IIRC Kubuntu/Ubuntu and DSL in 2003-5ish, and IIRC programs were compiled on the local machine back then.
I mostly sticked with Windows cause most of the 3D packages are on Windows (I’m a 3D generalist). Was exposed to centos variants while working in the industry.
After covid, I had a lot of time to get back onto GNU Linux.
I regularly do edits in my city. Its way better than google maps. Especially when you're travelling and want to visit the less tourist crowded parts of whereever you are.
Complex CAD is extremely demanding on CPU hardware. The tree is built sequentially and the math is all single threaded. Once CAD gets past a certain threshold the software needs to start tuning the way the Kernel works. The regular settings optimised for throughput and latency become a problem and the software needs away to change this. I’ve been messing with the Linux CPU scheduler to try to improve performance for FreeCAD designing complex assemblies on an older machine. I finally gave in and ordered another machine, but am still curious about CPU schedulers in general. I don’t know how other software accomplishes improved performance on the hardware. I can only speculate, but I am willing to bet there are methods used to alter kernel parameters like the CPU scheduler in programs like Solidworks. The way these things are done is probably not portable to any other kernel.
I’ve been using system76-scheduler for a while now and it works great. You can create a profile for your desired software and all of its related processes and then assign a high priority (low niceness) to them.
Matrix multiplications could be at least somehow multi threaded and few fields has been more optimized than displaying 3D. Do you mean simulations maybe?
I would have thought they were done mostly on the GPU nowadays?
Open a .step file in a text editor and you’ll understand better. All the coordinates are calculated like they appear in a step file and they are mostly relative to each other.
Windows has just become worse and worse over the years. I was building a new PC and realized I wasn’t going to give MS my money for a terrible OS when Linux was free.
I love the idea of OSM, been trying to use MagicEarth on iPhone which leverages OSM, but I run into similar issues that your describe. I’ll be honest tho, I never even thought of trying to contribute, may look into it as a little hobby in my free time.
Windows XP was the last Windows that I wanted to use. When it became totally obsolete, I upgraded to Linux Mint. I will never go back to Windows. I did not even start off using Windows. MS-DOS was my first operating system.
I’ve not contributed to the main one, but I have for the humanitarian osm team, you get recently disaster stricken areas and copy roads and buildings and the like
I don’t like flatpak or snap or any of them. System libraries exist for good reason, just because your computer is stupid fast and you have enough disk for the library of Congress a couple times over doesn’t mean you should run a veritable copy of your whole operating system for each program. IMO it’s lazy.
Sandboxing is a different thing though, if that’s the purpose then it’s doing it right.
I have a ton of flatpaks which means packages are shared between them, so no it’s not lazy or a copy of the whole system. It makes a ton of sense for stability.
Updates are diff’s so downloading and updating is fast. Not entire packages.
Making every package work with only a certain version of a dependency and hoping it is stable doesn’t make a lot of sense.
Besides that it’s only partially true (unless we speak Nix systems) That’s also not the point of it. It’s more about having runtime environment that an app was built against and tested with.
Sure. Except you gain universal compatibility for all distros that have flatpak and aren’t building all the different package formats. Makes it much more attractive for actual developers to package since it’s only done once.
There’s no right answer here, but there are definite benefits.
I’ve had many little issues since I moved to Linux years ago, most of which would never have been an issue if flatpaks were there at the time. My experience has been better with them.
Makes it much more attractive for actual developers to package since it’s only done once.
I maintain a few apps that are included into some distributions with no participation from my side apart from tagging what I consider releases in my git repo. How is doing something only once is more attractive as not doing it at all?
Because you can make sure it was done right. You don’t have to worry about bugs or other issues being the result of faulty packaging if you’re the one doing the packaging. Plus It makes reproducing bugs easier when everyone’s using the same package, and declaring the flatpak as the official package makes it much more likely that people will use the flatpak.
Yeah, that’s why Arch is almost the only distro that keeps everything installed natively. All other distros either have a troublesome workaround or only support flatpaks.
Rolling release just keeps everyone on the same pace. Yes, they break sometimes, but on the long run it just works.
I like them for the opposite reason. I’m still quite new to Linux, so I’m figuring out which software is best for me. I set up my server with Xubuntu and installed everything through Apt. I uninstalled a lot of software, but inevitably missed some things like libraries and config files.
Using Flatpak seems to keep track of everything, so uninstalling gets rid of everything that I would otherwise miss.
If it’s doing what it says on the tin, Flatpak is making my life much easier :)
Flatseal is super easy for anyone with a tech background to use. You can very quickly expand or reduce the access an app has to your system. Even below what the app comes with by default.
I do kinda wish the guis for installing flatpak apps were more forthcoming with the permissions, and possibly integrated some of the features of flatseal so you could modify the permission set before installing.
linux
Active
This magazine is from a federated server and may be incomplete. Browse more on the original instance.