- From: saku via GitHub <noreply@w3.org>
- Date: Fri, 13 Jun 2025 12:55:16 +0000
- To: public-css-archive@w3.org
@fantasai Thank you for pointing out the previous draft. I've taken a further look at both the current specs and, as you mentioned, I believe it's better to have a restoration because of: 1. Undefined implementation tiers: Current spec lacks a clear distinction between mandatory (-9999 to 9999) vs optional (extended range) implementation, causing cross-browser inconsistency in implementation & tests 2. Missing algorithmic rules: The 2013 draft had proper algorithmic definitions for "drop ones" behaviour (e.g., 10000 → 万, not 一万) and zero handling that were removed but are essential for correct CJK representation. As the current spec doesn't mention this, implementation errors are widespread across browsers So, the above are the concerns in the specification. Implementation-wise, I can observe some bugs as follows at the moment: (1) Firefox implements extended range but breaks at ~10^10 due to missing overflow protection (in `color: red;` part in the image) (2) All implementations fail to properly implement "drop ones" rules, producing incorrect output like 一千万 instead of 千万 (in `color: orange;` part in the image) (3) Cross-browser inconsistency under -9999 and over 9999 <img width="1597" alt="shown in attached browser rendering result - red indicates overflow issues, orange shows incorrect 'ones' handling" src="https://github.com/user-attachments/assets/10a521e4-6aaa-4838-b6e7-e2cf611b1510" /> Furthermore, the current tests expect extended behaviour beyond 9999 while specs only mandate up to 9999, creating systematic test failures and implementation confusion. - https://github.com/web-platform-tests/wpt/blob/master/css/css-counter-styles/japanese-informal/counter-japanese-informal.html So, what I would propose is to restore the 2013 algorithmic approach Section 7.1.2 with corrections, defining a clear two-tier implementation (mandatory/optional), and restructure WPT tests accordingly. The current editor's draft already acknowledges this need [here](https://drafts.csswg.org/css-counter-styles/#complex-cjk:~:text=NOTE%3A%20Implementations%20are%20encouraged%20to%20research%20and%20implement%20counter%20representations%20beyond%2010k%20and%20report%20back%20to%20the%20CSS%20Working%20Group%20with%20data%20when%20a%20generally%2Daccepted%20answer%20is%20discovered.%20Some%20previous%20research%20on%20this%20topic%20is%20contained%20in%20an%20earlier%20draft.) - restoring and refining those previous definitions would solve these implementation confusion and cross-browser incompatibility issues. -- GitHub Notification of comment by sakupi01 Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/12300#issuecomment-2970304821 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Friday, 13 June 2025 12:55:17 UTC