Custom Dockerfile on Azure Environment with python poetry

Question

I am new to docker and environments. This could be basics but i have been trying to install packages in my pyproject.toml file in Dockerfile without success.

I have tried using poetry to export requirements.txt file and using it with the
Environment.from_pip_requirements('requirements.txt') function and a Dockerfile.

But could there be any elegant solution to use toml file directly for creating a custom environment ?

Answer 1

Thanks for the response, @Ram R
Using the Dockerfile :

FROM python:3.8-slim-buster  
ENV PYTHONUNBUFFERED=1 \  
    PYTHONDONTWRITEBYTECODE=1 \  
    PIP_NO_CACHE_DIR=1 \  
    PIP_DISABLE_PIP_VERSION_CHECK=1 \  
    POETRY_VERSION=1.1.7 \  
    PYLINT_VERSION=2.9.4  
  
RUN pip install pylint==$PYLINT_VERSION \  
    && pip install "poetry==$POETRY_VERSION"   
  
COPY pyproject.toml ./  
RUN poetry config virtualenvs.create false   

Answer 2

@Antara Das Thanks for the question. Can you please share the sample/document that you are trying. I would recommend to use yml file that is relatively easy to get from pip requirements file

from azureml.core import Environment
from azureml.core.conda_dependencies import CondaDependencies

env = Environment(“myenv”)
env.python.conda_dependencies = CondaDependencies(“my_yaml_file”)

https://learn.microsoft.com/en-us/azure/machine-learning/how-to-train-with-custom-image