I'm trying to set up a simple Nutanix test environment. The community edition requires an account with a lot of details to sign up - once you go through that pain, you get a link to a 6 year old forum thread with downloads that were last updated in 2024. Is this really it?
2
1
10
Second comment on first-time experience: when presented with this dialog, I had the feeling I had to choose between one or the other, but digging deeper it turns out I can do both native repos and mirrored repos. I also hesitated because I wasn't sure if it could be reverted
1
0
1
@GoEntire sometimes exposing more options causes more friction - now that I have to choose a primary region, I can't help but notice that Canada isn't available, since a US region is selected by default. I've never really asked myself that question for GitHub
2
0
4
@Crypt32 All I'm asking for is that it becomes just as easy to find the publisher ID than it is to find the certificate thumbprint, which is one of the most common ways to pin a signing certificate right now. I don't care how, it just needs to be surfaced
2
0
1
@Crypt32 The Azure Artifact Signing publisher ID is encoded in an EKU OID that begins with 1.3.6.1.4.1.311.97. - the rest of the OID is the actual publisher ID. Here's a modified screenshot that shows how it could be displayed in a much more natural manner to the end user:
1
0
1
Azure Artifact Signing introduced a stable publisher id inside the certificate that is *perfect* for pinning, but Microsoft didn't follow through. There are no easy ways to inspect this publisher id, either through the GUI, Win32 APIs or PowerShell commands
1
2
14
@Phatmandrake ooooh you mean the context size for the model. the default context window is small (266K), but I can request the long context (1.0M). I haven't played much with the long context, I get good results with the small one, at least for code reviews.
0
0
1
@Phatmandrake in any case, all I had to do was type "make fixes based on findings" and now it's all fixed 😉
0
0
0
Gemini 4 Argon is my new favorite model for deep code reviews using Copilot CLI's rubber duck agent 🦆👇
2
0
13
Fans are running loud on my laptop as I give my AI agents all the VMs they need to perform live testing of complex RDP features like session redirection in an RD Gateway farm. This would have taken forever to test by hand
2
0
26
@normandeveloper @timrogers I have a working prototype for an agentic workflow that can build and run UniGetUI as a Windows desktop app that I can control in a live localhost RDP environment, with screenshots to show before/after the changes: github.com/Devolutions/Un…
2
0
1
If you have an Android phone and receive a lot of spam calls, I highly recommend you install PrefixBlocker (prefixblocker.com). It works locally with simple prefix-based rules. I only use the phone for local calls anyway, and I just cut 80% of unsolicited calls so far
2
2
7
I worked around the permission issue for now - the next issue is that Windows Sandbox is a clean environment, uncustomized, so it takes quite a while for the agent to figure out what to do with a first-run launch of RDM to test a feature. Is there a way to customize sandbox?
0
0
1
@metulev I'm trying to use the winappcli plugin in GitHub Copilot app with Windows Sandbox - enabling Windows Sandbox is a bit difficult or confusing, but once I got it running, I'm hitting problems with file sharing permissions with the sandbox. There's an unreleased fix?
3
0
1
It's so annoying that the GitHub Copilot app can only rename a branch for a session *once* - why is that?
0
0
4
JSON or BSON are excluded, as they can't preserve 100% of PowerShell object properties the same way CLIXML would. I've tested the CBOR and Protobuf encodings to make sure all edge cases survive an encoding round trip
0
0
3
I've been experimenting with PSObject binary encodings to replace CLIXML - CBOR and Protobuf are both great candidates, with significantly smaller size, but also much faster encoding and decoding performance! Which one would you like to be supported natively in PowerShell?
4
2
18
@Phatmandrake For SQLite in PowerShell my main concern is usability: maybe I'm wrong, but reading/writing files in scripts feels a whole lot easier or natural than dealing with SQL tables and columns. Another idea I had would be to port AgentFS (backed my SQLite) as a PSProvider
1
0
3
Weird, the OpenRouter docs cover most coding agents, including those that don't officially support BYOK, but there's nothing about the GitHub Copilot app or Copilot CLI. I've used OpenRouter with the GitHub Copilot app and it works just fine - some basic docs would still be nice
4
0
8
@rcarmo @WithinRafael @anaisbetts If you want to surface PRs that are ready to go, and just need attention, we're at a point where it's easier to have developers send us the list of PRs to fast track in the IronRDP matrix room: matrix.to/#/#IronRDP:mat…
1
0
0
@Tech2Wild We have two of those in the office, is it even worth looking at those newer PCs? It feels quite similar, just with a worse form factor. Price is in CAD in the screenshot, and it increased quite a lot since we bought them
0
0
1
@DamianEdwards @davidfowl @MStrehovsky There's more to investigate on our side, but if there was a way to somehow increase that 32MB limit in ILC, it would help get the full metadata for diagnostics in the meantime. I'll see about recompiling ILC with patches
0
0
0
@DamianEdwards @davidfowl @MStrehovsky I can't even get full diagnostic metadata because it just crashes trying to compile RDM. We're hitting limits somewhere
1
0
0
@DamianEdwards @davidfowl @MStrehovsky I let GitHub Copilot run overnight. It had no problem building UniGetUI, but it struggled to build RDM with .NET NativeAOT to produce the files for sizoscope without crashing. Bad news: the .exe had grown to almost 1GB. Good news: it brought it back down to 800MB
1
0
1
@DamianEdwards @davidfowl Anyway, I've got a bigger problem just with sizoscope trying to load a 28GB SGML file with 65 million nodes into memory all at once to process it. I've got to improve sizoscope to handle large files better
1
0
2
@andrewlocknet I'm trying it... it's so large that sizoscope-cli can't process it all at once. As for the sizoscope GUI, it just crashes on it
0
0
0
@johnzabroski @Quantradin GitHub Copilot app with a parent session in one project, multiple child sessions working on another project, and inter-session communication that you can just prompt and the sessions coordinate and report back automatically without being asked
1
0
1
@mike_pendon It's still working on it, I doubt this is the kind of PR you just want to YOLO-merge, unfortunately. I'll leave it running, it could take a couple of hours to finish. If I were you I'd just look at the branch to see what the impact is so it can be done the proper way
1
0
0
@mike_pendon Performance is really good, even if there are some things that I should follow up on:
2
0
1
What's the proper pattern in Claude Code to split changes into multiple PRs that you can track in their own sessions? As Claude Code discovers things that could be worked on separately, it shows a popup that can start a new session in its worktree, but I can't prompt this
3
0
0
I figured out why "dotnet restore" performance is so bad on Windows runners as opposed to Linux runners: GitHub Actions doesn't override NUGET_PACKAGES, NUGET_HTTP_CACHE_PATH to point to the D:\ drive which is where the builds are done with faster disk I/O 👇
The new HydraFusion model in GitHub Copilot is really good, I managed to generate a fake, large .NET enterprise application with a size and complexity similar to RDM Windows for less than $100 in AI credits, for build system performance testing: github.com/mamoreau-devol…
9
2
39
@_MichalKaminski Yes, just use the workflow on AtlasOps for benchmarking building for Windows from both Windows and Linux runners. When benchmarking, do the dotnet restore separately, it varies wildly and Windows is usually much slower. Core build is slower as well
1
0
1
Alright @davidfowl, I just got copilot max, lets see how this goes.
0
0
1
How does the GitHub Copilot app fetch models from the catalog of a given provider? Is there a way this API could provide the correct values for max prompt tokens, max output tokens and supported reasoning levels? I need to look at docs every time to find the correct values
1
0
0
The Atlassian MCP server has been unstable for the past few weeks, it's unreliable, fails randomly and keeps asking for sign in. It's not just me, it's everyone in the company that's observed this, in GitHub Copilot, Claude Code and all the tools.
2
0
4
Spawning child sessions in different projects that coordinate with the parent is by far the most powerful feature in the GitHub Copilot app. I tried doing similar workflows in Claude Code but it is just too limited on this front
5
3
26
If you had unlimited access to GitHub Copilot, Claude Code, Codex and Cursor, which one would be your favorite?
34
2
9
I made ChatGPT/Codex plugins that wrap Copilot CLI, so you can now review changes using any of the models available in GitHub Copilot! Just follow instructions in the repository: github.com/awakecoding/co…
I made Claude Code plugins that wrap Copilot CLI so you can easily launch reviews using any of the models supported by GitHub Copilot. It doesn't matter how good the latest Anthropic models are, different models catch different things: github.com/awakecoding/co…
1
0
4
Why is it that Windows is consistently slower than Linux at building software *for* Windows? I've got real numbers from CI builds of RDM Windows (.NET) on Windows and Linux runners as proof: Windows has a 1.33x median penalty. It's not a one-off, I re-ran the benchmarks 5 times
19
1
48