Can you explain what you mean by “inside the switchboard”? Maybe a photo?

Normally, you would use standoffs to mount it.

They also dominate compute. There’s still a lot of software that depends on CUDA.

A good photo can really go a long way. Back up and zoom in as much as possible to reduce perspective distortion. Try to get the camera square to the part.

Another nice trick for small parts with a flat face is a flatbed document scanner. Unlike a camera, the scanner ensures no perspective distortion. They also have a known scale (the DPI). Or, for more accuracy, you could calibrate the scale factor by scanning a ruler.

It makes way less difference to the sound than most people think. In a blind test with different string gauges, I think few people would be able to tell which is which.

Also be aware that changing string gauge also changes the tension. You will need to readjust intonation, spring tension (unless you have a fixed bridge), and possibly truss rod.

For a beginner, I would highly recommend sticking with the standard 10-46. Aside from the adjustments needed, heavier strings are also a bit harder to play. Even as an experience player, I find zero benefit of heavier gauges.

If you like OpenSCAD, you should definitely give CadQuery a try. I’ve used both, and CadQuery absolutely blows OpenSCAD out of the water.

What does the probed mesh look like? If you run multiple probe cycles, are the results consistent?

If milk jugs are acceptable, that’s probably a better option anyway. I don’t think there’s any standard for detergent bottles, so you’d have to make a different version for each type of bottle. At least in my country, milk jugs caps are fairly universal.

However, that’s not really any better for privacy. There’s absolutely nothing preventing someone from logging a history of the changes.

Looks like an SDI video switcher.

If it’s that rusted on the outside, the inside is probably a mess as well. Probably best to replace the whole thing.

https://youtube.com/watch?v=PHiFNWJXWgI&feature=share

For the vast majority of people, writing zeroes to the disk is good enough. Even nation states and other advanced attackers would need a good amount of luck to get anything useful off a single pass zeroed HDD.

If you really need more than that, then there are various standards and recommendations for multiple passes of random/zeros/ones. Or just go the destructive route.

Of course SSDs are a totally different scenario. Writing zeroes is not sufficient or recommended for those. Generally, the secure erase command is the best way. However, this requires trusting that the firmware does that properly. If you don’t trust the firmware, destructive is the only option.

They contribute to a ton of stuff. It’s too much to duplicate here, so I’ll just leave a link: https://www.redhat.com/en/about/open-source-program-office/contributions

The recent actions with CentOS and now RHEL are a huge change from what RedHat used to be. Now it’s becoming increasingly apparent that the IBM acquisition is destroying the company.

The udevadm command just notified other running components of the change. Rebooting would also work.

As someone who has been using Linux since the 90s and gone through many different init systems, I like systemd way more than any of the past ones. It makes adding services dead simple, and is much smarter about handling dependencies and optimizing startup sequences.

The main complaints I’ve seen about it seem to be people that don’t understand that systemd init is a separate thing from all the other systemd stuff. If you don’t like all the other systemd things, you don’t need to install them at all.

When it comes to mass manufacturing, inject molding is the undisputed king. 3D print times are measured in hours per part, injection molding is measured in seconds or even milliseconds per part. Injection molded parts also have a smoother surface finish and are generally stronger than 3D prints.

Setting up an injection molding run is expensive and time consuming. You have to design molds (requires specialized design skills) and have them machined (costly in time and materials). Setting up a 3D print just requires slicing a model and sending to the printer, which can be as little as a few minutes.

3D printing can also create geometries that are impossible in injection molded parts. With injection molding, there are quite a few specific design requirements to allow the plastic to flow into the mold, cool, and be ejected. With 3D printing, there fewer restrictions on the design.

So, if you want a huge number of something it’s definitely worth it to spend the up front time and money to do injection molding. If you are doing smaller quantities, need to get started fast, or require geometries that are impossible with injection molds, then 3D printing may be a better option.

https://rockylinux.org/news/2023-06-22-press-release/

While this certainly makes things difficult, I wouldn’t count Rocky out just yet.

The number of users who care about emulation is utterly insignificant compared to the hold Nvidia has on the compute market. There is a lot of stuff that either requires or is more optimized for CUDA.

Open drain means the output has no pullup. The output is either ground or high impedance. Basically the same as open collector, if you are familiar with that.

Do you have any specs for the source device? Max output current, max sink current, max low output voltage, output impedance, etc?

Currently, these systems have no way to separate trusted and untrusted input. This leaves them vulnerable to prompt injection attacks in basically any scenario involving unvalidated user input. It’s not clear yet how that can be solved. Until it has been solved, it seriously limits how developers can use LLMs without opening the application up to exploitation.