Internationalization Core Working Group Teleconference

Agenda and Minutes

Action Items

close ACTION-101

<trackbot> ACTION-101 Extend number of ports for I18N call to 15 closed

ACTION-109: koji: haven't heard back, will ping again

<trackbot> ACTION-109 Check about Korean layout task force notes added

close ACTION-130

<trackbot> ACTION-130 Write to www-international, including decriptions of other lists from about page, announcing changes in our use of list closed

Info Share

andrew: working for state gov't creating guidelines on how they use translations on the web
... especially focusing on new/emerging languages
... that have limited support
... will be probably public

IRI and IRI bidi documents

http://www.w3.org/International/wiki/Review_radar

http://tools.ietf.org/html/draft-ietf-iri-3987bis-09

http://tools.ietf.org/pdf/draft-ietf-iri-bidi-guidelines-00

http://tools.ietf.org/html/draft-ietf-iri-3987bis-11

http://tools.ietf.org/html/draft-ietf-iri-comparison-01

Norbert: thought comparison document looked fine

matial: no problem with it

http://tools.ietf.org/html/draft-ietf-iri-bidi-guidelines-02

matial: may be different for hebrew vs. arabic users
... or even between individual users
... still trying to do usability tests
... one thing I don't like in document is that there is too much freedom
... not as critical what we decide as long as mandatory
... e.g. in Section 3
... talks about components
... domain, segment, etc.
... each one should laid out independently of other components
... and then components can be subdivided, such as labels in domain names
... but if labels allowed to swap (or not) in implementations, it should be consistent
... be normative rather than optional
... once we have a consensus, it should be enforced

Norbert: trying to specify "what is a good bidi iri"
... but spoofers are not interested in "good bidi IRIs"
... so standard needs to specify what happens (reject, color address bar)

addison: are we ready to say something (and what)

matial: way to display IRIs should become unique for all browsers

addison; consistent, right?

matial: same IRI displayed same way in all user-agents

<matial> draft email

addison: draft email about sense of WG? or come back next week? preference?

norbert: there is no consensus in IRI community

addison: problem is that there is diversity of opinion and we're unlikely to achieve a consensus of "100% minus one person"

foo/bar/baz/example.com://http

aharon: side of bus is RTL, and works when typed in, but when some LTR thing is added to the URL, the whole thing changes direction, alarming user

addison: some sense of a base implicit direction of LTR

aharon: anyone with a RTL domain name will be unhappy if the domain is in LTR context
... think there will be strong objections to consistently LTR IRIs

addison: other possibility is that it's insoluable and doesn't get (need to get) solved
... just guidelines for IRI beauty

aharon: if you have an all RTL domain name/site name, you should take pains to create all RTL URIs (including path)
... if you don't, it'll look ugly
... but limit to how much those guidelines will be possible to follow
... since parts such as query can't be fully controlled
... or parts of the domain may be just an alias for an LTR name
... would be value in publishing guideline

andrew: two basic scenarios
... IRI that is fully LTR and IRI that is fully RTL---both work consistently
... when you mix that the problem occurs
... so guidance on how to display
... and guidance to e.g. CMS developers
... so that they can implement in ways that makes sense and works

addison: so... what to do/say at this point?
... cut to chase with "guidelines"? or "normative display"? or?

aharon: what about giving user preferences time to shake out, give time to user testing

matial: some testing planned, but not happening yet
... users of different proficiency how they perceive IRI (such as napkin/bus)
... or when told, how they write it
... hopefully these not-as-extensive-as-maybe-would-like will give us a sense
... maybe get results in the next two or three months
... but may get different results
... maybe indication
... might not be perfect
... but solve simple cases
... based on TLD, which is unlikely to change

com.EGYPT

EGYPT.com

com.tpyge

<Norbert> مصر.com

norbert: e.g. if you have entirely RTL domain name, you can do one thing

<koji> sorry, I've got go...

norbert: but also have to say how to make spoofs detected

AOB?

norbert: will there be user tests of IRIs in Arabic speaking countries?

gwyneth: can follow up with Shawn

addison: for next time?

norbert 3987bis and ruby?

addison: no objections?
... will send links

<andrewc> مصر.com

andrew: which is TLD?