[Bug]: Streaming can buffer a whole file's audio before the first keyframe

未关闭
#114 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-2 天
新手友好度
52/100
Issue 类型
缺陷
描述清晰度
描述清楚
活跃度
活跃
技术栈
rust

调研方向

Start in src/media/stream.rs around line 322, where audio packets accumulate in held_audio until the first video keyframe sets video_start; trace both where video_start is assigned and where held_audio is later drained, in copy mode. Decide on a bound (packet count or duration) or a timeout for giving up on the keyframe, and keep the existing A/V sync behaviour intact. Done means memory stays bounded for a file with no keyframe flag or a far seek, verified by a test or a synthetic input rather than a live server, since the report was not reproduced.

由索引模型根据 Issue 内容生成。

描述

bug

[!NOTE]
This issue was written by Claude (Claude Opus 5.5) from a code review session with @PhrolovaMoe. It was found by reading the code and hasn't been reproduced on a running server.

Version

0.40.0 (feat/desktop, 017a837b)

Area

Streaming / playback

What happened?

In copy mode, audio packets wait in held_audio until the first video keyframe sets video_start (src/media/stream.rs:322). The vector has no limit, so for a file whose video packets never carry the keyframe flag (or a seek that lands far from one), the stream reads the whole file's audio into memory.

Expected behavior

Memory use stays bounded however the file is laid out.

Suggested fix

Cap held_audio (by count or duration) and drop the oldest packets, or give up waiting for a keyframe after a few seconds of input.

主要语言
TypeScript
星标
4
派生
1
平均合并
4 小时 49 分钟
30 天内合并 PR
5

环境准备

  • 提供 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 没有贡献指南

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。

tinystream-dev/tinystream 的其他 Issue

查看 tinystream-dev/tinystream 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。