That’s a very noble goal. The reason I asked was actually because I wanted to clarify the intent here a bit more so someone with more knowledge about the inner workings of Fedora, SIGs, Initiatives, etc. than us can give you an answer on where this best belongs. My gut feeling (note that, similar to you, I’m just a “regular” Fedora user with minimal packaging experience) is that the SIG would be a good match to exchange ideas in that regard. Whether there’s a SIG that focuses on GPU-based software is something I’m not aware of currently, so I can’t really comment more on this.
In regards to packaging, if we use OpenGL or Vulkan, that means that Fedora’s stack will depend on Mesa3D. This shouldn’t a huge worry, since a given packager would simply build the software with a given Mesa version. If there’s something I’m missing, let me know (since I’m not a packager).
Me neither but the good thing as you also rightfully pointed out is that there’s people here who have that experience and who can give a definitive answer. From what I know, Mesa is already installed in the base system. Vulkan and OpenGL should be too. The only thing that are usually missing if you want to build from source are the -devel packages but they’re also easy to install.
The way a spec file (which is how Fedora infra builds packages) works is that you basically define your runtime dependencies and build dependencies (they’re called requires in the spec file). To give you a concrete example, the build process for llama.cpp is currently defined like this.
I think that the biggest benefit of these forums, is that the engineers working on the stack would now be able to communicate directly with its users on the Fedora forum, since the stack would be Fedora-maintained. I believe this communication is paramount to the success of this stack, since the engineers will have direct knowledge of what the users want, along with their pain points. The engineers will be able to produce new software, if the users’ needs change.
Would you envision this as something like a standalone category similar to the one for Project Discussions or a post like the meeting minutes where interested users could comment on? I’m a little worried about the number of posts because technology moves quite fast, especially when it comes to current trends like AI.
One of my pain points with a lot of software personally, is that if I use any APIs that are exposed to other software engineers, I don’t have any avenues to figure out how the software functions, since communication with their developers is limited
I fully share that sentiment. The good thing is that with FOSS, you can at least look at how the software functions and address issues upstream. I still remember having hour long debates with HERE (the geocoding/routing provider) about their updated APIs sending bogus responses and doing the same with Vodafone half a year later… fun times but a completely different story.
This is actually a common misconception. Fedora is as much of a community owned distro as any other distro out there. The main difference is that RedHat sponsors Fedora development and many RedHat folks also actively engage in the Fedora community because RHEL is based on Fedora. But of course this also means that if Redhat has an interest in certain pieces of software, they’ll try getting it into Fedora as well giving Fedora more leverage than other distros (maybe).
Before anyone asks, yes, I’ll do the work myself if I have to, in order to get the ball rolling, although this may choke the initiative a little bit.
I think as soon as we as a community have figured out what we want, there’ll be more people who will put in some work, so don’t worry, I don’t think you’ll have to do all of it by yourself and will get help from experts in fields you’re not as experienced in