Dev container, setup instructions, and virtual environment: conda vs venv #3590
Replies: 4 comments
-
Hi @BurtHarris Thanks for the feedback.
|
Beta Was this translation helpful? Give feedback.
-
Thank you. Am I to understand that the promptflow dev team does not use dev containers? They use conda environments instead? I'll certainly go back and retry with the quick start instructions thank. It is my hope to contribute, starting with documentation improvements, but not the core code (I've got a lot to learn about python.) I run vscode on win11. My initial intent was simply to fork and look around.
Among the log, it says:
Note: The instructions in quick start's Setup a connection for your API key were a little unclear here, my apprentice didn't understand that the OpenAI and Azure OpenAI options are alternatives. I spotted it, as I had read of the concept of connection in this context. I'll continue to move ahead with this, perhaps offer a PR with improvements to the quick start documentation. |
Beta Was this translation helpful? Give feedback.
-
I guess having the .devcontainer setup at the root of the project had the process confused, perhaps this was because I git cloned from within vscode, rather than another way. |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
Trying to get started in this repro. The instructions in dev_setup.md don't seem compatible with the base image used in the devcontainer.json file.
The dev_setup instructions start with a conda command, but conda isn't installed in the image built by the dev container.
Not having much experience with python, nor with dev containers, this has been a confusing experience, even getting many of the examples to run has been painful. I've tried using the dev container, both locally and in a github codespace.
Should I use a standard python venv virtual environment? Or install conda? Or is a virtual environment needed at all, since I'm working in a dev container?
To further complicate things, building a venv from the command line inside vscode doesn't seem to let vscode know the environment should be activated, but if I used the vscode's tool to create the venv, that seems a step forward.
Beta Was this translation helpful? Give feedback.
All reactions