Coding with AI (Ollama / Continue)

If there is one task that is boring yet essential, it is commenting our code.

Commenting code is an essential practice in software development for several reasons:

1. Clarity and understanding: Comments help explain what the code does, especially in complex or non-obvious sections. They make it easier for other developers who may read or maintain the code in the future to understand it.
2. Maintenance: Code evolves over time, and well-placed comments can help developers remember the logic or intent behind certain design decisions. This makes maintenance and updates more efficient.
3. Collaboration: In collaborative development environments, comments let team members quickly understand the functionality and structure of the code, which improves communication and cooperation.
4. Documentation: Comments act as built-in documentation, explaining how functions, classes and modules are used, as well as the expected parameters and return values.
5. Troubleshooting and debugging: When a problem occurs, comments can give clues about the original logic of the code, helping to identify and fix errors more quickly.
6. Good practice: Commenting your code is part of good programming practice. It shows a professional and rigorous approach to software development.

This is where the LLM comes in! We can generate comments automatically.

Of course you still have to read them through to correct / validate them, but it is a great help all the same!

Nothing could be simpler: select the method to comment, then right-click, Continue and Write a Docstring for this code.

Once the comment has been generated, accept it with OPTION + CMD + Y.

You can also use the Write comments for this code option - in that case the LLM will also suggest changes to the code to make it clearer.

You can ask the LLM to explain what the selected code does.

This can be useful to confirm what we had understood of it.

The method is a little different: select the code to explain, then right-click, Continue and Add Highlighted Code to Context or more simply use CMD + L.

This opens a new context in the chatbot. Then ask the chatbot to explain the code by typing Explain this code under the highlighted code and pressing Enter.

You can also ask it to generate a commit message by using /commit in the prompt. Very handy!

There are loads of existing commands; for those, I suggest you read the documentation.

By downloading the Ollama model Starcoder 3b, Continue can offer programming assistance.

First, download the model:

ollama pull starcoder2:3b

Then add the following configuration to Continue:

"tabAutocompleteModel": {
"title": "Starcoder 3b",
"provider": "ollama",
"model": "starcoder2:3b"
},
"tabAutocompleteOptions": {
"useCopyBuffer": false,
"maxPromptTokens": 400,
"prefixPercentage": 0.5
}