Objectives

The learning objectives for this practical are:

• Configure Visual Studio (VS) Code for AI pair programming with GitHub Copilot.
• Learn using GitHub Copilot for AI pair programming.
• Learn how to give context to GitHub Copilot to get better code suggestions.

Setup and background

To do this practical you need an installation of the text editor VS Code, R and RStudio. You can find the instructions in the setup link on how to install, VS Code, R and RStudio in your system. Make a directory called practical11 for this practical.

Adding the GitHub Copilot extension to Visual Studio Code

To add the GitHub Copilot extension to VS Code, follow these steps:

1. Open VS Code.

2. Click on the Extensions icon in the left sidebar (or press Ctrl+Shift+X).
     
     

3. In the search bar, type “GitHub Copilot” and press Enter, and once you see the GitHub Copilot extension in the search results, click on the Install button.
     
     

   This will install in fact two extensions, GitHub Copilot and GitHub Copilot chat. The first one allows you to get in-line code suggestions as you type, while the second one is a conversational AI assistant that allows you to have a chat with GitHub Copilot to help you with your coding-specific tasks.

Link your GitHub account to Visual Studio Code

Once the GitHub Copilot extension is installed, you need to authenticate your GitHub account in VS Code. This will only work if your GitHub account has access to GitHub Copilot. If you don’t have access to GitHub Copilot, you can request access as a student by following the instructions in this link.

To authorise VS Code to access your GitHub account, follow these steps:

1. Click on the GitHub Copilot icon in the bottom right corner of the VS Code window, and then click on Sign in to use AI Features.
     
     
2. This will open a popup window in VS Code asking you about what option you want to use to sign in to your GitHub account. Click on Continue with GitHub.
     
     
3. This will open a new browser window where you will be asked to sign in to your GitHub account, if you have not done it yet, and then authorise VS Code to access your GitHub account. Click on Continue.
     
     
4. In the next web page, you will be asked to authorise VS Code to access your GitHub account. Click on Authorise Visual Studio Code.
     
     
5. Once you have authorised VS Code to access your GitHub account, you should see a message in the web page saying VS Code is now authorised to access your GitHub account. You can now close the web page and go back to VS Code, where you may be asked whether you trust the authors of the files in this workspace. Click on Trust Workspace & Continue.
     
     

Start using GitHub Copilot

If you GitHub acccount is properly linked to VS Code, you should see a tiny Copilot icon in the bottom right corner of the VS Code window. If this icon is not visible or is crossed out, it means that GitHub Copilot is not properly linked to your GitHub account. In such a case, go back to the previous section and make sure you have properly linked your GitHub account to VS Code.

Sometimes, the extensions may be installed and your GitHub account properly linked, but VS Code may ask you to either trust the current workspace/file that you have opened, or to click a the popup window saying Enable AI features.

To start using GitHub Copilot, open a new file in VS Code called odd_even_numbers.py, in the practical11 directory, and type the following two comment lines:

## write a program in Python that asks the user for an integer
## number and then prints out whether the number is odd or even.

As you type the second comment line, you should see a suggestion from GitHub Copilot appearing in light grey text, like in the image below. While the primary supported language for prompts and chatting with GitHub Copilot is English, you can also use other languages, such as Catalan or Spanish, so feel free to write the comments in any language you prefer.

This suggestion is the code that GitHub Copilot thinks is the most likely to be what you want to write based on the comments you have written and the context of the file. To accept the suggestion, press the Tab key, and now save the file and run it in the shell to see that it works as expected.

Exercise: Open in VS Code a new file called perfect_numbers.py in the practical10 directory and, using GitHub Copilot, write a Python program that asks the user for an integer number and then prints out whether the number is a perfect number or not. A perfect number is a positive integer that is equal to the sum of its proper divisors, i.e., excluding itself from the sum. For example, 6 is a perfect number because its positive divisors are 1, 2, and 3, and their sum is 6. Once the program is ready, save the file and run it in the shell to see that it works as expected.

Modifying code with GitHub Copilot Chat

Open the Copilot Chat extension (clicking on the right of the search text box) and ask GitHub Copilot to modify the perfect_numbers.py program to take the input number from the command line instead of asking the user for it, by typing the following prompt in the chat window:

modify this program to take the input number from the command line instead of
asking the user for it.

After some thinking, GitHub Copilot will suggest a modified version of the program, and you can accept the suggestion by clicking on the Allow button shown in the image below.

GitHub Copilot will modify the program as you have asked, and will give you the option to keep the modified code or undo the suggested edits. Click on the button Keep shown in the image below.

Now save the file, go to the Unix command line and verify that the modified program works as expected by calling it as follows:

$ python perfect_numbers.py 6

How to get explanations from GitHub Copilot

GitHub Copilot can also provide explanations for the code it suggests. For instance, in the previous example, type in the Copilot Chat window the following prompt, selecting the first option that appears in the dropdown menu as shown in the image below.:

/explain the function is_perfect_number

How to give context to GitHub Copilot

Before we start this section, close all files you have opened in VS Code, and close VS Code itself.

To get suggestions from GitHub Copilot adapted to your needs, you should give it context about what you want to do and how you want to do it. The first step to enable GitHub Copilot to learn the context in which you want to work is to create a directory specific for the project you want to work on, and then open that directory with VS Code. For instance, let’s say we want to learn how to write simple programs in Python using only while and if statements, avoiding the use of the range() function and the += adding notation. Let’s create a directory called my_first_python_programs outside the practical11 directory, and then open that directory with VS Code by typing on the shell:

$ code my_first_python_programs

VS Code will ask you whether you trust the authors of the files in this directory, and you should click on Yes, I trust the authors as shown in the image below.

Now, open the GitHub Copilot Chat extension and type the following prompt in the chat window and press Enter.

/init help writing simple programs in Python to learn programming, using only
while and if statements, avoid using the range() function and the += adding
notation. Write code outside a function.

VS Code will start scanning the files in the directory, which is empty, and will ask you whether it should scan other files in your system, click on Skip as shown in the image below.

GitHub Copilot should have created a hidden directory called .github with a Markdown file called copilot-instructions.md in it, and will ask you whether you want to keep that file, click on Keep as shown in the image below.

If somehow, VS Code has not created the hidden directory .github with the file copilot-instructions.md in it, then type in the Copilot chat window the following prompt and press Enter:

create a .github directory containing the copilot-instructions.md with that information

Using the file explorer on the left, navigate to the file copilot-instructions.md and open it to see the first draft of the instructions that GitHub Copilot has created for you. You should see something similar to the image below.

To see how GitHub Copilot uses these instructions, create a new file called perfect_numbers.py in the my_first_python_programs directory and type the following comment line in it:

## write a Python program that asks the user to enter an integer number
## and prints out whether the number is perfect or not

At the end of the second comment line that you have just typed, Copilot may keep suggesting further instructions, disregard them and press Enter to go to the next line. Then, you should see a first line of code suggested by GitHub Copilot, press Tab to accept the suggestion, and a multiple-line code chunk will be suggested by GitHub Copilot, which may not be adapted to the context you have given it. In this situation, hover the mouse over the suggested code and you should see a popup window starting with < 1/2 >, click on the right bracket > to see the next suggestion, as shown in the image below.

You should obtain a second chunk of suggested code, this time adapted to the context specified in the Markdown file copilot-instructions.md, using only while and if statements, and without using the range() function or the += adding notation, as shown in the image below. Click on the Accept button, or press the Tab key, to accept this second suggestion.

If somehow, GitHub Copilot did not suggest a second code chunk adapted to the context, then type in the Copilot chat window the following prompt and press Enter:

you did not use the copilot-instructions.md

This time, GitHub Copilot should suggest a code chunk adapted to the context specified in the Markdown file copilot-instructions.md.

You may find more comprehensive instructions on how to accept, reject or navigate through the suggestions from GitHub Copilot in VS Code in this link.

Exercise: Create a new file called prime_numbers.py in the my_first_python_programs directory and, using GitHub Copilot, write a Python program that takes a number from the command line and prints out whether the number is prime or not.

Exercise: Create a new file called relatively_primes.py in the my_first_python_programs directory and, using GitHub Copilot, write a Python program that takes two numbers from the command line and prints out whether they are relatively primes or not.

Using GitHub Copilot with R and RStudio

In this link you can find instructions on how to use GitHub Copilot with R and RStudio.