- From: Xiaoqian Wu <xiaoqian@w3.org>
- Date: Fri, 24 Apr 2026 22:25:31 +0800
- To: public-editing-tf@w3.org
Hi all, The minutes of the Web Editing WG April meeting are now available at: https://docs.google.com/document/d/1p9cUrHpwGPwEgiGjZ9nZ0ppJArYCc759sHgwJDuKbQc/ Thank you for taking part in this Google Meet tooling experiment with us. Text version: WebEditing WG Meeting Apr 9, 2026 Present: Johannes Wilm, Xiaoqian Wu, Michael Aufreiter, Shweta Bindal, Dan Clark, Rakesh Goulikar, Wenson Hsieh, Olli Pettay Chair: Johannes Wilm Summary Discussion covered Input Events composition, the caret movement spec, and AI-generated content policy for the WebEditing WG. Composition Event Web Compatibility Discussion addressed redundancy between Input Events and Composition Events, acknowledging web compatibility risks if the last input event before composition end is marked with is composing false. The preferred solution is to introduce new explicit properties like has composition ended on the input event for backward compatibility. Caret Movement Spec Issues The group updated the caret movement spec explainer and discussed whether caret movement should be based on visual screen position instead of DOM order. It was agreed that addressing empty elements with placeholders by using a <br> tag is the recommended approach to ensure space and addressability. Policy for AI-Generated Content The group established a policy that using AI for non-substantial tasks like grammar and spellchecking is acceptable. Any human participant who uploads content, even if AI-generated, must remain fully responsible for reviewing it first. Details Meeting Setup and Note-Taking Agreement: The participants agreed that a summary of the meeting, rather than a full transcript of every word, would be shared publicly. It was agreed that a scribe was not needed because the transcription feature was enabled. Clipboard API Issue 248 Review: Rakesh Goulikar brought up issue 248 in the Clipboard API to call for attention and complete the Pull Request after incorporating any necessary comments. Olli Pettay noted that there were no assigned reviewers for the PR, suggesting that Edgar or Anne be added as reviewers. Input Events Issue 176: Composition Events: Discussion centered on Input Events issue 176 regarding the redundancy between `onbeforeinput`, `input` events, and composition events. Michael Aufreiter suggested consolidating everything in the `onbeforeinput` stage to clearly distinguish all cases, acknowledging that their current implementation relies on `is composing` being true throughout the composition process. Olli Pettay shared concerns from Masayuki about web compatibility if the last input event before composition end is marked with `is composing` false, noting that some websites might not expect that change. Web Compatibility and Proposed Solutions for Composition Events: Johannes Wilm mentioned that JavaScript editors have long wanted final control over the composition process, a goal previously hindered by technical limitations on Android. Proposed solutions included implementing the Firefox approach (two events, one cancelable, one not), setting `is composing` to false for the final character, or introducing new, explicit properties like `has composition ended` and `has composition started` on the input event. Michael Aufreiter favored introducing new explicit properties as it would be backward-compatible and eliminate the need for custom state management within their application. Need for Android Input on Composition Events: The group agreed that the limiting factor in solving composition event issues has historically been Android. The primary commitment is to seek input from Android developers to determine the feasibility of proposed solutions, with Dan Clark and Rakesh Goulikar taking on the assignment of contacting those code owners or getting information via GitHub. Michael Aufreiter also raised a separate, related question about why cancelability is relevant to this issue. Charter Status and Carrot Movement Explainer: Xiaoqian Wu updated the group that the charter review is proceeding smoothly with votes in support and no objections so far. Rakesh Goulikar provided an update on issue 529, the explainer for the caret movement spec, stating they would incorporate the feedback and share an updated draft with the group by the 21st. Carrot Movement Based on Visual Screen Position: Michael Aufreiter introduced issue 533, asking whether carrot movement should be defined by visual screen position rather than DOM order, citing an example of a flipped layout where DOM order causes unintuitive cursor movement. Wenson Hsieh raised concerns about how this visual movement would interact with selection extension (Shift+Arrow) and bi-directional text, potentially requiring multi-range selection. Johannes Wilm noted that this change could be especially relevant given the rise of "pre-text" editors which allow non-continuous editing areas. Alternative Solutions for Visual Carrot Movement: The group explored alternatives to purely relying on visual order, such as specifying the order of editable elements using a mechanism similar to `tab index` . Olli Pettay mentioned that Firefox's next focusable element logic uses DOM order, which is defined by the HTML specification, and noted that selection is already complicated when reordering elements using Shadow DOM and slots. Michael Aufreiter concluded the discussion by stating that this item might not require immediate action but is a fundamental issue relevant to the caret movement spec. Issue 528: Handling Empty Elements with Placeholders: Michael Aufreiter discussed issue 528, which aims to provide a recommended approach for handling empty elements that contain a placeholder. The current proposal is to use a `<br>` tag inside the empty element to ensure space is reserved for the line height and make the element addressable, a technique already used by some editing libraries. The group noted no opposition to this approach, which would resolve inconsistencies in placeholder rendering and interaction. Firefox Up/Down Arrow Navigation (Issue 523): Michael Aufreiter requested that issue 523, concerning the inability to access absolutely positioned elements with the down key in Firefox, be made a priority. Olli Pettay agreed to ask Masayuki to look into the related bug. Wenson Hsieh confirmed that they are still investigating the three priority test Safari items, including the anchor issue. Meeting Transcript Review: Xiaoqian Wu confirmed that the meeting transcript would be shared as a Google Doc, allowing all participants a week to correct the transcript before a final, public-facing version (with any non-public parts removed) is sent to the mailing list. AI-Generated Content Policy Discussion: Xiaoqian Wu asked the working group for their opinion on the use of AI tools for generating content, such as Pull Requests or issues, to help formulate a policy guide. Johannes Wilm and Olli Pettay agreed that using AI for non-substantial tasks like grammar, spellchecking, or improving English clarity should be acceptable. The consensus was that any human participant who uploads content, regardless of whether it was AI-generated, must always remain responsible for reviewing it first. Suggested next steps [Rakesh Goulikar] Notify Shweta: Inform Shweta to add her GitHub account credentials to the W3C account. [Rakesh Goulikar] Add Reviewers: Add Edgar and potentially Anne as reviewers to PR 248. [Rakesh Goulikar, Dan Clark] Contact Google: Nudge Google/Android code owners to comment on Input Events issue 176. Request them to join a future call regarding composition event handling. [Rakesh Goulikar] Share Draft: Incorporate feedback and share the updated draft explainer for the carrot movement spec. Share the draft by the 21st. [Olli Pettay] Request Review: Ask Masayuki if he can review issue 523 regarding Firefox up/down arrow navigation. [Xiaoqian Wu] Share Transcript: Share the generated Google Doc transcript with attendees for review and correction. [Johannes Wilm] Send Minutes: Send the final, edited minutes document to the mailing list one week after the transcript is shared.
Received on Friday, 24 April 2026 14:25:31 UTC