Incus: are we macOS yet?

Hmm, hopefully you get a response from Github and can get the flag cleared.

Lets not forget about less corporate owned sites. Forgio etc?

Sure, it’s not dramatic and the data is already replicated in other places. But yeah, I’ll have to setup mirrors.

Looks like it’s up again, with a very confusing message from the support.


This is absolutely ridiculous.

Random spitball, but I’m curious if someone mistakenly flagged your repo/org for potential violation of Apple’s EULA and/or ToS since you aren’t allowed to run macOS VMs outside of Apple hardware; and they didn’t look close enough to see that you aren’t distributing prebaked VM images or OS install images.

This is probably it, although OpenCore itself is on GitHub. The thing is, nothing in the repos suggests to run macOS outside of macOS (as you can definitely run Incus with Colima on macOS and use my repositories to get macOS running, although I haven’t tried it). I think it’s the same rationale with OpenCore: you do what you want with it, but it doesn’t violate the ToS per se.
I also thought it could be related to the logo.
But anyway, the organization has been reinstated, all good.

Well, there’s also the angle of running Linux on older Intel Macs, especially now that they’re going out of support, then running MacOS in a VM on that, potentially letting you run a more recent version of the OS.

That would still be running MacOS on Mac hardware and so should be compliant with the ToS.

I’m finishing my installation guide, and a big part of it is being able to reproduce my testing environment. For that, I have developed a quick tool to setup the environment needed to compile, package and sign the installation ISOs. However, I’m stuck on very strange problems that I’m shamelessly pinning down to OVMF. I’m trying to identify the exact bits but EDK2 2025 is obviously problematic (the latest unstable OpenCore is based off of 202502, but it doesn’t seem to boot with that specific version in my case, latest stable is based off of 202405). I think I also encountered PLENTY of problems with SecureBoot, so I just rolled out a quick experiment where I sign everything with my key, so we shall see (tests are quite expensive as a recovery install from recoveryOS takes easily a good hour…). I’d be quite sad to have worked on a working SecureBoot chain if I have to recommend people to disable it…
I’m having a really hard time understanding where the problem is, but it looks like some disks are not properly probed.

Edit: well I’ll be damned, extracting all the .fd from Incus 6.14 seems to do the trick. More debugging needed, obvs.

I have spent a lot of time (and quite frankly most of my week-end) on it, and I’m happy to release Orchard, the build infrastructure for the macOS on Incus project. (By the way, this new repository deprecates some of the previous repositories advertised here.)
Orchard is a tool allowing you to set up a development environment, compile, sign and package the necessary files to get macOS up and running on Incus. Orchard is still a very primitive tool, as it has only been tested on Debian 13, relies on the apt package manager, compiles all its sources (no binary distribution yet), and is not very configurable wrt. dependencies and artifacts locations. The deal is that now that it works, reliably and reproducibly, on two different Intel machines of mine, I’m open to contributions from people wanting to port it to their distributions, because I just won’t be doing it (except maybe for Arch) :slight_smile:

I’m currently writing a quick guide, and I think that will be it; it’s nice to finally be able to get closure, 15 months after the project initially started. I remember seeing my first Apple logo on Incus more than a year ago, and a lot of work has been done since, both on Incus and on the actual installation ISOs! To be fair, all the nights spent on it were well worth it, if only for all the accumulated knowledge. (I’m obviously accepting beers if we meet)

@stgraber could you add the macOS agent build to the daily Incus builds? (I’m not seeing it)
That’d help with the final debugging. I already noticed a small bug in the install script, but I’d rather have the agent officially packaged to test it in a clean vanilla environment.

Yeah, I’ll add them to both daily and stable