Hello! Appreciate this might be a trivial question as I am very new to pants but I've been searching online to no avail. Our team recently migrated all our code to the

pants

framework. I need to include the

pytorch_quantization

library which can only been installed successfully from Github My

requirements.txt

now includes

pytorch-quantization @ git+<https://github.com/NVIDIA/TensorRT.git@a1820ecdf6c14b7e50d2feb4fff4cc149dd6846b#egg=pkg&subdirectory=tools/pytorch-quantization>

in the

python/BUILD

file - I've included

overrides={
    "pytorch-quantization": {"dependencies": [":default#torch"]},

When I try and generate the lockfiles,, the following error occurs

error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> See above for output.

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

How do I diagnose this? The lockfile was create successfully when I didn't include

pytorch-quantization

in the

requirements.txt

I've tried generating the lockfile with

--print-stacktrace -ldebug

but it doesn't yield anything informative

have you tried

--keep-sandboxes=on_failure

? That should let you inspect the exact commands that are being run by pants and rerun them yourself.

Can you create a small demo github repo that exposes this problem?

Apologies for late reply - I signed off for the weekend.

Can you create a small demo github repo that exposes this problem?

sure - will try! thank you just to reiterate what the main problem is I need to include

pytorch_quantization

as part of my

requirements.txt

The usual way to install the package

pip install --extra-index-url <https://pypi.nvidia.com> pytorch-quantization

causes an

ImportError

at run-time and NVIDIA recommend the way to install it is by cloning the repo and doing

• cd tools/pytorch-quantization && pip install .

hence I now includes in my

requirements.txt

pytorch-quantization @ git+<https://github.com/NVIDIA/TensorRT.git@a1820ecdf6c14b7e50d2feb4fff4cc149dd6846b#egg=pkg&subdirectory=tools/pytorch-quantization>

I can confirm this works in a normal

pip install -r requirements.txt

@brainy-petabyte-49210 did you ever figure out the issue here