tests: discuss removing dotenv environment mutation during TTS collection
维护者通常 1 天内回复
还没有人认领这个 Issue。
评估
调研方向
Read tests/test_tts.py, especially the module level load_dotenv(override=True) call and the collection time PLUGIN filtering. First reproduce the overwrite in a subprocess using a temporary .env and dummy values, then determine how provider selection and credentials should be prepared explicitly for TTS tests. Done means collect only preserves preexisting environment values while intentional TTS configuration still works.
由索引模型根据 Issue 内容生成。
描述
Problem
While auditing test import-time side effects after #8134, I noticed that tests/test_tts.py calls load_dotenv(override=True) at module scope.
At main commit 2a1c02d17cb90244dc27a8161ff4825707430665, this is at line 56:
https://github.com/livekit/agents/blob/2a1c02d17cb90244dc27a8161ff4825707430665/tests/test_tts.py#L56
Importing the module during collection can therefore replace pre-existing process environment values with values from a discovered .env, even when no test executes. Later in the same module, the PLUGIN environment variable controls parameter filtering, so preserving intentional collection-time configuration needs care.
This module already has the tts category marker. Category filtering protects --unit runs; this report concerns collection when the TTS module is selected or pytest runs without a category filter. I am not claiming the current unit gate imports it or that this causes a model download.
Verification
I parsed the test module, extracted its module-level load_dotenv(override=True) expression, and executed that expression using the real python-dotenv loader with an explicit temporary .env path. With a dummy variable initially set to from-shell and the temporary file setting it to from-dotenv, the expression overwrote the original value. No real credentials were used or inspected.
This is an isolated verification of the statement's overwrite behavior, not a full pytest collection reproduction. A regression test should exercise collection in a subprocess with a temporary .env and dummy values.
Proposed scope for discussion
Would you welcome a small follow-up that removes implicit environment mutation from importing this module and makes dotenv preparation explicit for TTS test execution?
The implementation would preserve intentional provider selection and credential setup, and verify that collect-only does not overwrite dummy environment values. Since PLUGIN is read during collection, simply moving everything into an autouse fixture would be insufficient; I would agree on the intended configuration path before implementing it.
No production behavior, changelogs, package manifests, or version files would change. I have not started an implementation and would appreciate guidance on the preferred scope.
- 主要语言
- Python
- 星标
- 14.6k
- 派生
- 3.9k
- 平均合并
- 2 天 6 小时
- 30 天内合并 PR
- 227
环境准备
这个项目没有提供开发容器、Dockerfile 或贡献指南，环境需要你自己搭建：先看它的 README，通用步骤见我们的新手贡献指南。
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
livekit/agents 的其他 Issue
-
难度 1/5 1 小时以内 新手友好度 78/100
维护者通常 1 天内回复
-
enhancement
难度 2/5 1-3 小时 新手友好度 66/100
维护者通常 1 天内回复
-
ci: unit tests pin Python 3.12 while the package advertises 3.14 support可能已有人在做 @huong8373tt-beep 于 1 天前认领。 未关闭
难度 2/5 1-3 小时 新手友好度 74/100
维护者通常 1 天内回复
-
Assistant markup is stripped from chat history when expressive mode is off可能已有人在做 @ShivamSharma43 于 22 天前认领。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
docs: openai.LLM vs openai.responses.LLM for HTTP OpenAI-compat hosts可能重新可做 @AbarnaaSree 于 62 天前认领，目前没有进行中的 PR。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 79/100
aws-samples/sample-per-model-bedrock#639 ·
维护者通常 1 天内回复
-
bug
难度 1/5 1 小时以内 新手友好度 86/100
awslabs/visual-asset-management-system#937 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 62/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 79/100
project-codeflare/codeflare-sdk#1195 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 65/100
pandas-dev/pandas-stubs#2051 ·
维护者通常 1 天内回复