Hacktoberfest 2026：メンテナが10月に向けて印を付けた、オープンで初心者向けの issue。 Hacktoberfest の issue を見る

dynamic: remaining SC1090/SC2020 lowering gaps for a pi-coding-agent worker (follow-up to #472; related #213 #217 #220 #221 #223)

オープン
  • 何か問題がありますか？
#577 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
35/100
issue の種類
バグ
明瞭さ
おおむね明確
活発さ
活発
技術スタック
node.js, typescript
領域
compilers

調査の方向性

Start by reproducing the scriptc build --dynamic diagnostics with the described Node 26 worker; the issue lists ten remaining sites and their codes, with the eleventh tracked separately in #471. Compare each gap with the open related work in #213, #220, and #221, and decide which cases remain untracked (notably row 2). Done means triage identifies the remaining work and the listed diagnostics are addressed or assigned clear follow-ups.

索引モデルが issue の本文から書いたものです。

説明

Summary

This is a follow-up tracking the --dynamic lowering gaps that still block one real-world worker: a Node 26 ESM program built on @earendil-works/pi-coding-agent, pi-ai, pi-mcp-adapter and typebox. These gaps are distinct from the .node addon case (#471) and the runtime-only failures (#575, #576).

With scriptc build --dynamic, the worker has 11 errors on 0.1.7 and the same 11 on main at 0d9d946. Ten of them are listed below (9× SC1090, 1× SC2020); the eleventh is the .node addon from #471. Coverage is 86/89 statements (96%). The static build has 10 errors.

Remaining diagnostics (from the main build)
# Site Code Diagnostic
1 tools: [...options.tools.map((tool) => tool.name), ...(options.extensionToolNames ?? [])] (object passed to createAgentSession, a package API) SC1090 spreading 'any' into a 'string[]' literal (only a same-element-type array spreads)
2 const tools = [...createRepositoryTools(root), createDelegateTool(...), createSubmitReportTool(...)] SC1090 spread arguments are not supported yet
3 pi.setActiveTools([...new Set([...pi.getActiveTools(), ...toolNames])]) (pi is a package ExtensionAPI) SC1090 spreading 'any' into a 'string[]' literal
4 let unsubscribe = () => {}; unsubscribe = session.subscribe((event) => { ... }) SC1090 using an 'any' value where '() => void' is expected
5 let unsubscribe = () => {}; unsubscribe = pi.events.on(EVENT, (value) => { ... }) SC1090 using an 'any' value where '() => void' is expected
6 pi.on("tool_call", (event) => { if (...) return; ...; return { block: true, reason: "...", terminate: true }; }) SC1090 functions with this signature crossing into dynamically-executed code ...
7 factory: createMcpAdapter({ config }) assigned to a field typed as the package's ExtensionFactory ((pi) => void) SC1090 using an 'any' value where '(any) => void' is expected
8 return { extensions: [], ready: Promise.resolve() } from a function whose return type is used by a package API SC1090 passing a value of type '{ extensions: number[]; ready: Promise }' into dynamically-executed ('any'-typed) code
9 const request = JSON.parse(readFileSync(0, "utf8")) as RepositoryAuditRequest; (the type contains a package TSchema-typed field) SC1090 a checked cast of 'unknown' to 'any' ...
10 lastAssistant?.content.filter((b) => b.type === "text").map((b) => b.text).join("\n").trim() ?? "" SC2020 'string[] | undefined.join' has no scriptc lowering yet

Two notes:

  • #10: #472 was closed as completed, and the fix is in 0.1.7, but this site (an optional chain continued through .filter().map().join()) still fails on both 0.1.7 and main. See my comment on #472.
  • #8: the array literal is inferred as number[] instead of the declared element type, which looks like a separate inference issue.
Related upstream work

All of the PRs below are still open and unmerged, so none of them is in a release. I haven't built them against this worker, so the mapping to rows is a guess.

  • #472: closed; row 10 still reproduces
  • #220 "fix: erase non-validatable 'as T' casts instead of SC1090 in --dynamic mode": probably row 9
  • #221 "fix(dynamic): map unmappable function params/returns and type parameters to JSVAL": probably rows 4–7
  • #213 "fix: three jsval-representation gaps in checked-dynamic lowering": probably rows 1, 3 and 8
  • #217 "fix(dynamic): lower globalThis, JSVAL property access, and 'in' on JSVAL": general JSVAL surface
  • #223 "fix(dynamic): propagate JSVAL through .then() callbacks on JSVAL-settling promises": general JSVAL surface

As far as I can tell, nothing tracks row 2 (non-any spread arguments / array spread of a function result).

Related, but not part of this issue: #471 (the .node addon, won't-fix), #575 (.ts npm sources not type-stripped) and #576 (export cycle through a namespace import).

Ask

Could these be triaged together, ideally with a decision on #213, #220 and #221? I'm happy to re-test any of those branches against the worker and report back.

Versions

scriptc 0.1.7 (npm) and main at 0d9d946 (built from source), linux-x64-gnu, Node 26.10.0.

主要言語
TypeScript
スター
5.1k
フォーク
130
平均マージ
1時間 13分
マージ済み PR（30日）
254

環境構築

このプロジェクトには開発コンテナ、Dockerfile、コントリビューションガイドがありません。まず README を読み、一般的な手順ははじめてのコントリビューションガイドを参照してください。

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

vercel-labs/scriptc のほかの issue

vercel-labs/scriptc の issue をすべて見る

似ている issue

TypeScript の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。