forked from DonJayamanne/pythonVSCode
-
Notifications
You must be signed in to change notification settings - Fork 1.3k
Open
Labels
feature-requestRequest for new features or functionalityRequest for new features or functionalitytriage-neededNeeds assignment to the proper sub-teamNeeds assignment to the proper sub-team
Description
Workflow
- Clone repo https://github.com/Azure-Samples/azure-search-openai-demo/ and open in vscode
- Ask copilot to generate
copilot-instructions.md - Then ask copilot to setup Python env or do something that results in model calling the tool to configure python env.
- Model will call configure python env tool (great)
- Now model will create virtual env, (great, & uses
uv, even better) - However when model tries to install packages from
requirements.txt, now things grid to a halt aspip install -r ...is very very slow compared touv.
In the repo I'm testing, this last step took 5 minutes in terminal on my Mac M1
Usinguvto install from requirements.txt took 2s.
Suggestion: support installing pacakges from requirements.txt files via a tool.
Then the whole setup (above workflow) would be complete in a matter of seconds (creating venv and installing packages using uv will be done in < 5s)
Suggestions:
- Create a new tool to install the packages from requirements.txt
- Update existing tool to support installing from
requirements.txt - As part of creating the virtual env, we should ask the user to install packages from
requirements.txt(i believe we had some code)
This should be part of the tool call itself, instead of displaying prompts to the user and taking them out of the chat view.
Note: Even if this isn't done, i think its fine as the model will end up installing fromrequirements.txt(thats what i see today)
Metadata
Metadata
Assignees
Labels
feature-requestRequest for new features or functionalityRequest for new features or functionalitytriage-neededNeeds assignment to the proper sub-teamNeeds assignment to the proper sub-team