[kixelated]

Problem

moq-dev/moq#6843 makes moq-net's Decoder::str call std::str::from_utf8, so just ts moq at the moq revision #144 moves to (c764ace) refuses std::str::from_utf8 and std::str::Utf8Error. Finishes quest/m1/ts-utf8.md.

Consumer: moq-net's coding::Decoder::str, and setup::decode_implementation and the like.

Approach

• str::from_utf8 (both core::str::from_utf8 and str's own) is rt.strFromUtf8: the fatal TextDecoder on a view of the slice, which rejects exactly what std rejects (surrogates, overlongs, past U+10FFFF, truncated), keeping a leading BOM as std does.
• On failure it reports std's valid_up_to and error_len from the scan String::from_utf8 already used.
• Utf8Error is rt.Utf8Error, a plain { validUpTo, errorLen }; valid_up_to and error_len read its fields; Display and Debug are runtime functions.
• rt.FromUtf8Error now holds an rt.Utf8Error as std's does, so the two share one Display and Debug.
• Tests: a new lang op, utf8::decodes, decodes from every offset of raw bytes or of an alphabet of UTF-8 boundary bytes and prints valid_up_to, error_len, Display, Debug, and {:#?}. caller.ts adds named cases for the BOM, surrogates, overlongs, past U+10FFFF, and truncations at the end or cut by another byte. Mutation-checked: widening the surrogate range in the runtime yields 51 differences.

Impact

• Translator output: std::str::from_utf8, str::from_utf8, Utf8Error::{valid_up_to, error_len}, and Utf8Error's Display/Debug now translate instead of being refused.
• Runtime, breaking: rt.FromUtf8Error's validUpTo and errorLen move to e.error.validUpTo and e.error.errorLen (MIGRATE.md entry). New exports: Utf8Error, strFromUtf8, fmtUtf8ErrorDebug, fmtUtf8ErrorDisplay.

Alternatives

• Keep FromUtf8Error's flat fields and duplicate the Debug: dropped for one shared representation matching std's.

Checks

• just check: pass.
• just ts test: pass (unit, fixtures incl. lang differential, runtime).
• just ts moq target/moq at c764ace: pass (translate + tsc). The pin itself stays at 88448d9; feat(ffi): export moq's owned Path #144 moves it.
• Not run: just ffi/driver/uniffi test (untouched).

Follow-ups

• just ts census at c764ace still panics in vec_view; that is quest/m1/ts-vec-view.md, still needed before feat(ffi): export moq's owned Path #144 merges.

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-28T09:16:29.268742Z	75689a5	New commits
🔒 Security Review	✅ Completed 2026-10-28T09:18:25.572495Z	75689a5	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[kixelated]

Automated review by review (OpenAI)

Reviewed commit: 5d5abe4

No actionable bugs found in the changed code. The direction is sound: reuse the fatal decoder and existing UTF-8 error scan, share error formatting between borrowed and owned decoding, and document the runtime representation change rather than maintain duplicate error logic. The differential fixtures cover offsets, BOM preservation, invalid boundary sequences, truncation, and formatting.

Verification limits: static review of the diff and relevant runtime/lowering code through GitHub; I did not run the translator, differential tests, or consumer build.

[kixelated]

Merge summary:

• Merged main twice to resolve conflicts with #158, #124, and #139.
• lang op numbers: main appended wrappers (55) and edits (56) after skips (54). This PR had inserted utf8::decodes at 54, so it moves to 57 and main's numbers stay as they are. caller.ts and the harness.ts snapshot are updated to match.
• quest/m1/ts-utf8.md stays deleted. fix(ts): view a Vec by its own element type #158 had only edited it to drop the vec-view link.
• Checks: just check and just ts test pass locally on 75689a5; CI Check passed; Codex code and security reviews of 75689a5 found nothing.

(Written by Claude Opus 5.5)