Marc-André Moreau
@awakecoding.com
Remote desktop protocol expert, OSS contributor and Microsoft MVP. I love designing products with Rust, C# and PowerShell. Proud to be CTO at Devolutions. 🇨🇦
I was on a call this morning discussing how session recording could be done on a *different* Devolutions Gateway instance than the one used for the connection. Devolutions Gateway can act as both a recording server and a just-in-time connection proxy 😜
November 12, 2025 at 2:03 PM
I was on a call this morning discussing how session recording could be done on a *different* Devolutions Gateway instance than the one used for the connection. Devolutions Gateway can act as both a recording server and a just-in-time connection proxy 😜
GitHub Copilot Fight: VSCode ⚔️ Visual Studio 2026
For once, Visual Studio actually performs at the same level as VSCode! I guess there's finally hope that Visual Studio can catch up
For once, Visual Studio actually performs at the same level as VSCode! I guess there's finally hope that Visual Studio can catch up
November 12, 2025 at 1:09 AM
GitHub Copilot Fight: VSCode ⚔️ Visual Studio 2026
For once, Visual Studio actually performs at the same level as VSCode! I guess there's finally hope that Visual Studio can catch up
For once, Visual Studio actually performs at the same level as VSCode! I guess there's finally hope that Visual Studio can catch up
GitHub Copilot in VSCode against GitHub Copilot in Rider - this time I'm testing against a large .NET project (IAP Desktop) with the question "explain the various protocol tunneling features in IAP desktop along with the most relevant source files". VSCode still wins easily 🏆
November 12, 2025 at 12:54 AM
GitHub Copilot in VSCode against GitHub Copilot in Rider - this time I'm testing against a large .NET project (IAP Desktop) with the question "explain the various protocol tunneling features in IAP desktop along with the most relevant source files". VSCode still wins easily 🏆
Cursor with Composer 1 against VSCode with GitHub Copilot and Claude Sonnet 4.5, for a question that requires deep analysis of a large code base. When using the same model, both perform similarly, but with Composer 1, Cursor is noticeably faster by a few seconds
November 11, 2025 at 9:45 PM
Cursor with Composer 1 against VSCode with GitHub Copilot and Claude Sonnet 4.5, for a question that requires deep analysis of a large code base. When using the same model, both perform similarly, but with Composer 1, Cursor is noticeably faster by a few seconds
Cursor against VSCode with GitHub Copilot for a question that requires deep analysis of a large code base. Both are using Claude Sonnet 4.5, and take approximately the same time to produce a good, detailed answer. In comparison, Claude Code doesn't even come close to this (well over two minutes)
November 11, 2025 at 9:41 PM
Cursor against VSCode with GitHub Copilot for a question that requires deep analysis of a large code base. Both are using Claude Sonnet 4.5, and take approximately the same time to produce a good, detailed answer. In comparison, Claude Code doesn't even come close to this (well over two minutes)
Here's VSCode with GitHub Copilot side-by-side with Claude Code, on the same git repository, with the same question that requires deep analysis of a large code base to produce a decent answer.
GitHub Copilot: 1 minute, excellent answer
Claude Code: 2 minutes later, still waiting
GitHub Copilot: 1 minute, excellent answer
Claude Code: 2 minutes later, still waiting
November 11, 2025 at 9:09 PM
Here's VSCode with GitHub Copilot side-by-side with Claude Code, on the same git repository, with the same question that requires deep analysis of a large code base to produce a decent answer.
GitHub Copilot: 1 minute, excellent answer
Claude Code: 2 minutes later, still waiting
GitHub Copilot: 1 minute, excellent answer
Claude Code: 2 minutes later, still waiting
Is it worth keeping Visual Studio 2022 once I have Visual Studio 2026?
November 11, 2025 at 4:12 PM
Is it worth keeping Visual Studio 2022 once I have Visual Studio 2026?
💡 Pro tip for .NET devs:
If you randomly hit file locking issues during builds, lingering MSBuild worker nodes may be the cause.
🧹 Kill them:
dotnet build-server shutdown
🛠 Disable them:
setx MSBUILDDISABLENODEREUSE 1
#dotnet #MSBuild #VisualStudio #Roslyn
If you randomly hit file locking issues during builds, lingering MSBuild worker nodes may be the cause.
🧹 Kill them:
dotnet build-server shutdown
🛠 Disable them:
setx MSBUILDDISABLENODEREUSE 1
#dotnet #MSBuild #VisualStudio #Roslyn
November 11, 2025 at 2:59 PM
💡 Pro tip for .NET devs:
If you randomly hit file locking issues during builds, lingering MSBuild worker nodes may be the cause.
🧹 Kill them:
dotnet build-server shutdown
🛠 Disable them:
setx MSBUILDDISABLENODEREUSE 1
#dotnet #MSBuild #VisualStudio #Roslyn
If you randomly hit file locking issues during builds, lingering MSBuild worker nodes may be the cause.
🧹 Kill them:
dotnet build-server shutdown
🛠 Disable them:
setx MSBUILDDISABLENODEREUSE 1
#dotnet #MSBuild #VisualStudio #Roslyn
Light snow again this morning
November 8, 2025 at 3:00 PM
Light snow again this morning
First snow of the season at the cottage ❄️
November 8, 2025 at 1:12 AM
First snow of the season at the cottage ❄️
I'm trying something new at Devolutions: quick videos of me showing off GitHub Copilot in VSCode for real cases I've worked on during the week. I include subtitles to make it easy to watch without sound, and a text summary for those not sure if they want to watch it 👇
November 7, 2025 at 8:54 PM
I'm trying something new at Devolutions: quick videos of me showing off GitHub Copilot in VSCode for real cases I've worked on during the week. I include subtitles to make it easy to watch without sound, and a text summary for those not sure if they want to watch it 👇
What's the best way to experiment with all the amazing open-source models on 🤗 Hugging Face without being limited by local hardware? Something like Ollama, but cloud-hosted with enough GPU to try the latest models? I see HF has paid plans 👇
November 7, 2025 at 8:31 PM
What's the best way to experiment with all the amazing open-source models on 🤗 Hugging Face without being limited by local hardware? Something like Ollama, but cloud-hosted with enough GPU to try the latest models? I see HF has paid plans 👇
GitHub Copilot in VSCode pitfall: forgetting to select a premium model like Claude Sonnet 4.5, and leaving GPT-4.1 selected for coding tasks.
Seriously, it makes a world of difference. I've talked to more than one person that struggled with GitHub Copilot and that was the reason
Seriously, it makes a world of difference. I've talked to more than one person that struggled with GitHub Copilot and that was the reason
November 7, 2025 at 7:06 PM
GitHub Copilot in VSCode pitfall: forgetting to select a premium model like Claude Sonnet 4.5, and leaving GPT-4.1 selected for coding tasks.
Seriously, it makes a world of difference. I've talked to more than one person that struggled with GitHub Copilot and that was the reason
Seriously, it makes a world of difference. I've talked to more than one person that struggled with GitHub Copilot and that was the reason
Want Claude Skills in VSCode with GitHub Copilot? All you need is the right set of custom instructions 😎
November 5, 2025 at 7:05 PM
Want Claude Skills in VSCode with GitHub Copilot? All you need is the right set of custom instructions 😎
Oopsie. It looks like running 3 instances of VSCode each migrating a batch of 10 WinForms dialogs to Avalonia in parallel with GitHub Copilot is enough to hit rate limits 🤷
November 2, 2025 at 2:43 PM
Oopsie. It looks like running 3 instances of VSCode each migrating a batch of 10 WinForms dialogs to Avalonia in parallel with GitHub Copilot is enough to hit rate limits 🤷
Full builds can take up to a few minutes. Would the LLM consume tokens just to wait? It's not a problem if the build fails early on because it triggers terminal output and the LLM is like "yipee, something's happening!". If it could know that it should wait, it might work without consuming tokens
November 2, 2025 at 2:24 PM
Full builds can take up to a few minutes. Would the LLM consume tokens just to wait? It's not a problem if the build fails early on because it triggers terminal output and the LLM is like "yipee, something's happening!". If it could know that it should wait, it might work without consuming tokens
I noticed GitHub Copilot in VSCode has a "runTasks" tool to run tasks asynchronously, but when it tried doing that for "dotnet build" it thinks the build finished successfully right away, even though the task is still running, and that the build fails. I use "runCommands" tools
November 2, 2025 at 2:04 PM
I noticed GitHub Copilot in VSCode has a "runTasks" tool to run tasks asynchronously, but when it tried doing that for "dotnet build" it thinks the build finished successfully right away, even though the task is still running, and that the build fails. I use "runCommands" tools
"dotnet build" often triggers the idle warning in VSCode for long builds that don't fail right away. My ideal solution would be to capture walls of text that aren't directly fed into the LLM, but available to fetch with filters once the build completes (successfully or not)
November 2, 2025 at 2:01 PM
"dotnet build" often triggers the idle warning in VSCode for long builds that don't fail right away. My ideal solution would be to capture walls of text that aren't directly fed into the LLM, but available to fetch with filters once the build completes (successfully or not)
Image generation with Mistral AI is quite fast, and can correctly reuse elements of an existing image, instead of generating a new one from scratch. These are my dogs sunbathing at home, and this is my dock near the lake at my cottage 👇
November 2, 2025 at 1:25 AM
Image generation with Mistral AI is quite fast, and can correctly reuse elements of an existing image, instead of generating a new one from scratch. These are my dogs sunbathing at home, and this is my dock near the lake at my cottage 👇
Is there a way to tell when the next "summarizing conversation history..." is going to interrupt my flow in VSCode with GitHub Copilot? I don't mind waiting, but I'd like a visual indicator of some kind with a percentage, etc. As in, when it reaches 100%, it compresses the context for the chat.
October 29, 2025 at 7:33 PM
Is there a way to tell when the next "summarizing conversation history..." is going to interrupt my flow in VSCode with GitHub Copilot? I don't mind waiting, but I'd like a visual indicator of some kind with a percentage, etc. As in, when it reaches 100%, it compresses the context for the chat.
"Workspace Index: Remotely Indexed" - that's the part I took for granted with GitHub Copilot in VSCode. When trying Claude Code on the same repository, I immediately noticed the difference, but couldn't figure out why. There's no automatic workspace indexing for code search
October 28, 2025 at 8:32 PM
"Workspace Index: Remotely Indexed" - that's the part I took for granted with GitHub Copilot in VSCode. When trying Claude Code on the same repository, I immediately noticed the difference, but couldn't figure out why. There's no automatic workspace indexing for code search
Achievement unlocked: I maxed out my number of GitHub Copilot premium requests for the month on a GitHub Copilot Business plan 👇
October 28, 2025 at 2:05 AM
Achievement unlocked: I maxed out my number of GitHub Copilot premium requests for the month on a GitHub Copilot Business plan 👇
I really wish GitHub Copilot in VSCode had a "share" button to easily share entire chat histories with colleagues within the company. I use this feature a lot in ChatGPT to show sample chats, it's the best way to teach tricks that work in practice, by far
October 24, 2025 at 7:48 PM
I really wish GitHub Copilot in VSCode had a "share" button to easily share entire chat histories with colleagues within the company. I use this feature a lot in ChatGPT to show sample chats, it's the best way to teach tricks that work in practice, by far
Is there a way to configure GitHub Copilot in VSCode to automatically wait and poll for long-running commands that can take up to a few minutes? It's annoying to have to click "Yes" every time - I just want it to wait for the build to finish.
October 24, 2025 at 7:08 PM
Is there a way to configure GitHub Copilot in VSCode to automatically wait and poll for long-running commands that can take up to a few minutes? It's annoying to have to click "Yes" every time - I just want it to wait for the build to finish.
It was a busy week with @adamrdriscoll.com in the @devolutions.net Lavaltrie office, but we already have a few really exciting ideas for AI features in PowerShell Universal! 😁
October 24, 2025 at 6:17 PM
It was a busy week with @adamrdriscoll.com in the @devolutions.net Lavaltrie office, but we already have a few really exciting ideas for AI features in PowerShell Universal! 😁