Changxin Miao

希望支持import语句

3

GUI 版本很好用，感谢，希望能够支持import语句，现在`import "base.proto";`这种语法，会报`no such file or directory`错。

gzip doesn't work

3

Hey all, Thanks for your work! I'm having trouble getting gzip to work on heroku. Here is my [nginx config file](https://github.com/polyrabbit/hacker-news-digest/blob/master/config/nginx.conf.erb#L13) and here is my web app deployed on heroku...

Hacker News 是了解科技圈资讯的一个高质量数据源，但首页 30 条新闻读起来还是挺费劲的。所以我用 ChatGPT 替我们阅读新闻，直接生成摘要：  ## Hacker News Summary * 整个服务其实是个静态页面，由定时器定时部署到 GitHub Pages 上：https://hackernews.betacat.io/ * 另外还有篇中文博客，简单介绍了撰写提示词的技巧，调用 OpenAI API 的过程，以及怎样用本地类似的 GoogleT5 模型作为 API 的兜底：https://blog.betacat.io/post/2023/06/summarize-hacker-news-by-chatgpt/ * 根据反馈，中文版翻译也已上线：https://hackernews.betacat.io/zh.html

Another case to remove leaked objects automatically.

Actively flush dirty pages when under memory pressure to avoid wrong OOM killer from cgroupv1

3

Purpose: 1. Ensure staged/cached files is persisted in disk, close doesnot guarantee this - see https://linux.die.net/man/2/close 2. Avoid massive dirty pages in memory, which may cause OOM - see https://github.com/kubernetes/kubernetes/issues/43916...

Short-circuit fileReader.visit function if visitor returns false

2

Avoid unnecessary loop

**What would you like to be added**: Just like `used_space` and `used_inodes`, it will be very helpful to spot capacity issues when we have similar metrics on trash. Also it...

Dedup prefetch requests at the entrance to avoid massive read amplification

1

We observed a large read amplification when user is doing an incremental but non-continuous sequential read on large amount of files. After labeling object GET requests with purpose, we found...

We observed several volumes having much larger usage than the actual data size. It's confirmed when two of them are emptied but still have a large usage, and `juicefs summary`...

Flush staging files on close as much as possible

2

Flush staging files as much as possible, making writeback safer to turn on.