Netflix/metaflow

Improve visibility of Metaflow Code object

Open

#1,009 创建于 2022年4月13日

在 GitHub 查看
 (1 评论) (0 反应) (0 负责人)Python (7,202 star) (695 fork)batch import
enhancementhelp wanted

描述

The code tarball saved by Metaflow is accessible with the client but it is not very easy to work with. We may want to make it more visible possibly even in the Metaflow UI.

Another idea would be to link it to the source control that the Metaflow code is in (if there is one). In particular, we could store the branch/commit information as well as a "dirty" patch file if needed. We can keep the same tarball to work internally but also linking it to the user's development environment may make it easier to work with.

贡献者指南

技术栈
python
领域
backenddata
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
1-2 days
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
stale
清晰度议题是否清楚说明期望改动、验收标准和下一步。
needs investigation
前置要求
Basic familiarity with MetaflowPython programmingUnderstanding of version control (git)
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
30
研究方向
Investigate the current Metaflow client API for Code objects, specifically in the `metaflow/client.py` or similar files. Review the Metaflow UI code to understand how code visibility is currently handled. Check if there are any existing methods or properties for linking to source control. The single comment on issue #1009 may provide additional context. Consider exploring how to store branch/commit info and dirty patch in the tarball metadata, and propose a method on the Code object to expose this information.