Using Copilot Effectively

[colindembovsky]

GitHub Copilot is aptly named. While some have feared that generative models will replace developers, I do not believe we are there yet: Copilot is an assistant, not a replacement. However, developers will need to adjust their skills, both to stay effective as well as to stay marketable through the disruption that the AI age is bringing.

I have a friend who is a commercial airline pilot, and asked him about how autopilot works on commercial airplanes. I think the analogy of how autopilot works is useful in framing how developers should approach learning how to use GitHub Copilot.

How Autopilot works on Commercial Flights

Most of us have flown in a commercial airplane. We all know that there are two human pilots in the cockpit, and we even know that they engage autopilot to fly the aircraft. However, even though we are all comfortable with the idea of planes flying themselves, we would be a little nervous if there were no humans in the cockpit before we take off!

Here is my understanding of how autopilot works during a commercial flights:

1. The pilot taxis the plane and takes off - the autopilot cannot take off automatically
2. Once the plane reaches around 15,000 ft altitude, the pilot engages the autopilot system. Some pilots will fly manually until they are at cruising altitude.
3. Once engaged, the autopilot is programmed to fly the plane along the current flight plan. The autopilot can navigate through bad weather and turbulence.
4. The pilots man the radios and watch for weather and wind conditions. At times, pilots will tell the autopilot to fly around weather, or change altitude to get better wind conditions.
5. The autopilot lands the plane.
6. The pilot takes over to taxi the plane.

Note: Even if the above is not 100% correct, it's good enough to make an analogy for GitHub Copilot! Errors and omissions are my own.

GitHub Copilot

Understanding how autopilot works, we can make a useful analogy when we consider GitHub Copilot:

1. Developers must "take off" since Copilot can't take off by itself (context)
2. Developers can use Copilot "mid-stream" but will need to make adjustments for "turbulence" (work in small chunks)
3. Developers must "man the radio" to monitor the code that they are writing with Copilot (good DevSecOps))
4. Copilot can "land the plane" but getting to the final destination is up to the developer (remember to solve the right problems)
5. Quality control is beyond the purview of Copilot

Let's dig into these a little deeper.

Taking off - providing context

Just as autopilot can't take off automatically, in the same way a blank project or file isn't a good way to get going with Copilot. Even before that, developers need a "flight plan" - some idea of what they are going to be coding. Spending a little time to analyze requirements and think about how code is going to be written, tested, scanned, packaged and deployed will go a long way to better productivity and efficiency.

When using Copilot, you get the best results when supplying good context - think of this as the flight plan. Context is the file that you're currently editing as well as other tabs open in the solution. If you have other files already, open a couple of them to assist Copilot. Open test files to help Copilot with tests and examples of how your methods are being called.

Where none of this exists, take time to think about what you want the code to do and write the intent in comments at the top of the file. The more context you supply, the better your results will be.

I love doing the Advent of Code in December - and using Copilot while solving the puzzles has been great. However, I think one of the main advantages of using Copilot was that it subtly changed how I develop: rather than simply diving into code, I take a few moments to think about how I can best prompt Copilot to give me what I want. This makes a big difference and I found myself spending more time thinking and less time thrashing code - which is a more fulfilling experience as well as a more productive way to code!

Prompt engineering is a phrase that is being bandied about - I think there is something to this. Successful engineers will be those that can successfully guide AI to do the right thing.

Cruising Altitude - working in small chunks

Once you have a little bit of code, you're at "cruising altitude". This is where Copilot feels like magic - there is enough context for it to generate the code that you were thinking of. Keep working in small chunks (like inside a method body or inside a loop). The narrowed context produces far better results.

Remember, Copilot is a probability engine and there is some level of randomness inherent in how it works (this is true of all large language models). Broad, vague requests (low context) tend to produce results that show much more randomness (and less meaning and utility). Narrowing the context reduces the compounding effect of the randomness and is more likely to produce meaningful code.

Man the Radio - fast feedback

While you're having fun coding with Copilot at your side, don't forget to "man the radio". Remember, code in an of itself isn't the goal - solving business problems is! Moving faster isn't an end - it's a means to an end.

Why do we want developers to be more productive and efficient? The value of going faster is that we get feedback faster. The faster we get feedback from our end-users, the faster we're able to adjust course. Scrum and Agile didn't succeed because of daily stand-ups and retros - Agile succeeded because it focused on flow and shortening the feedback loop. Copilot, by making developers more productive, is wasted unless you're shortening the feedback loop. Listen to the feedback from end-users, and adjust accordingly. This will give Copilot purpose and value beyond just developer happiness.

Land the Plane - Good DevSecOps

Landing the plane is crucial - after all, if your plane doesn't land, you can't get to your destination! Again, the landing of the plane is a means to an end - you have to get off the plane to reach your destination!

Copilot will help you land, but you'll have to taxi in yourself. Copilot is designed to speed the "inner loop" of development - but you'll have to make sure you have an efficient "outer loop" too - peer code review, build automation, linting, unit and integration testing, scanning and automated deployment are critical if you're going to get the most out of Copilot.

Having said that, some Copilot X features are bringing AI to the "outer loop" such as Copilot for PRs, which can suggest missing test cases for code changes in a PR.

Autopilot is only for flying

Copilot allows developers to move faster - which means you need to match that speed when it comes to quality gates and deployment - otherwise you'll get an impedence mismatch, which if you know your electronics, is a Bad Thing. Copilot, by making developers faster, requires your quality gates and processes to be faster.

The autopilot on planes do not check the fuel levels or the ailerons or do any of the preflight checks itself - quality control is still up to the pilots and ground crews. Copilot is not meant to do everything for you - it's meant to augment your developers and make them faster. You must have good DevSecOps practices in place to maximize your usage of Copilot.

Conclusion

GitHub Copilot is a powerful tool, but to get the most out of it developers should understand how to feed it context, work in small chunks, and ensure the rest of the DevSecOps pipeline is running smoothly.

Next steps - try out Copilot and share your insights about what you do to get it working for you!

[wswright3]

I just enrolled in the copilot free trial today. I can't get it to work with Visual Code Editor on and M1 Macbook Pro (I haven't tried on any other hardware because this is where I intend to use it). I couldn't find any copilot support sites and didn't find a discussion forum for it. Do any exist? I Googled up some info on the problem and found that the problem is common and some people have found that going back to a lower version of the copilot extension helps, but I've tried going back to multiple versions all the way down to 173 with no luck. I don't see any Icon or any sign that this is installed any way. Perhaps I don't know what I should see but based on some YouTube demos that I found, it seems like I should have a new icon for "chat" displayed in the sidebar menu but I don't.

Is there a way I can solve this problem? If not, I'll have to cancel my subscription. Thanks.

[wswright3]

I meant to say that I am trying to use it with "Visual Studio Code" Version 1.78.0 (Universal)

[colindembovsky]

Hi @wswright3 - the Chat function is part of Copilot X, so you will not see that yet. In VS Code, can you open the extensions pane and check that the Copilot extension is installed and enabled? Then in the toolbar, you should see the Copilot icon:

Also, you can hit Cmd-Shft-P to open the command palette and type "copilot" to see all the Copilot commands. Run "Collect Diagnostics" to dump some debug info.

[AlejandrosTesla8]

[Nadaa00]

I just enrolled in the copilot free trial today. I can't get it to work with Visual Code Editor on and M1 Macbook Pro (I haven't tried on any other hardware because this is where I intend to use it). I couldn't find any copilot support sites and didn't find a discussion forum for it. Do any exist? I Googled up some info on the problem and found that the problem is common and some people have found that going back to a lower version of the copilot extension helps, but I've tried going back to multiple versions all the way down to 173 with no luck. I don't see any Icon or any sign that this is installed any way. Perhaps I don't know what I should see but based on some YouTube demos that I found, it seems like I should have a new icon for "chat" displayed in the sidebar menu but I don't.

Is there a way I can solve this problem? If not, I'll have to cancel my subscription. Thanks.

[DanDevel]

I can suggest a few steps you can take to troubleshoot the issue:

Ensure Copilot is Installed: Double-check that the Copilot extension is installed in your Visual Studio Code (VS Code) editor. You can do this by going to the Extensions view (Ctrl+Shift+X) and searching for "GitHub Copilot". If it's not installed, you can install it from there.

Restart VS Code: Try restarting your VS Code editor after installing or updating Copilot. Sometimes a restart can resolve issues with extension visibility.

Check for Icon and Activation: After restarting, look for the Copilot icon in the sidebar menu of VS Code. It should appear as a speech bubble or chat icon. If you can't find it, try enabling the extension by going to the Extensions view, locating Copilot, and clicking on the "Enable" button.

Review GitHub Copilot Documentation: Visit the official GitHub Copilot documentation (https://github.com/github/copilot-preview) for any specific installation requirements or known issues. It may provide additional troubleshooting steps or direct you to relevant support resources.

Contact GitHub Support: If the issue persists, you can reach out to GitHub Support for assistance.

[dtlevin]

Copilot stopped working today. 9/10 times all I get is an error. This is really disappointing for a paid product.

Error on ghost text request FetchError: Stream closed with error code NGHTTP2_INTERNAL_ERROR

I have reinstalled it.
No firewalls, no VPN
MacBook Pro (16-inch, 2021)
Apple M1 Pro
Monterey 12.5
VSC 1.78.2

[johngagefaulkner]

For what it's worth, I have the same setup as you and I've never had even the tiniest of an issue.

Could there be a conflict with one of the extensions you have installed on VS Code?

[Santosandrade996]

Muito bom!

[magnusrydin]

Using a free organization, how will that affect telemetry ? Currently unable to see any settings on it so I guess free organizations are also considered "business level" (and thus no telemetry is collected on your code) ?

[meheller]

Weird error in Copilot X for CLI on M1 Mac:

martinheller@Martins-M1-MBP Code % ??
zsh: command not found: HF

WTF?

[Alxdelira]

GitHub Copilot is an incredible tool that revolutionizes the way we write code. With its ability to auto-complete and generate real-time code suggestions, it has become a valuable programming assistant. Copilot utilizes artificial intelligence to understand the context of the code being developed and offers relevant suggestions, saving time and effort. Its ability to learn from existing code and adapt to the developer's preferences makes it a powerful tool for increasing productivity and code quality. With GitHub Copilot, programming becomes a more fluid and efficient experience, allowing developers to focus more on the logic of the code itself and less on the syntax. It is truly impressive how Copilot can accelerate the development process and help overcome programming challenges. It is a tool that every developer should try and take advantage of to elevate the level of their projects.

[wswright3]

Thanks for the quick response! I see that it is installed. I have the icon you indicated and "Collect Diagnostics" worked. I still don't seem to know how to use it. How do I ask it to do something? I expected it to comment on my code and allow me to ask it questions. I've been using chatGPT so I expected it to do something like that. Do I need to activate it somehow? The icon you showed me indicates that it is activated since toggling it would deactivate it.

[colindembovsky]

Hey @wswright3 Copilot now does not include the "chat" experience - that's coming with Copilot Chat. Read more about Copilot X (which includes Chat) here. Here's a video showing some of these features: https://youtu.be/ZtZ0xdk5wTM

[colindembovsky]

For now, you can use Copilot as you type. See this video for a walkthrough: https://youtu.be/gZKY1GBJNHs

[iDereje]

Now, let me tell you a story, my friend.
I use Siri+ python to chat with chatGPT.
that is awesome I am telling you guys

[huangyocai]

When I used copilot to automatically generate unit tests for my c++ project, it couldn't use the classes or interfaces defined by mine correctly, resulting in a compilation failure. I feel copilot's C++ support could be better.

[hichembd]

Ok good

[cmtfelix]

Thanks

[qinshulei]

Context and prompt are key factors.

[xitxat]

just signed up 15 mins ago and WOW. It's reading my mind but better. I wish I was using this sooner.

[DanDevel]

me too my friend copilot looks amazing! did you watch the oficial video on youtube? check it out asap!

[dsk114u]

I want to work with copilot X. ;)”help!”

[colindembovsky]

You can sign up for access using the sign up buttons here: https://github.com/features/preview/copilot-x?ref=sidebar

[Hayderhammami]

Thanks

[branden1992]

GitHub Copilot is aptly named. While some have feared that generative models will replace developers, I do not believe we are there yet: Copilot is an assistant, not a replacement. However, developers will need to adjust their skills, both to stay effective as well as to stay marketable through the disruption that the AI age is bringing.

I have a friend who is a commercial airline pilot, and asked him about how autopilot works on commercial airplanes. I think the analogy of how autopilot works is useful in framing how developers should approach learning how to use GitHub Copilot.

How Autopilot works on Commercial Flights

Most of us have flown in a commercial airplane. We all know that there are two human pilots in the cockpit, and we even know that they engage autopilot to fly the aircraft. However, even though we are all comfortable with the idea of planes flying themselves, we would be a little nervous if there were no humans in the cockpit before we take off!

Here is my understanding of how autopilot works during a commercial flights:

1. The pilot taxis the plane and takes off - the autopilot cannot take off automatically
2. Once the plane reaches around 15,000 ft altitude, the pilot engages the autopilot system. Some pilots will fly manually until they are at cruising altitude.
3. Once engaged, the autopilot is programmed to fly the plane along the current flight plan. The autopilot can navigate through bad weather and turbulence.
4. The pilots man the radios and watch for weather and wind conditions. At times, pilots will tell the autopilot to fly around weather, or change altitude to get better wind conditions.
5. The autopilot lands the plane.
6. The pilot takes over to taxi the plane.

Note: Even if the above is not 100% correct, it's good enough to make an analogy for GitHub Copilot! Errors and omissions are my own.

GitHub Copilot

Understanding how autopilot works, we can make a useful analogy when we consider GitHub Copilot:

1. Developers must "take off" since Copilot can't take off by itself (context)
2. Developers can use Copilot "mid-stream" but will need to make adjustments for "turbulence" (work in small chunks)
3. Developers must "man the radio" to monitor the code that they are writing with Copilot (good DevSecOps))
4. Copilot can "land the plane" but getting to the final destination is up to the developer (remember to solve the right problems)
5. Quality control is beyond the purview of Copilot

Let's dig into these a little deeper.

Taking off - providing context

Just as autopilot can't take off automatically, in the same way a blank project or file isn't a good way to get going with Copilot. Even before that, developers need a "flight plan" - some idea of what they are going to be coding. Spending a little time to analyze requirements and think about how code is going to be written, tested, scanned, packaged and deployed will go a long way to better productivity and efficiency.

When using Copilot, you get the best results when supplying good context - think of this as the flight plan. Context is the file that you're currently editing as well as other tabs open in the solution. If you have other files already, open a couple of them to assist Copilot. Open test files to help Copilot with tests and examples of how your methods are being called.

Where none of this exists, take time to think about what you want the code to do and write the intent in comments at the top of the file. The more context you supply, the better your results will be.

I love doing the Advent of Code in December - and using Copilot while solving the puzzles has been great. However, I think one of the main advantages of using Copilot was that it subtly changed how I develop: rather than simply diving into code, I take a few moments to think about how I can best prompt Copilot to give me what I want. This makes a big difference and I found myself spending more time thinking and less time thrashing code - which is a more fulfilling experience as well as a more productive way to code!

Prompt engineering is a phrase that is being bandied about - I think there is something to this. Successful engineers will be those that can successfully guide AI to do the right thing.

Cruising Altitude - working in small chunks

Once you have a little bit of code, you're at "cruising altitude". This is where Copilot feels like magic - there is enough context for it to generate the code that you were thinking of. Keep working in small chunks (like inside a method body or inside a loop). The narrowed context produces far better results.

Remember, Copilot is a probability engine and there is some level of randomness inherent in how it works (this is true of all large language models). Broad, vague requests (low context) tend to produce results that show much more randomness (and less meaning and utility). Narrowing the context reduces the compounding effect of the randomness and is more likely to produce meaningful code.

Man the Radio - fast feedback

While you're having fun coding with Copilot at your side, don't forget to "man the radio". Remember, code in an of itself isn't the goal - solving business problems is! Moving faster isn't an end - it's a means to an end.

Why do we want developers to be more productive and efficient? The value of going faster is that we get feedback faster. The faster we get feedback from our end-users, the faster we're able to adjust course. Scrum and Agile didn't succeed because of daily stand-ups and retros - Agile succeeded because it focused on flow and shortening the feedback loop. Copilot, by making developers more productive, is wasted unless you're shortening the feedback loop. Listen to the feedback from end-users, and adjust accordingly. This will give Copilot purpose and value beyond just developer happiness.

Land the Plane - Good DevSecOps

Landing the plane is crucial - after all, if your plane doesn't land, you can't get to your destination! Again, the landing of the plane is a means to an end - you have to get off the plane to reach your destination!

Copilot will help you land, but you'll have to taxi in yourself. Copilot is designed to speed the "inner loop" of development - but you'll have to make sure you have an efficient "outer loop" too - peer code review, build automation, linting, unit and integration testing, scanning and automated deployment are critical if you're going to get the most out of Copilot.

Having said that, some Copilot X features are bringing AI to the "outer loop" such as Copilot for PRs, which can suggest missing test cases for code changes in a PR.

Autopilot is only for flying

Copilot allows developers to move faster - which means you need to match that speed when it comes to quality gates and deployment - otherwise you'll get an impedence mismatch, which if you know your electronics, is a Bad Thing. Copilot, by making developers faster, requires your quality gates and processes to be faster.

The autopilot on planes do not check the fuel levels or the ailerons or do any of the preflight checks itself - quality control is still up to the pilots and ground crews. Copilot is not meant to do everything for you - it's meant to augment your developers and make them faster. You must have good DevSecOps practices in place to maximize your usage of Copilot.

Conclusion

GitHub Copilot is a powerful tool, but to get the most out of it developers should understand how to feed it context, work in small chunks, and ensure the rest of the DevSecOps pipeline is running smoothly.

Next steps - try out Copilot and share your insights about what you do to get it working for you!

Can you restore my data to its original format from my request. I used a lot to disclose of the info

[Semsemq]

Ok

[JunLovin]

Wow, i wish use GitHub Copilot soon

[Semsemq]

👍🏻

[PriNova]

Hello Community,

I don't know if this is the right channel to ask questions about Copilot X. Please direct me if my copilot suggested me a wrong route :-)

I wonder, how big is the context size? It seems to be scoped only on the currently opened file (and where the cursor is located).
I would like to have more information about:
How long the context is before the cursor?
How long the context is after the cursor?
How long the overall context is?

Secondly, when is it possible to have context over all opened documents/workspace?

And finally, in my opinion, including web access is a very important step, because the trained data is out-dated. The programming landscape is evolving really fast, especially with the AI era. Every chat-/codebot not connected to the internet is like reading old newspaper.

Another idea comes in my mind: Automatic embedding of all GitHub Repositories (after opt-in by user/Organizations) after a commit/merge/PR etc. and this embeddings are then connected with a VecDB which Copilot X has access to.

[mspaansen]

😀

[danisztls]

Can a non-developer use it?

No. As the post says it can't takeoff or land by itself. It requires a developer to be used safely and effectively and honestly it can be helpful a handful of times but can also feed bad code.

If you configure it you can use it for things other than code like writing notes but you'd be better using something not prompted for code.

[ShadReyes]

Thank you for all the advice and help!

[wswright3]

Thanks! The videos were helpful.