You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
links to tox-uv and hatch issues for posterity, same issue comes when running either tool, tox-uv bounced me here.
I am running [tox-uv / hatch] in an azure devops yaml pipeline via bash script. the tools themselves seem to run as expected end to end, however after completion of a successful test run, a number of debug statements are written to stderr, causing the pipeline task to fail despite an exit code of 0.
The specific statements themselves are also quite odd. whilst several are just entire debug messages like '##[error]Resolved 47 packages in 1.72s', it also includes the names of some packages, and at times even just the packages version number (i.e. '##[error] + anyio==4.8.0', '##[error]==2.2.1' or even '##[error] '.
for context:
we run on linux self hosted agents, using ubuntu:20.04, and install uv via pip in eitehr a python 3.9 or python 3.10 environment (both produce the same outputs, due to our reliance on azure machine learning we cannot go beyond python 3.10 currently)
we use a fresh agent with no other installs on to install first uv, then the required packages for testing (i.e. tox, tox-uv, pytest, ...)
we dont see the same error when running uv commands manually in bash scripts, only when using these tools which wrap calls to uv.
we can write the entire output to dev/null and still hit these error messages, suggesting they may not even be being written to stderr?
these errors specifcally appear after the stage completes, i.e. if i were to run tox or hatch test and then follow it with echo statements, the error trace would appear after the subsequent echoes.
changing the rust log level to for instance trace doesnt fix the issue any further than sending to stdout / stderr does
hmmm potentially? although im not convinced this is the issue - it appears to me the logs are specifically being written somewhere other than stdout / stderr, and are being interpreted as errors when they arent. its possible running in quiet may resolve this (is there a means other than setting the rust log to 'trace' as described in the docs to achieve this when uv is run by a third party tool?) but im not sure it deals with the core issue?
thanks very much for getting back to me so quickly though, really appreciate it man!
links to tox-uv and hatch issues for posterity, same issue comes when running either tool, tox-uv bounced me here.
I am running [tox-uv / hatch] in an azure devops yaml pipeline via bash script. the tools themselves seem to run as expected end to end, however after completion of a successful test run, a number of debug statements are written to stderr, causing the pipeline task to fail despite an exit code of 0.
The specific statements themselves are also quite odd. whilst several are just entire debug messages like '##[error]Resolved 47 packages in 1.72s', it also includes the names of some packages, and at times even just the packages version number (i.e. '##[error] + anyio==4.8.0', '##[error]==2.2.1' or even '##[error] '.
for context:
tox
orhatch test
and then follow it with echo statements, the error trace would appear after the subsequent echoes.example code:
with output:
The text was updated successfully, but these errors were encountered: