W3C

– DRAFT –
Internationalization Working Group TPAC 2024

27 September 2024

Attendees

Present
addison, Frankie, Frankie Wolf (remote), jyasskin, r12a, TristanNitot, Willy Yeh, xfq
Regrets
-
Chair
Addison Phillips
Scribe
xfq, addison

Meeting minutes

addison: is tehre something not on the agenda that people would like to discuss?

jyasskin: I'm especially curious about the <time> question
… which is on your list

Info Share

jyasskin: I might be able to help with you find people in Google to participate

addison: I can name names in the Google space of people who've shown interest in the past
… I'm happy to connect offline

<jyasskin> I co-tech-lead web standards at Google, so recruiting there is kinda my responsibility. I'm not super optimistic, but I'll look.

addison: people saving their horizontals for the end

<jyasskin> 1 other topic that I forgot is that Bliss Symbolics in the APA is looking for a way to include themselves in HTML, and Elika and I think they should use Ruby.

addison: not doing horizontal engagement early
… when it would do the most good
… tehre's not a lot of appetite to add more gates to teh process
… I also wen to the sustainability breakout
… it is a new horizontal
… which is about limiting things like impact of web tech wrt things like power useage etc.
… third, I met WebAuthn people
… i18n people may dimly recall that they had invented a serialization for addig language and dir metadata to fields
… at our behest, we asked them to put lang metadata and they did by suing the Unicode lang tagging chars
… no one has implemented this
… if they did, no one would understand what this encoiding is
… this morning I added a comment about what we discusseed doing
… which is OK, don't implement a thing that attaches garbage to the end of oyour strings
… adn potentially a dcouemnt level metadata instead

jyasskin: I was in the APA-whatwg session
… they were talking about Bliss symbolics
… the sugestion is to attach it when they like to use ruby to attach them
… when I talked to fantasai about it, she said that ruby actually has some of the same semantic variation that Bliss has
… they will almost certainly need a language tag for bliss
… that doesn't seemm unreasonable
… you don't like the idea

r12a: we've been around this for a couple of times before witht he same people
… we definitely don't think it's a good idea to repurpose ruby as a glossing mechanism

jyasskin: it is a glossing mechanism

r12a: ruby was designed with a specific sort of target in mind
… and it has been develpoed with a specific target in mind which is CJKM
… I don't think it's a good idea

jyasskin: I feel strongly that HTML elements should have semantics and not just behaviour

r12a: a good example is that if you have japanese and you want to have normal ruby and you want to have Blissymbols

jyasskin: if you have effectively 2 different explanations fo the same text
… you have to mark them with the differences between them
… so we need a language tag for Blissymbols

r12a: in Japanese and Chinese in particular, depend on what kind of relationship the annotation has with the base text
… you're developing a tech that's not going to be widely standardized and used

jyasskin: you can say the same thign about ARIA

r12a: you could, but there's no reason to do more

jyasskin: fantasai was going to flesh out the semantic of the various ways Ruby is used

addison: OK, I don't want to be the last to know about it, though

<addison> Tristan

<addison> xfq: met with a11y guidelines

<addison> ... will do early review of WCAG 3

<addison> ... they will send it to us and incorp our checklist into their wcag3 process

<addison> ... will try this process and see if it is working

r12a: we spent a lot of time about the short review checklist
… I don't think we want another gating point
… we want peoople to apply the principles of i18n as they're doing the work right from the beginning
… that's what I was saying to the a11y people
… not look at the short review checklist after they develop something, but they should look at it before they develop something
… and they should follow the links in the right hand column and read that
… FPWD is too late
… it's a sort of gating point
… I don't know if we need to create a video that says here's the short review hcecklist
… and here's the longer checklist and how to get there
… and you should read all these things and go through these points one by one
… and they can play that video before FPWD

addison: @@1

Glossary Publication at Note

addison: I had a hallway conversation with plh

https://www.w3.org/TR/i18n-glossary

addison: as you know we have the i18n glossary
… which is a Note track document
… which contains all the i18n terminology that is not found in INfra
… we tell people to use these terms in their specs
… one of the challenges is that they get a warning from spec that says you have used a non-normative ref in your normative section
… wheyn they link to a term like code point to our glossary
… and we're a special case
… it was suggested in a coulple of places that we might become a living stdardn that would avoid that problem
… but doesn't seem to make sense
… that's too heavyweight
… becae we update the glossary all the time
… and it's not a registry
… the main thing we need to do is publish it with the Note status and not Draft Note status, which is what is currently is
… and then continue to publish it as a Note
… and then adjust the tools like respec to make it special

jyasskin: do you feel like the terms of the glossary affect the behavior of any normative specs that refer to it or are they just for understanding?

addison: they are for understanding
… it's helpful to not have people to have to copy the defs out

jyasskin: if everything in the glossary wouldn't change the test cases it seems fine for it to be a note

addison: like code point, we don't mean UTF-16 code point

addison: any other comments on glossary?

Advancing LTLI

https://www.w3.org/TR/ltli

addison: we have a document called Language Tags and Locale Identifiers, LTLI
… I have an issue that I own in specdev to harmonize the reqs between these 2 documents
… mostly mean importing the guidance from LTLI to specdev
… I think that's worthwhile
… I have a draft
… some time real soon now I'd like to advance LTLI
… I don't think there's other work on that
… I think it's mostly up to date
… I'd like to do a wide review again
… and publish that out
… so I want the WG aware

PR and Issue Review

addison: any comments?

w3c/bp-i18n-specdev#136

<gb> Pull Request 136 Change the guidelines about establishing base direction and mention ALM (by aphillips)

addison: this one is about guidelines about establishing base direction and mention ALM

<jyasskin> Speaking of not being last to know, there was progress on whatwg/html#7039 (comment) `Element.currentLang`, this week.

<gb> Issue 7039 Proposal for Element.currentLang and Element.currentDir (by claviska) [addition/proposal] [needs implementer interest] [i18n-tracker]

addison: fixes 2 of the oldest issues against specdev

w3c/bp-i18n-specdev#133

<gb> Pull Request 133 Add a new guideline about ruby markup (by xfq)

r12a: I had a quick look at specdev #136, you might need to explain or point to an explanation of what ALM does

<gb> Issue 136 not found

r12a: which most people have no clue

addison: it's only 11 years old

r12a: it's a bit of a weird one

<r12a> https://r12a.github.io/scripts/arab/block.html#char061C

addison: I resisted the temtation to add the descriptoin
… but I can add that

r12a: ALM determines the order of digits in a sequence
… it's used for Arabic, @@, and Syriac
… but it's not used for Persian
… I mean the Arabic language, not the Arabic script

[addison does a demo]

<r12a> https://r12a.github.io/scripts/arab/arb.html#expressions

w3c/bp-i18n-specdev#133

addison: the second PR is the new guidelines about ruby markup
… it is about our earlier conversation about ruby

jyasskin: I feel like it's a mistake, but I don't have the context to actually say that authoritatively

addison: we can hold off on it since we actually had a conversation today

w3c/i18n-glossary#77

<gb> Pull Request 77 Update orthographic syllable related definitions (by r12a)

r12a: we can publish it and if it turns out to be incorrect advice then we can change it

addison: this one is against the glossary

xfq: this is to address the comments from Norbert Lindenberg

ACTION: richard: tidy specdev PR#77

<gb> Created action #133

w3c/bp-i18n-specdev#21

<gb> Issue 21 Value of default direction (by r12a) [help wanted] [devt] [Agenda+]

<r12a> #133 note w3c/bp-i18n-specdev#21

<gb> Added comment

addison: Value of default direction

w3c/bp-i18n-specdev#21

addison: it was proposed to make auto the default direction in place of ltr

https://w3c.github.io/bp-i18n-specdev/#sec_default_base

jyasskin: does this have any implication on implementations?

addison: it should have no implications for impls
… this is guidance to spec writers

jyasskin: OK

addison: some standards have a longstanding default base direction of ltr
… auto seems reasonable
… since in most cases if you have strong rtl chars it's at the beginning

<jyasskin> +1

addison: auto's default is ltr
… in the absence of any strong rtl chars

addison: is there any objection to my making the change?

xfq: sounds good to me

Preparing for WHATWG

addison: a bunch of of bidi issues

w3c/i18n-activity#1819

<gb> Issue 1819 Should dir=auto with no strong characters inherit directionality from parent or be ltr? (by w3cbot) [close?] [tracker] [s:html] [i:bidi_text] [spec-type-issue] [alreq] [hlreq] [t:bidi_markup] [Agenda+I18N+WHATWG] [whatwg]

addison: I have a call out on an issue that is related to bidi which is ^

whatwg/html#10097

<gb> CLOSED Issue 10097 Should dir=auto with no strong characters inherit directionality from parent or be ltr? (by dbaron) [i18n-tracker] [i18n-alreq] [i18n-hlreq]

addison: fantasai has expressed the opinion that dir=auto @@

r12a: @@ ALM

[addison demonstrates the issue]

jyasskin: this is a probably too aggressive suggstion, bu t I wonder if the auto algorithm isn't paying enough attention to whether the content is weakly directional

addison: no, it's looking for a strongly directional char
… if it doesn't see one, is uses its default

jyasskin: thinking about how to get the impls to change, whatever we write in the spec, the impls will be highly constrained by compat
… unless there's a strong argument
… if the person that broke was clearly doing the wrong thing, that has a chance of convincing the employers
… but if they weren't clearly doing the wrong thing, then probably everyone's gonna prioritize compat

addison: you can see that it's complicated

jyasskin: yes

jyasskin: just reading the bug it looks like all 3 impls are inclined to not change anything, to keep auto to ltr

addison: it is very frequently ltr speakers are making decisions for rtl language users
… and it is hard to work with rtl everywhere
… as a group we tend to bias towards rtl users
… if you don't have any rtl, none of this bothers you, but if you have rtl text, this bothers you all the time
… and if you talk to individual non-technical users there, @@

jyasskin: I think your bias is correct
… you should prioritize the needs of the rtl speakers

[r12a shows a demo]

addison: different Arabic script languages do have different preferences wrt neutral separated number sequences

r12a: Arabic and Persian would expect it behaviour differently

jyasskin: that makes it really hard to justify a change in the default

addison: that's only one corner case, though
… there are a lot of neutrals that are not number neutrals in weak directionals that are not numbers
… because you want the isolation
… and don't want spillover effects
… between the value that's being inserted inside the span and the value on the outside

jyasskin: this is again for convincing the implementers, not just arguing

r12a: yeah

[addison shows a demo]

jyasskin: I like your currency example because it shows broken behaviour with the current thing in a realistic situation
… I would suggest posting this example at least
… I don't want to wait on what this group prefers

addison: this example includes strong chars inside of the isolate, and this issue is about only neutrals and weaks

jyasskin: it would be good to have a realistic example where you have a reason to isolate

Break (until 11:00 Pacific)

WHATWG prep (continued)

r12a: I was hoping the blackletter discussion would create further discussions about how the browser knows what to do when somebody says they want more

https://www.w3.org/events/meetings/7b543379-edbe-4922-b6c4-91f0b03bc8c2/

addison: I updated the agenda for the whatwg meeting with some links
… if you get a chance you might want to click on the summary links
… do we want to spend more time on base direction of neutral isolates?
… I got an email from Shane Carr who wants to dial into our meeting with whatwg
… I had an action to talk to him
… he's the author of semantic skeletons
… draft at Ecma
… adding time formatting attributes to input type=@@
… in our Monday meeting we also talked about the fact that HTML has language about respecting the locale of an input type=@@
… but it's inconsistently implemented

https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3AAgenda%2BI18N%2BWHATWG

https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3At%3Aloc_time

https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3At%3Abidi_markup

addison: as I mentioned before, there's a couple of broad categories
… if you go to t:loc_time
… then there are things related to time

https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3At%3Abidi_markup+label%3Awhatwg

addison: inncluding some that are not agenda+
… any thoughts?

<r12a> is:issue is:open label:t:bidi_markup,bidi_text label:whatwg

<r12a> https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3At%3Abidi_markup%2Cbidi_text+label%3Awhatwg+

whatwg/html#7039

<gb> Issue 7039 Proposal for Element.currentLang and Element.currentDir (by claviska) [addition/proposal] [needs implementer interest] [i18n-tracker]

<gb> Issue 326 TPAC 2024 meeting (by past)

<r12a> https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3At%3Abidi_markup%2Ci%3Abidi_text+label%3Awhatwg+

addison: create a TF to do a regular call to work through these?
… with WHATWG, like our CSS call
… our total whatwg list is 86 open issues

r12a: we have 17 needs-resolution issues

r12a: +1 to creating a TF

Bert: it's an effecient way of working but it depends on whether there are people available to come to that meeting

xfq: I'm not opposed to the idea, but depends on whether we have time, the frequence, the time slot etc.

addison: let's propose it first

whatwg/html#7039

xfq: have a look at ^

addison: Lea added a comment 20h ago

whatwg/meta#326 (comment)

addison: unhappy to see this first time this morning
… the proposal is better than the current situation

r12a: I guess the question is how they're struggling with
… if you're a content author and you knew that you'd labeled certain things with a particular language tag
… then you remember what that language this was

addison: I think fantasai mentions that intermso f providing some kind of method for computeLangMatches
… so that you can do some queries like that

r12a: if ou want to dothis as returnign an boject rahter than as a string

addison: I would return the string
… if you force it to go through an ojbect like Intl.Locale
… you might get garbage

r12a: if you use :lang(fr) lang is fr-CA, taht still matches, right?

addison: yes
… but the el.computedLang is fr-CA

addison: does el.computedLang === 'fr'

r12a: we need to advise authors on how to use that in that string
… you need to get the string so that you can do something

addison: it occurs to me that computedDir might have a subtlety in it
… if the element directoin is auto
… would that return ltr from the actually computed value as opposed to just saying auto

r12a: I was just wondering the same thing

addison: I think that's a comment worth adding

r12a: why do they want to know what the direction is?
… for diagnostics could be useful sometimes

addison: if you grab an element and you change the content of this element then the value of computedLang might change

r12a: probably just want to know *at the point* when you ask what the direction/language is

addison: might be super interesting for an input box what the behaviour might change
… after some inputs been put in the field

r12a: in input fields where you don't have an explicit declaration of the direction

`dirname=foo` so `foo=ltr` on submit

<r12a> https://www.w3.org/International/questions/qa-html-dir.en.html#reportingdirection

whatwg/html#2404

<gb> Issue 2404 A tag to display date and/or time to the user in his preferred format. (by lostmsu) [addition/proposal] [needs implementer interest] [i18n-tracker]

addison: anything else we want to talk to them about?

https://www.w3.org/2024/09/font-i18n-privacy.html

r12a: for Syriac Noto try to harmonize all the styles
… including Latin
… they tend to be sort of simple and blocky
… the user should be able to choose whether they want all guards off
… or all guards on
… or something in between
… it could work a bit like cookies

xfq: let the user add an allowlist or something

The possibility of a configurable, per-user opt-in to exposing some or all User-Installed Fonts, or a per-origin opt-in, is being discussed.

r12a: in chris's document ^

r12a: Safari has implemented this
… you can only systems font
… so I never use Safari these days for multilingual stuff
… that's really bad
… although I think you could do webfonts

r12a: you could be working on an intranet
… you should be immune from this privacy issue in those situations
… so the browser shoudln't hinder from using whatever you want
… or localhost

Summary of action items

  1. richard: tidy specdev PR#77
Minutes manually created (not a transcript), formatted by scribe.perl version 229 (Thu Jul 25 08:38:54 2024 UTC).

Diagnostics

Succeeded: s/tiem/<time>/

Succeeded: s/@@/the various ways Ruby is used/

Succeeded: s/thtis/this

Succeeded: s/@@ spec/WCAG 3/

Succeeded: s/ th e / the /

Succeeded: s/it's better/the proposal is better

Maybe present: Bert

All speakers: addison, Bert, jyasskin, r12a, xfq

Active on IRC: addison, Frankie, jyasskin, r12a, xfq