I REGRET buying an nvidia adapter when I had the opportunity to buy an AMD/Radeon adapter.

During the pandemic, I purchased an GeForce GTX 1650. It's an older, Turing hardware-based card, so you'd think the driver support would be pretty mature, right? It has been NOTHING but problems.

On nouveau, it's stable, but 3d acceleration just doesn't work right. Under the nvidia open source driver, it corrupts the screen after boot and locks up entirely second later. Under the proprietary driver, it freezes on boot a good amount of the time.

Now, once I get it booted, it's solid as a rock. I've gotta crank the engine over five or six times every time I DO boot, though. If I had it to do over again, I'd definitely have stuck with AMD.

This almost seems like a good idea... if unicode weren't already shaky enough.

UTF-8 is, honestly, pretty amazing. It lets you do things like compose latin-character text, and then interpose words like 𰻞.

That's 'biáng', which is, to my understanding, a kind of Chinese noodle dish. It's apparently the most complex Chinese character, comprising more than 50 strokes. (https://www.compart.com/en/unicode/U+30EDE).

In hex it's encoded as: 0xF0 0xB0 0xBB 0x9E

So, yeah, only 8 bytes to describe a character that looks like white noise to me unless I zoom WAY in on it! (My vision's getting pretty bad, tbh. I need it to be about the size it shows up on compart.com to make out the individual radical characters.)

If you were to count strokes on 'biáng', you end up with 5 bytes to encode 11 pen strokes or 2.2 strokes per byte. At 8 bytes to 57 pen strokes, the information density goes up to 7.125 strokes per byte.

So in Latin characters provided by UTF-8, you end up with very similar storage requirements. To encode the much more complex character, you get more than 3 times the information density.

I'm always nervous when hearing about new filesystems since a certain high profile news incident a several years back.

I really, really, really hope that Kent Overstreet has a really good relationship with any partner or spouse he may or may not have.

There are a couple factors that play into future-planning. The first, and most important factor is that most people neither care what OS their hardware uses or actually need more than the barest baseline. They want to spend time with their friends doing the things their friends are doing.

This is what has allowed Android to gain such massive prominence in the mobile space. It's all that's needed to play crap web games, listen to music, watch videos, and commune on social media. Expect more and more consumer hardware to be ARM-based devices running Android for the next few years.

The next big factor is that Linux has become a sort of driver dumping ground for reputable hardware manufacturers. Want to sell a piece of hardware? Better make damn sure it's got Linux driver support so that it can be part of an Android device. This means that more manufacturers are contributing drivers and code to the rest of Linux. It doesn't necessarily mean that code that works with Linux is going to be open source or play well with others. nvidia has proven to be an absolute bastard in this regard.

I don't think that means the future for Linux is going to be dim. I do think we need to expect and plan for more corporate presence. Some of that presence will be good. It doesn't take much to be a good member of the community. However, we do need to keep our collective eyes out for nvidia-like presences that will only serve to anchor everyone else down.

Where I'd personally LIKE to see Linux going is to provide more power to older hardware. We have a wealth of hardware that's in the 10-20 year-old range that can be doing useful work. The problem there is maintainership. It's harder to get volunteers to work with older hardware. If you can get people to work on supporting that hardware, it means fewer PCBs in landfills and more doing hobbyist or scientific work.

In the 'modern' Desktop Linux space, I'd like to see a renewed focus on privacy. I'd like to see privacy features baked into the kernel alongside security features. In a lot of cases those are the same feature.

I'm... very frustrated with GIMP and its development team. I really badly need a good raster editor and GIMP is just not that editor. The GIMP team tends to discourage suggestions and volunteer work that does not originate from within their group, so I don't have a lot of hope of that changing.

Photoshop on WINE can be made to work, but it's a terribly bad solution for many reasons.

I certainly don't want to recommend either of them the way I do other applications and OSes in my list. Even the ones I mention that have frustrations are things I'm still willing to use (and enjoy) on a daily basis.

LEX writes in another reply:

Dump Gimp. Krita is the way.

Krita is great. It's not perfect, and doesn't provide some functionality I personally need. However, the Krita team really does seem intent on improving it and making it a better application, and that shows in its development and featureset. In time, I hope to completely replace anything I have to go back to GIMP or PS for.

Inkscape simply eliminated any dependence I had on other vector editors like Visio or Illustrator. It's amazingly good. I'm hoping that Krita gets to that same place in the future.

tbh, this was my first post on Lemmy after deciding to dump reddit like the hot garbage it became. When I saw that the post could have an image attachment in addition to the text, this sprang to mind immediately.