Internationalization Core Working Group Teleconference

12 Sep 2012


See also: IRC log


Addison, Richard, Felix, Aharon, Norbert, Mati, Fantasai, Koji
Addison Phillips
Addison Phillips


Action Items

Check about Korean layout task force

koji: pinged, didn't get a response
... preparing to translate the korean version

<r12a> http://www.w3.org/International/track/actions/open

koji: will ask murata-san about what to do with the document

Work on tidying up Additional Requirements for Bidi in HTML & CSS over coming couple of months

Update qa-byte-order-mark document targetting end-of-August for review

Prepare some examples of ITS2NIF showing NFC issue for consideration next week

Move TURTLE comments to tracker and send the comments to the WG using the usual process

close ACTION-146

<trackbot> ACTION-146 Move TURTLE comments to tracker and send the comments to the WG using the usual process closed

Info Share

<fsasaki> http://www.dagstuhl.de/en/program/calendar/semhp/?semnr=12362

felix: last week at a seminar
... multilingual semantic web
... a kind of specific version of like Dublin ML-Web workshop
... may be some interesting things to watch in the future
... such as best pracices
... ML-Web LT project sponsoring workshops
... next one in Rome
... contact me for info on participation

<r12a> richard would certainly like to be involved

<fsasaki> great, r12a, I'll give you info

<r12a> http://www.w3.org/International/tests/

richard: working on tests before my holiday
... converted all tests in html5 and css sections (the first two)
... results pages are all updated live from database
... using latest data
... stuff we've been changing all done

<fsasaki> update to workshop: it's another MLW workshop in this series http://www.multilingualweb.eu/ , and we just started the program committee work. Let me know if you want more info on that, we very much welcome new PC members

richard: updated whitespace tests, rewriting line-breaking tests
... made a hash of the JP/CN ones??
... and just now got ability to remove test results so can correct
... i18n charter review finishes on 14th

<r12a> http://www.w3.org/International/questions/qa-resizing-backgrounds.en.php

richard: someone wanted to translate above article
... so I updated the above, 'cause it needed it
... added a couple of newer CSS methods that can be used, added examples

<Norbert> http://wiki.ecmascript.org/doku.php?id=globalization:specification_drafts

norbert: ecmascript I18N
... tc39 meeting next week
... this will be the standard unless you complain now

Bidi (Isolation etc.)

richard: would be useful to recap what's going on
... and how that affects advice to html authors

Aharon's update: http://lists.w3.org/Archives/Public/public-i18n-bidi/2012JulSep/0067.html

The editor of the relevant CSS specs (Writing Modes Level 3 and Text Level 3) have already met and worked out the disparity between CSS and Unicode isolates.

aharon: fantasai and I met and she changed both of the above in editor's latest
... already had isolates, but didn't/couldn't make use of Unicode isolates
... para breaks in isolates now act as para breaks to surrounding para
... and isolates now are a special form of embedding
... and documents will change again once uax#9/Unicode are updated
... to explicitly reference how UBA does this
... another bidi issue is that bidi parenthesis algorithm from MSFT
... they decided to try to add it into uax#9

<koji> fantasai: probably too busy for 6.2 for now but I hope after that

aharon: some folks at MSFT and Google tasked to provide new version
... very interested in seeing both in at same time
... to ensure compatibility between the two
... have talked to msft folks
... they are implementing isolates
... so far no problems?
... parenthesis
... if typing parens in a bidi para without using controls
... all sorts of things go wrong

"hello!" -> "!hello"

aharon: and parens in particular were a problem
... if you have 'BOOK(S)'

<matial> Sorry, I have to quit the call.

<r12a> )book(s WERBEH

aharon: you sometimes get )BOOK(S
... and there are some other cases
... looks for parens pairs and balances their bidi levels (and some other details)
... so that they match
... all of the mirroring enclosing punctuation
... a negative side effect is that UBA gets harder
... since it needs more than just bidi properties
... but will probably happen in unicode 6.2.1

Various HTML5 bugs: http://lists.w3.org/Archives/Public/public-i18n-bidi/2012JulSep/0071.html http://lists.w3.org/Archives/Public/public-i18n-bidi/2012JulSep/0068.html

it makes no sense to limit the placeholder attribute to values of the same direction as the <input>

People don't know about these characters because they haven't needed to use them. But if they want this to work, then they need to use them. So I don't see why this would remain a truism forever. There's nothing more magical or difficult about &#x202B; than <span dir=rtl>, IMHO. If anything, I would say that the mistake we made was ever using high-level markup for this character-level issue, as it means that every time we have a non-markup-level problem, we

norbert: why is that an attribute

richard: xhtml2 rid us of title and alt, but... couldn't get rid of it in html
... but this is a new invention

aharon: hixie basically said "oh, just use formating characters"
... different bug 17829

<r12a> https://www.w3.org/Bugs/Public/show_bug.cgi?id=17829

Bug 17829 - dir=auto does not work for user-visible attribute values

aharon: even more ludicrous
... if you have <div dir=auto>
... then attributes inside that (like span or something with @title)
... then contents get direction from *surrounding* text, with *no* consideration for contents of the attribute in question
... proposed new attribute @attribsdir
... to allow direction setting on all attributes of an element
... could say auto (or ltr/rtl)
... reopened this bug although hixie rejected
... and this would also solve placeholder
... original proposal included a list but that was too complicated

richard: stop using attributes for natural language text

norbert: or add a new attribute for every attribute they add
... title attribute gets its own window/context, while placeholder is embedded into document

addison: do we want to reopen this one?


still shows new

richard: copy 17814 to 15488 and then go on to comment there (in 15488)

aharon: 15488 doesn't offer solution
... before we opened the @attribsdir

addison: 1. quit making attributes with natural language text
... and 2. but then use @attribsdir (or something) to fix it

aharon: would there be a placeholder *element* instead?a

addison: generated text or placed text can't always provide bidi controls

richard: attributes with mixed direction needs can only have controls
... only talking attributes here

aharon: there is room to say that inside attribute values you will need to use bidi controls anyway
... that's true, but that doesn't translate to *always* requiring their direciton to be indicated that way
... less common to mix text anyway
... but not end of world to have to use them in attribute values
... not enthusiastic to continue arguing it?
... often input is a different direction from that of page (and of placeholder)
... e.g. phone number or email address is ltr, but placeholder needs to be rtl

richard: make placeholder dir=auto by default??

aharon: would prefer a way to control is, because auto means "first strong"

<scribe> chair: time check

richard: can continue on email?
... been writing a bidi document for a year :-)
... maybe need to rewrite further
... but need to do that soon
... so add as agendum in an upcoming telecon
... aharon made some radical change requests

next time


tpac: we need a phone

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2012/09/20 15:20:00 $