W3C

– DRAFT –
Internationalization Working Group Teleconference

15 May 2025

Attendees

Present
Addison, Atsushi, Fuqiao, JcK, Richard, Zainab
Regrets
-
Chair
Addison Phillips
Scribe
xfq

Meeting minutes

Agenda Review

Action Items

<addison> https://github.com/w3c/i18n-actions/issues

<addison> #169

<gb> Action 169 reply to html 10843 and css 9832 (on aphillips) due 2025-05-08

<addison> #168

<gb> Action 168 add instructions on dark mode to i18n-editors (on xfq) due 2025-05-01

<addison> #165

<gb> Action 165 add a conformance section to suppress the respec warning to specdev (on aphillips) due 2025-04-10

<addison> #162

<gb> Action 162 poll I18N/CSS for new day/time (on aphillips) due 2025-03-25

<addison> #157

<gb> Action 157 write glossary proposal identifying options and next steps for those options (on aphillips) due 2025-02-20

<addison> #136

<gb> Issue 136 follow up on XML errata (by aphillips) [task]

addison: xml-editor list has been revived
… now you can send errata about the xml spec

<addison> #135

<gb> Action 135 follow up on XR issue 1393 about locale in session (on aphillips) due 2024-10-17

<addison> #127

<gb> Action 127 make a list of shared topics of interest between TG2 and W3C-I18N (on aphillips) due 2024-09-30

<addison> #33

<gb> Action 33 Close issues marked `close?` or bring to WG for further review (on aphillips)

<addison> #7

<gb> Action 7 Remind shepherds to tend to their awaiting comment resolutions (Evergreen) (on aphillips, xfq, himorin, r12a, bert-github) due 18 Jul 2023

<addison> #4

<gb> Action 4 Work with respec and bikeshed to provide the character markup template as easy-to-use markup (on aphillips) due 27 Jul 2023

Info Share

Zainab: my name is Zainab Rizvi
… I work at Google on Chrome
… I work on a team that looks at anti-tracking
… anti-fingerprinting measures, detection, and possible interventions
… a lot of the interventions we think about don't necessarily think about i18n because the goal is to have uniformity

r12a: Unicode 17 beta will start on 20th
… I've already updated Uniview
… but haven't released it yet

<r12a> https://r12a.github.io/scripts/index.html#scriptnotes

r12a: because there might be last minute changes to the data
… I've been going through my orthographic notes
… and updating them significantly

Review RADAR

<addison> https://github.com/orgs/w3c/projects/91/views/1

addison: we have no new incoming requests this week
… we got an early review request for DID

Pending Issue Review

<addison> https://github.com/w3c/i18n-activity/issues?q=is%3Aissue+is%3Aopen+label%3Apending

TPAC Planning

ACTION: addison: reserve two days worth of meetings at tpac

<gb> Created action #170

Discuss <cite>Limiting Access to Local Fonts</cite> proposal

addison: does anybody have any specific preferences for tpac?

<addison> https://lists.w3.org/Archives/Member/member-i18n-core/2025May/0000.html

<addison> explainers-by-googlers/limiting-local-fonts-access

Zainab: Limiting Access to Local Fonts is an early proposal by our team
… we're looking for specific feedback from this group before we incubate it further
… the motivation is to combat font-based fingerprinting

<addison> (our comment: https://github.com/explainers-by-googlers/limiting-local-fonts-access/issues/3)

<gb> Issue 3 W3C I18N WG comments (by aphillips)

Zainab: specifically looking at canvas measure text but as we all know that's not the only way
… there's also like CSS font-family that's used to uniquely identify users across the web
… very simply the proposal is to have a list-based approach
… specifically for Windows and Mac
… and then use those lists as source of truth when determining whether a font is considered user installed
… towards Safari already does
… the reason why we're just considering Mac and Windows right now is because Linux distributions are varied and we don't think a list-based approach will be sustainable
… feedback is welcome
… the first thing I wanted to talk about are use cases that cannot be covered by web fonts or the local fonts access API

r12a: I feel very poorly prepared and unfortunately I've written about these things in so many different places and so many different threads that I'm not sure I could immediately find
… the list of things I want to mention
… but let me start with a couple of things
… first, I no longer use Safari because of my work with scripts from around. the world Safari is no use to me anymore
… I hope that Chrome does not go the same way
… the system fonts are often way out of date
… even Google Noto fonts
… there are things like the Urdu font
… which has had significant changes recently to support Urdu and Kashmiri and so on
… and Mac prevents you using the latest version of the font even if you download it
… it's a pain in the neck
… I'm not sure I've covered all of them, I need to find that list that I did before but those are issues around the idea that system fonts are not enough
… the idea of using web fonts is not all the time great because Chinese fonts is too big
… some page contains a lot of languages
… needs lots of bandwidth by having to have every font downloaded

w3c/csswg-drafts#4055 has some example issues

<gb> Issue 4055 [css-fonts] incorporate mitigations for font based fingerprinting (by pes10k) [css-fonts-4] [i18n-tracker] [Needs Design / Proposal] [privacy-tracker]

r12a: people who are researching stuff like Egyptian hieroglyphs and stuff like that they want to be able to switch between fonts to understand whether they support things correctly and so on
… without any penalties or having to create web fonts before being able to do anything
… and they're doing that locally on their own system
… from my point of view they should be immune

addison: let's pause here for a second
… this has been a long discussion in a lot of places

Zainab: I have a few different thoughts
… thank you for the feedback, r12a
… extremely useful
… I'm a native Urdu speaker, I can read Kashmiri and I also find Safari unusable, so I definitely relate in that front
… you're right that we would be disadvantaging folks
… I've been thinking of a few ways we could mitigate some of these concerns
… I have a few very early ideas
… the localhost one is very obvious
… we should not have interventions in place at localhost
… I think that's agreed upon
… there's not thread model for privacy
… the other thing is opt-in mode of some sort and maybe starting with Chrome's incognito mode
… which has less users than regular browsingt
… also somewhere we're trying out like new privacy features, when a user has opted into a higer privacy mode
… and that can be perceived as an active choice being made by users
… that's one option
… the other option would be to have an explicit toggle that would apply to the entire browsing sessions
… but they would need to opt into it
… and then we can talk about how this would work for enterprise users

r12a: so if you don't do that , you can continue as we currently go on at the moment right?
… then you've got the question of educating people to say if you do want to fly under the radar then use this mode
… that sounds like an interesting way
… the "more than x fonts" thing I'd break that barrier pretty quickly
… I wrote one of the Unicode tehnical notes on Indic scripts and how they work, I'm using example for all of those scripts, so I'm probably going to hit 10 or more
… in a single article of that kind

Zainab: a quick response
… when you're talking about building web pages that use like more than whatever the limit might be would you be open to then requesting permission through the local font access API explicitly
… then you can do as many as you want, there is no limit
… it's an existing API

ACTION: zainab: revise local-font-access to exclude localhost and talk more about opt-in model

<gb> Cannot create action. Validation failed. Maybe zainab is not a valid user for w3c/i18n-actions?

Zainab: it's saved in their preferences

r12a: yes I think that's what I had suggested previously

ACTION: addison: remind zainab: revise local-font-access to exclude localhost and talk more about opt-in model

<gb> Created action #171

Zainab: I have one question
… I don't know if folks are familiar with the Windows features on demand
… which are the extrea things that you can install
… based on your region
… we were exploring to have sublists based on different countries and locals
… but we don't want to be in the business of maintaining 100 different lists
… but if it's a useful thing we're open to it

r12a: it's also a problem of determing which is the right list for you

<Bert> gb, help alias?

<gb> Bert, I am a bot to look up and create GitHub issues and

<gb> … action items. I am an instance of GHURLBot 0.5.

<gb> … Try gb, help commands or

<gb> … see https://w3c.github.io/GHURLBot/manual.html

addison: it's tricky
… the world is a big place

WebNN

https://github.com/webmachinelearning/webnn/pull/841/files

<gb> MERGED Pull Request 841 Add notes regarding label usage, provided by i18n review (by inexorabletash)

<addison> w3c/i18n-activity#1999

<gb> Issue 1999 String metadata and localization for operator labels (by xfq) [close?] [needs-resolution] [s:webnn] [t:loc_localization] [wg:webmachinelearning]

[xfq introduces the issue and PR]

addison: I think this is a classical tension
… particularly with develop facing stuff
… the statement that the string isn't intended to be natural language
… I think the notes that they inserted look good
… maybe the warning about Trojan Source is a good one but that's sort of a general purpose warning and maybe it doesn't need to be right there
… I imagine they have other strings besides this which is Unicode data
… if this is the only one then here is fine

xfq: sounds good to me

ARIA key-shortcuts

<addison> w3c/aria#2141

<gb> Issue 2141 `aria-keyshortcuts` needs attention? (by aphillips) [i18n-tracker]

<addison> https://lists.w3.org/Archives/Public/public-i18n-core/2025AprJun/0035.html

addison: aria-keyshortcuts
… we have a longstanding issue
… daniel invited us to their next call
… y'all are invited to come too

Specdev "isomorphic string" PR

<addison> https://deploy-preview-158--bp-i18n-specdev.netlify.app/#char-string-byte-oriented

addison: a couple of weeks ago I showed you the early version of this section's rewrite
… I took your feedback
… and made some changes
… hopefully better illustrating how bytestrings work

<addison> w3c/bp-i18n-specdev#158

<gb> Pull Request 158 Address isomorphic string and ByteString guidance (by aphillips)

addison: comments welcome
… particularly interested in whether we got the mustard guidelines right

Font filtering & generic fonts

r12a: we now have generic(nastaliq) in CSS
… I'm not sure whether the font filtering stuff that we were just talking about has an impact here
… one of the things you'd be able to do with generic fonts is you could author a document without calling for a specific font at all

Zainab: I'll have to look into the specific lists for these different system fonts

r12a: in Safari's case you would fall back to the system nastaliq font
… but as a user you should be able to prioritize the order in which the fonts get chosen

AOB?

Summary of action items

  1. addison: reserve two days worth of meetings at tpac
  2. zainab: revise local-font-access to exclude localhost and talk more about opt-in model
  3. addison: remind zainab: revise local-font-access to exclude localhost and talk more about opt-in model
Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).

Diagnostics

Maybe present: r12a, xfq

All speakers: addison, r12a, xfq, Zainab

Active on IRC: addison, Bert, r12a, xfq