And now we have free threads so I can’t say at least you don’t have normal concurrency problems 🤣

I think the confusing part is that the rule is presented without the problem it solves.

The problem is when you take two vectors in 3d and want to find the vector orthogonal (perpendicular) to both, you have 2 valid choices.

The right hand rule is a way to pick the same one every time if you always label the two vectors you start with consistently i.e. make your thumb vector 1, pointer vector 2, then your middle finger points the direction of a perpendicular vector that matches the handedness of the hand you used.

This matters anywhere a vector cross product is used in physics, like calculating what direction an electron feels force when moving through a magnetic field. The physics doesn’t change, but you’ll have negatives in different places when comparing results calculated with a right handed convention (coordinate system) with a left handed convention.

When checking the handedness of your coordinate system, you point your thumb and pointer finger in the positive direction of any two dimensions then check if your middle finger points in the positive direction of the third dimension.

Imho no amount of paywall or legislation protects us from a dangerous model. It’s software that will eventually become widely available.

I enjoy red hat’s paid support articles that end by saying this is untested and may not work but it was added to the knowledge base 10 years ago

Don’t worry Java is alive and well on Android… For now 😹

We can already run arm seamlessly on x86 Linux, why not use Qemu-user + binfmt misc the other way around? I guess FEX must be much faster. Im also not super keen to run binaries that can’t be recompiled anyway so probably not the target audience.

Take that Java, everything is a portable binary now.

Lmao he did it again

😹 that’s great

ddrescue is probably your best bet

dd is the simplest: dd if=/path/to/disk/device of=/path/to/backup/file but it may fail with a broken device. ddrescue is similar but handles io errors appropriately and can retry bad reads.

ChromeOS does this well because it’s android, a walled garden that users aren’t allowed to break. You can buy it at Walmart, and it works well.

Other big “consumer” distro projects (Debian, Ubuntu, fedora, rhel, etc) are similar, especially if you’re installing stable releases on hardware that is supported.

The question for me is what do users want their OS to do? My guess is internet, office, print, scan, photos, games, updates, and get out of the way. Almost all big distros will give you that experience already, as long as you don’t expect to play Windows games or pick a specialized gaming distro.

Users who want to step outside using supported repos are back to googling for a solution when things are broken, and should see themselves as part of the tech-savvy group that need to fend for themselves.

Idk why I feel compelled to add this info, but / doesn’t have to be local as long as the necessary kernel modules for mounting it are available in the initrd or built into the kernel.

For me it’s I can make Linux do this when I see another system perform well, in contrast with they took my vertical taskbar in windows 11 and I have to gut the system to get it back

I do have to remind myself that I’m still used to living in a world where Linux enjoyed immunity to most “consumer” malware just because it wasn’t a popular desktop. Ultimately Linux is not more secure than any other system unless someone put in the work to make it that way.

That makes sense, I was thinking the executable is crashing without any output because something is wrong with the libraries available or executable itself. I should have made it clear that I don’t use jellyfin and the steps above are general debugging advice for applications that crash immediately.

For possibly more information on why the core is dumping (lol) try running jellyfin from the cli (probably just typing the path to the jellyfin executable and pressing enter)

If nothing interesting is printed, try adding strace before the jellyfin executable (Google strace, it intercepts all system calls and logs them) if that doesn’t work tell strace to follow forks.

Other than that you could start using binary debugging tools to see what shared libraries jellyfin is looking for? Maybe run it in gdb…

Maybe watch dmesg when you plug in the drive?

Or look at the logs later. Journalctl -k -f should also show more logs than dmesg when you plug it it which may be informative.

Parquet 4 eva

Csv is for arcane software or if you don’t know where it’s going.

Hdf5 is for Matlab interoperability

Otherwise I use parquet (orc could also work, but I never actually use it). Sometimes parquet has problems with Pandas or polars but I’ve always been able to fix it by using pyarrow

I found the solution, we’re running debug builds in prod from now on

Lol wait is a joule actually one Newton meter? 😅

Now I’m so confused

E=mc^2 so m is joule seconds^2 / meter ^ 2

F=ma so m is Newton seconds^2 / meter

A joule is 1 Newton / meter so they agree

Maybe it “worked” enough for the execs to realize it was a mistake?