I feel this so much.

Generally LLMs do generate some correct information, then also generate a bunch of useless and misleading text surrounding it.

Sometimes when starting out by prompting an LLM the resulting generated text makes it way faster to find what you are looking for. Other times it’s the opposite where a simple web search would have been faster. It feels like gambling for correct helpful information :(

But so many people trust LLM generated text without checking it. It’s exhausting trying to show them so mostly I don’t bother anymore…

While as a user it sucks that is exactly the reasons people do it. It takes the value away from reddit, if the content that users want to see it not there people will not go there.

What I find the best compromise is users that take their comments they had on reddit and post them again as it's own post to lemmy with the context needed. While not perfect the information is at least not lost completely and a google search in the future might actually bring someone to a lemmy instance instead of to a corporation like reddit. But that is obviously a lot of work to do, especially if you have lot of helpful comments on reddit.

My impression of lemmy changed a lot once I've read this updated from the lemmy devs from less than a month ago. TL;DR: Lemmy was developed by just two people and with reddit self-destructing everyone jumped to it, and lemmy wasn't really ready for that.

With that info I'm now all the more impressed that lemmy is working as well as it currently is and not crashing every few minutes!

As I posted in response to another comment along these lines:

While as a user it sucks that is exactly the reasons people do it. It takes the value away from reddit, if the content that users want to see it not there people will not go there.

What I find the best compromise is users that take their comments they had on reddit and post them again as it’s own post to lemmy with the context needed. While not perfect the information is at least not lost completely and a google search in the future might actually bring someone to a lemmy instance instead of to a corporation like reddit. But that is obviously a lot of work to do, especially if you have lot of helpful comments on reddit.

When you tinker and debug something on windows, you usually have little idea of what went wrong and can derive very little from the experience. At least that was the case back when I still used windows, in the XP and vista days.

I don't think that is completely fair, I feel like the reason is more that on Linux no easy to follow "solutions" to as many problems as on Windows exist. When you have a problem on Linux you most of the time have to dive deeply into the technical details. On windows it's often enough to search for a solution on the internet and follow the first tutorial (not the stupid SEO garbage sites). And once whatever problem you had is gone you don't go and try to understand why the solution worked.

That also really annoyed me a lot when I had to fix Windows problems for work, because I really like to understand why something is working or not. And after some research I actually found Sysinternals which are tools that help you dig deeper into Windows inner workings. There are also some wonderful videos on how to use those tools available by the author of those tools. And there are also books available both on how to troubleshoot with the tools and on how Windows internally works.

Edit: fiexd tyops ;)

On reddit I didn't comment much at all, because it mostly felt like posting things into the void.

Here there are mostly not so many comments that a single one gets lost and I've already had a few plesant discussions!

I'm not aware of any Website that goes into detail. There are however books that go into detail of how Windows works internally: https://learn.microsoft.com/en-us/sysinternals/resources/windows-internals

In general to poke around in Windows the sysinternals tools distributed by Microsoft are great. The developer of the Sysinternals tools also gave some talks going through how he uses the tools to debug problems that occurred. Those are freely available on the web.

But in general it is was harder to find information about Windows, than linux. Most of the time when it comes to a problem with Windows a solution is posted, that dosen't explain how the solution was discovered. I found with Linux there is often more information given. And obviously with Linux you can just look at the code in the Kernel if you need to dig deep, with Windows that is not an option.

[email protected]> Audiobookshelf

I didn't know that existed and now I love it and started up a docker container for it!

Thanks! :D

I would start out allowing mostly everything programming related and only creating more specific communities once the posts for a specific topic start spamming this one.

All nice organization doesn’t help if the created communities are not used by anyone as the seem to small to be worth the bother. That’s at least how I think about it. Just leave it open for most posts until the need arises to split specific topics off.

I run OpenSpeedTest Server.

That would be nice, but is at a whole other scale with bandwidth and storage space.

Although there is a fediverse alternative for Video Hosting: https://joinpeertube.org/

I’m actually already using mediawiki for my own notes, but the quality I write down for myself is not as good as I want to publish. 🙈

I also don’t find the style of mediawiki that nice and was specifically looking for something different that makes things look a bit more polished just from the styling itself.

But I suppose it would also have it’s benefits using a software I’m already familiar with. 🤔

Do non-US instances have to handle DMCA requests like US instance with european users have deal GDPR requests?

If you are on a smaller instance you should probably ask the admin(s) if they are okay with something like this, it would put a lot of extra strain on the server and might overload small instances.

Interesting. I've in the past specifically encoded images for a webpage I created to .webp using https://squoosh.app as I determined it visually looked best for the file size while also being supported by pretty much all modern browsers.

And since none of the other modern image formats you've listed are widely supported by many modern browsers I will probably do the same again when I have to create another website with some images. But I would hope that other formats gain more traction in browsers so that I can use them in the future.

Awesome.

In my mind Linux is something I use on servers, but on my desktop I'm far too used to Windows. And since everything I want "just works" on Windows I haven't really found the motivation to try / switch over to Linux for my desktop.

+1 for programming.dev

What benefit does rejecting image/webp give?

I like this idea, I've been thinking about running my own private instance but decided against it as I like the main feed with many different communities that larger instances offer.

I’m not quite sure what you mean with “IP Class”. What you seem to want is a different subnet for those two devices. The simplest way to do that is to just manually assign each one an IP in a different subnet than your router DHCP normally gives out.

If you want more isolation you would need to get a switch that supports creating VLANs.

And if you want remote access you’ll would need to set up a VPN on the phone and your router. Or do some port forwarding on the router.