I just want to say that the colour contrast on this post is atrocious.
#999999 on a white background is a 2.84:1 contrast ratio, when the minimum requirement for body text is 4.5:1. So this page technically violates the ADA and other accessibility laws.
Looks nice! But I'm always concerned about the performance impact of threejs. Considering they're never really seen in 3d, couldn't this be a CSS perspective transform on a 2d graphic to make them look squishy?
That is a great idea! I initially built them for another project where CSS was not available but it could really push the performance. Right now, their image is cached already when they are not moving but with CSS maybe one could render initial components with threejs and then animate them further with CSS, right? Or would you build them with CSS from the very beginning?
Maybe render them as an svg or a canvas element? I'm not too familiar with drawing stuff in 2d so I'm not sure, but the character creator could probably be converted to 2D somehow while still having the hairs and shading and stuff. 3D seems like overkill if they're always flat on the screen.
Three.js renders them onto canvas elements, actually, so perhaps there is an easy way to render them onto individual canvases and overlay them? It would be interesting to see how close one could get with 2D layers to the feel of the 3D version. I guess when they are that small, it would be very hard to notice, right?
The handful of popular DEs [0] are going to be much more polished and functional for like 99% of tasks, and nonexistent [1] for the other 1%. The OS settings are conveniently located in a sensibly organized system settings manager and it works great. Then you need to mount a network drive (temporarily or permanently) and simply cannot do so in a GUI. If you're comfortable dealing with that small handful of outlier tasks (and if you're here, you can probably manage), then they're all already there.
[0] I use KDE, but it's essentially true of all of them
IIRC in Thunar you just open a dialog and paste the connection string and you're done. Pretty much the same as in Windows but you don't need to execute some icky looking binary from some web page to e.g. get support for sshfs.
That was mostly just chosen as an example of something KDE lacks, but which feels like it should be possible to do in a UI. I have no doubt that Xfce has plenty of comparable examples.
But I was curious and looked up the process; it seems that Thunar can only temporarily mount a network drive. So it remains a good example of the kinds of things you can easily do in the UI in Windows or macOS and thus feel weirdly lacking in Linux.
I kind of disagree that it is done easily in Windows, network drives are pretty cursed everywhere in my experience.
But yeah, I'm a professional who lives in multiplexed terminals and none of the people who I've helped migrate to Linux were keen on NFS, CIFS, SMB and so on, so it's not like I have a solid grasp on the graphical tooling available for such things.
It has been much rarer for people to come back to me with problems after I've helped them move to Linux compared to the Windows users, though.
Yes, but I don't believe it can do so permanently. And by that I mean, solely through the UI, connect to a network drive and have it persist across reboots. You can do that in the Windows or Mac file explorers.
My initial comment muddied that complaint. You can do it transiently in the UI in Dophin.
Works for me, I have my nextcloud webdav saved there since I didn't want to deal with any client bloat. It's not only persisted across reboots, but persisted across platform upgrades (I run fedora KDE)
Many others have already replied and it is 6 days later, but I wanted to reply to say I am quite happy with the stock KDE Plasma 6 desktop. I use some themes, e.g. "Breeze AlphaBlack" gives a nice Win10-like pure black taskbar, lovely on OLED. But the rest is stock. It Just Works (tm).
OK, so I did this recently and it depends a lot on what software you're then running whether it works or not, and whether the keyboard will actually flip some Caps Lock switch somewhere regardless of being remapped.
ctrl2caps tends to break and activate Caps Lock, i.e. capital letters unless Shift is held.
The first sentence in the article immediately sounds AI written and also doesn't really make sense.
"I asked ChatGPT for the best AI note taking app. Seven words, no brand names." - The question was seven words? But later on the article says the prompt was "best AI note taking app 2026".
I don't know if there's a name for that AI writing pattern where it says something like "(Number) x, no y." but I see it all the time. It seems similar to https://tropes.fyi/tropes/negative-parallelism
Definitely. It's an inference crutch, like all the annoying uncanny features of LLM text. Models probably start doing this following reinforcement, because it's one of the nearest "paths" to staying on track with what the user wants.
"here's the honest take" (I'm switching things up so you have a less biased/overconfident answer)
"it's not x, it's y" (I don't want to get confused later about different valid interpretations of what I have so far)
and of course the original, "you're absolutely right"
#999999 on a white background is a 2.84:1 contrast ratio, when the minimum requirement for body text is 4.5:1. So this page technically violates the ADA and other accessibility laws.
reply