
[16:31]ludomp has entered the room.
[16:31]iea
[16:31]yone has entered the room.
[16:31]leslie has entered the room.
[16:31]randy_g has entered the room.
[16:31]tonyhansen has entered the room.
[16:31]hardie has entered the room.
[16:31]SAH has entered the room.
[16:31]resnick has entered the room.
[16:31]amarine has entered the room.
[16:31]agaton has entered the room.
[16:31]alexeymelnikov has entered the room.
[16:32]<randy_g> Paul: when this effort started, we had a lot of commitment, but we don't have any now
[16:32]<randy_g> Paul: there is a lot of user domand now, vendors are creating non-standard implementations
[16:33]<randy_g> Paul: IETF jumping in when it sees proprietary implementions doesn't usually work
[16:33]<randy_g> Paul: We need to know why email vendors aren't doing this
[16:33]<randy_g> (aren't supporting this work)
[16:41]<randy_g> Pete: IETF is supposed to bring engineers together so they interoperate; we don't have that now
[16:41]tonyhansen has left the room.
[16:41]tonyhansen has entered the room.
[16:41]<randy_g> Pete: If the vendors that are interested in supporting this aren't interested in interoperating, then what are we doing here?
[16:41]<randy_g> Comment: we need server support; without it no point in clients doing anything
[16:41]<randy_g> (not sure who is speaking)
[16:41]<randy_g> That was Anti
[16:41]<randy_g> Ted: experimental documents may be appropriate
[16:41]<randy_g> Ted: some vendors believe they can capture the whole market and hence non-interop is good from their point of view
[16:41]<randy_g> Ted: at some point they realize they can't capture it all, and have to interoperate; at that point we should have something
[16:41]<randy_g> Ted: doing only RHS isn't enoigh; we need to be able to do both LHS and RHS and then we may see more widespread eployment for multiple applications
[16:41]<randy_g> Pete asks if what Ted suggests is engineering
[16:41]<randy_g> Ted says if we do two experimental docs: one that requires infrastructure upgrade, and one that doesn't but could be deployed in an island, then it is valuable because it would demonstrate user demand (?)
[16:41]<randy_g> Ted: If we leave this entirely in the proprietary realm, we will have poor interoperability between islands and also between islands and email
[16:41]<randy_g> Pete clarifies that Ted is proposing that we could charter to do two experimental protocols
[16:42]<randy_g> Ted: If we do that we could get operational experience with both and then know which one is the better approach
[16:45]<randy_g> Ted: speaking personally he thinks it would be charterable
[16:45]<randy_g> (not sure who is speaking)
[16:45]<randy_g> (?): In Korea we have a lot of demand for this, and we have services for this now
[16:45]<yone> Taeha Park is speaking
[16:45]<randy_g> Taeha Park : We provide native-name email addresses and do it without violating the standards; people like it
[16:45]hta has entered the room.
[16:45]<randy_g> Taeha Park: We need something to use until full standard is deployed
[16:45]<alexeymelnikov> this is done using LDAP lookups
[16:45]<randy_g> Taeha Park  is from Netpia
[16:45]<randy_g> Greg V: A standards-track work item driven by end users is OK; we have a wealth of talent and engineering in this room; we can do a good technical solution even without implementators
[16:48]<randy_g> Greg V: If server vendors aren't interested, let's do a low-cost, easy to implement client-based solution
[16:48]<randy_g> Greg V: Problem with client-based solution is it separates transport from user-presentation and that has a history of debugging problems
[16:48]<randy_g> Greg V: We could work around that (people use cut-and-paste to address, reply to messages, etc. but don't type addresses

=== changing recorders ===

[16:42]<tonyhansen> jon: mentioned difficulty of having user display name diff from what is used underneath
[16:43]<tonyhansen> paul: all client-only-based systems will have problems
[16:43]<randy_g> Paul: client-based system can work; one way to create an island is to have addresses that only people in that island can type
[16:44]<tonyhansen> someone: i18n address ARE important to china
[16:45]<randy_g> He has ascii address on one side of his card and Chinese on the other
[16:45]<tonyhansen> someone: he prints i18n address on his business cards, on the chinese side, and ascii address on other side
[16:45]<tonyhansen> someone => xiaodong li
[16:45]<randy_g> Jon Rosenburg: question on scope: is restriction to IDs in email only?
[16:46]<randy_g> Jon R: many applications leverage off email; consider SIP
[16:46]<randy_g> Jon R: XMPP uses stringprep for user parts
[16:47]<randy_g> Jon R: we'd need extensibility so other applications can build on it
[16:47]<randy_g> Jon R: is group interested in solving this larger problem?
[16:48]<randy_g>   Asking person who has Chinese email address on one side how it is implemented.  Answer: Unicode
[16:48]<tonyhansen> ? to xiaodong: how are chinese addresses being implemented: punycode
[16:48]<tonyhansen> (idna on left side)
[16:48]<hardie> james seng to xiaodong li
[16:49]<tonyhansen> james: addresses are looked up in ldap to map to ascii
[16:49]<randy_g> james Seng suggests maybe an informational draft documenting what is being used now would be OK
[16:49]<randy_g> James' comment on LDAP was that it was Netpia's solution, and was different from anyone else's
[16:50]<randy_g> Pete clarifies that Netpia's solution is not an email protocol element, it is an LDAP query
[16:50]<randy_g> "Native Language Internet Service" maps native name key words to domain name
[16:51]<tonyhansen> ?: explaining another proprietary way of doing mapping used in another product
[16:51]<tonyhansen> jon@mike
[16:52]<randy_g> John: change mail clients so that everything gets looked up first and translated; compares to X.400
[16:52]<hta> hey, that's not X.400 - it's OUTLOOK!
[16:52]<randy_g> Paul notes that Bob Morgan had a document years ago on LDAP lookups for email names
[16:52]<hardie> someone closer thwap harald for me.
[16:54]<randy_g> John: integration was poor, so a recipient tended to see one's own address instead of the sender's, so the way to reply was to make a phone call.
[16:54] * hta ducks a thwap
[16:54]<randy_g> John: this is a promising area but risky
[16:55]<tonyhansen> someone: in china, chinese email embedded directory service into Outlook
[16:55]<randy_g>   in China there is a company that does something similar to Netpia; Chinese email that works with Outlook but is not email
[16:55]<randy_g> No protocol and no interoperability
[16:55]<tonyhansen> but no interop
[16:55]<tonyhansen> someone was xiaodong
[16:55]<tonyhansen> james seng: client-side requires patches to clients
[16:56]<randy_g> (I'm sitting behind everyone so unless I know them well I can't tell who they are)
[16:56]<tonyhansen> james: a million users a few years ago was a lot -- now it's nothing
[16:56]<tonyhansen> pete: wrapup
[16:57]<tonyhansen> pete: let's do experimental wg work
[16:57]<randy_g> Pete: we could do exp. protocols for both compatible and ng service
[16:57]<randy_g> Pete asks if people in the room are willing to work on this
[16:57]<randy_g> Pete: One is ASCII-compatible and one involves SMTP servers
[16:57]<tonyhansen> #1: ascii-compat, #2 smtp servers are updated
[16:58]<randy_g> Ted: you either need SMTP extension or you have end-to-end compatibility
[16:58]<tonyhansen> ted: paraphrasing pete's #1 & #2
[16:58]<randy_g> Ted: we have had some work on both on the lists
[16:58]<randy_g> Ted: we could charter to do more work on this and produce documents
[16:59]<tonyhansen> pete: directory lookup soltuion is not part of potential work
[16:59]<tonyhansen> pete: but perhaps someone would want to write that up anyway
[17:00]<tonyhansen> tae ha park: volunteering for directory-based
[17:01]<tonyhansen> paul: can we do it without a wg?
[17:02]<randy_g> Paul: we could do it all in one document, and that wouldn't be chartered
[17:02]<randy_g> Paul: Is there good reason for a WG?
[17:03]<tonyhansen> ted: multiple docs give us a way to evaluate the question "which one worked"
[17:03]<randy_g> Ted: If we're doing an experiment, having multiple documents is better (easier to see which one worked best)
[17:03]<tonyhansen> ted: add an informational doc as well that says how the experiment should go forward
[17:04]<randy_g> Ted: if people are using punycode then we can get a document with operational experience on it
[17:04]<tonyhansen> ted: we definitely need community review, but not necessarily a wg
[17:05]<tonyhansen> pete: exp work will go on, a proposed charter will also be flown to get comments
[17:06]<randy_g> (Going back to Harald's comment on Outlook, that model produces super-ugly addresses, with double quotes, single quotes, commas, etc.  Useful caution for this work?)
[17:07]<randy_g> John: "People interested in the work" might mean "people interested in spending 6 months or a year on the charter"
[17:08]<tonyhansen> jon: if we can get a wg, it will mean we'll have critical mass of people interested in working on it
[17:08]<tonyhansen> pete: doesn't think we need to try flying a charter
[17:08]<randy_g> Pete: personal opinion is to let people do individual experimental documents
[17:09]<tonyhansen> randy preshun: wait on floating a charter until after we get some experimental docs out
[17:09]<randy_g>   Many comments that there is a sense of urgency in many communities; this is an argument against a WG and for letting individuals do experimental documents for the different approaches, and then if need be, a WG to hash out a unified solution
[17:10]<tonyhansen> paul: will change name of imaa mailing list to more generic name
[17:11]<randy_g> Pete asks who is willing to work on exp. drafts:
[17:11]<tonyhansen> (preshun => presuhn)
[17:11]<iljitsch> I have an idea I may want to work out in a draft.
[17:12]<tonyhansen> volunteers: paul, ted, xiaodong, james seng, alexey, others to give me business cards
[17:12]<tonyhansen> jon: procedural question: if no charter for wg, and no draft being circulated, then is this an ietf activity?
[17:13]<tonyhansen> ted: more critical, is such an activity covered by "note well"?
[17:13]<randy_g> John: if this isn't an IETF activity, no one can claim to be involved in such as IETF activity
[17:13]<tonyhansen> harald: nothing officially exists in the ietf except wgs
[17:14]<tonyhansen> harald: this is a mistake; investigative activities should also be covered
[17:14]<tonyhansen> harald: we can't block people from experimenting
[17:15]<randy_g> Harald: Paul's mailing list is a private activity, and hence can enforce "note well" on its own
[17:15]<tonyhansen> harald: imc's mailing list is a private activity, and paul should feel free to impose whatever rules he chooses (such as "note well")
[17:15]<randy_g> Paul: can even be stricter
[17:15]<randy_g> John: Can anyone do an end-run on an investigational activity?
[17:16]<randy_g> John: that is, can authors of documents involved in this effort give up on it and submit individual submissions?
[17:16]<tonyhansen> harald: does jon see publishing such experimental docs as harmful?
[17:17]<tonyhansen> (jon shakes head no)
[17:18]<randy_g> Ted: "note well" or similar would be very valuable very early
[17:18]<randy_g> Ted: if you use your own language be sure to get it vetted by a lawyer
[17:18]<randy_g> Pete: we're going to take this to the newly-named mailing list
[17:18]<tonyhansen> ted: this will continue outside of ietf for the experiment to go forward
[17:18]<randy_g> Pete: folks can write their exp. drafts
[17:19]<randy_g> Pete: we can subsequently bring work into the IETF
[17:19]<tonyhansen> james seng: can "note well" be used outside of ietf?
[17:19]<tonyhansen> paul: yes
[17:20]<randy_g> Paul: also, any list postings are considered public disclosure from patent/IPR point of view
[17:20]<tonyhansen> pete: we are adjourning without forming a wg at this point
[17:21]<tonyhansen> ted: should ietf work on mail-ng? see ted and/or scott if interested in that topic.
[17:21]<randy_g> Ted asks if people are interested in working on mail NG and especially the deployment time frame; find Ted and/or Scott during the week
[17:21]<randy_g> Ted: note that was not BOF-related
[17:21]<randy_g> (*close*)
[17:27]<tonyhansen> additional volunteers: Jeon, Aye-Shil asjeon@xxxxxxxxx     antti pitkamaki antti.pitkamaki@xxxxxxxxx    yoshiro yoneya yone@xxxxxxxxxx       tony hansen