UTF-8 Encoding Errors Prevent Package Installation

[Paillat-dev]

Summary

uv is failing to install packages, reporting UTF-8 encoding errors. The root cause is likely non-ASCII characters in the Windows username, which uv may not be handling correctly.

Steps to Reproduce

1. Install from git:

   uv add git+https://github.com/feldberlin/timething.git
   

   Error:

   error: Failed to run `C:\Users\[USER]\AppData\Local\uv\cache\builds-v0\.tmp1pEQZM\Scripts\python.exe`
   Caused by: stream did not contain valid UTF-8
   

2. Install specific package:

   uv add timething
   

   Error:

   error: Failed to download and build `docopt==0.6.2`
   Caused by: Failed to run `C:\Users\[USER]\AppData\Local\uv\cache\builds-v0\.tmpwGOGdp\Scripts\python.exe`
   Caused by: stream did not contain valid UTF-8
   

Environment

• OS: Windows
• uv version: 0.4.11
• Python versions: uv: 3.12, project: 3.10

Impact

This issue prevents package installation, a core functionality of uv, significantly hindering development workflows for users with non-ASCII characters in their Windows usernames.

Additional Context

The root cause appears to be non-ASCII characters in the Windows username. For example, a user with the username "María García" might experience this issue, while a user with the username "John Smith" would not.

Example:

• Affected path: C:\Users\María García\AppData\Local\uv\cache\builds-v0\.tmp1pEQZM\Scripts\python.exe
• Non-affected path: C:\Users\John Smith\AppData\Local\uv\cache\builds-v0\.tmp1pEQZM\Scripts\python.exe

uv may not be correctly handling these non-ASCII characters in file paths, leading to the reported UTF-8 encoding errors.

Related Issues

• Support alternative requirements file encodings #1666
• uv pip wont accept my requirements file #2276
• uv pip install failed on Windows: stream did not contain valid UTF-8 #7549
• [Error]stream did not contain valid UTF-8 #7726

Please let me know if you need any additional information or if you'd like me to perform any specific tests to confirm this hypothesis.

[charliermarsh]

I'm sort of wondering if this is just a case of Windows long-path support not being enabled.

[Paillat-dev]

I think it was enabled already, double checked and made sure it was, restarted, but still the same issue:

uv add timething
⠸ docopt==0.6.2                                                                                                         error: Failed to download and build `docopt==0.6.2`
  Caused by: Failed to run `C:\Users\[USER]\AppData\Local\uv\cache\builds-v0\.tmpZ2ViUG\Scripts\python.exe`
  Caused by: stream did not contain valid UTF-8

[charliermarsh]

Unfortunately I haven't been able to replicate it on my Windows machine. uv pip install docopt==0.6.2 completes without error.

[kahojyun]

I found a way to reproduce it on my computer.

1. Set UV_CACHE_DIR to a directory containing non-ASCII characters like E:/缓存/uv
2. Create a new uv project and run uv add docopt

uv add docopt
⠼ docopt==0.6.2
error: Failed to download and build `docopt==0.6.2`
  Caused by: Failed to run `E:\缓存\uv\builds-v0\.tmp2qY1wK\Scripts\python.exe`
  Caused by: stream did not contain valid UTF-8

When Windows username contains non-ASCII characters, the default uv cache directory will trigger this error.

[Paillat-dev]

Right that is most likely the reason, as my username contains the é accentuated character.

[charliermarsh]

Great find, thank you for the PR!