Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

Complete recording coordination and recoverable stop semantics

未关闭
  • 哪里不对？
#3,745 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
12/100
Issue 类型
重构
描述清晰度
需要澄清
活跃度
活跃
技术栈
android, ios, typescript
领域
mobile, tooling

调研方向

Start with the linked Proposed ADR 0024 (docs/adr/0024-screen-recording-provable-signal.md) and the accepted-ADR follow-up under #3744, since this issue is an ordered design plan rather than a code edit. Coordinate with #2376 on process identity before touching any backend. Done means each backend (iOS, Android, HarmonyOS, transport) passes ordinary stop, recovery, retry, and stale-owner tests, and the superseded recovery controls are removed.

由索引模型根据 Issue 内容生成。

描述

enhancement

Purpose

Finish the shared recording lifecycle without loss of recoverable output. This continues the Proposed ADR 0024 (screen recording: one coordinator, two facts). Its implemented decisions move to an accepted ADR under #3744, and the proposal is then deleted.

Required behavior

  • Track export completion and recorder termination as two independent facts.
  • Keep the native-path disposition together with both facts.
  • Use one coordinator for a live stop and for an authorized recovery.
  • Keep signal selection and identity checks inside each backend.
  • Keep the manifest and the ownership fence.
  • Before a replacement, retain unresolved recordings by fence generation.
  • Return a playable committed export together with its recorder observation.
  • If collection, finalization, or commit fails, keep the retry material.
  • After the export completes, let recovery settle an unresolved termination.

Completion conditions

  • Each supported backend proves an ordinary stop, recovery after daemon loss, a retry after failed collection, and refusal of a stale owner.
  • A retry cannot apply processing twice.
  • A replacement cannot overwrite unresolved evidence.
  • The superseded recording recovery controls are removed only after equivalent behavior is proven.
  • HarmonyOS and transport recording use the shared finish sequence.

Decisions so far

  • Uncertain process identity is not a reason to discard usable output.
  • No public probe or signal levels: backends own the sequence.
  • Process markers are not a second lifecycle store. Keep markers only for launch windows that nothing else covers.
  • Recovery does not signal independently: that would bypass the coordinator guards.
  • Unresolved evidence is never retired.
  • Below Android API 34, keep the rotation. Evaluate unlimited recording on API 34+ separately.

Dependencies

  • Coordinate process identity with #2376.
  • Change other durable kinds only with their own failed-finish evidence.

Source proposal: https://github.com/callstack/agent-device/blob/0149438faf/docs/adr/0024-screen-recording-provable-signal.md

主要语言
TypeScript
星标
5.1k
派生
339
平均合并
12 小时 50 分钟
30 天内合并 PR
632

环境准备

  • 没有 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

callstack/agent-device 的其他 Issue

查看 callstack/agent-device 的全部 Issue

相似的 Issue

更多 TypeScript Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。