Yeah. This isn’t the first time the news app and the core nextcloud updates have fought each other in weird and mysterious ways (for me or others). I forget how I solved it last time (I think it was a similar case of needing to manually update to bleeding edge and then tweak things) but… I just don’t care anymore.

I don’t know who is right or wrong in how nextcloud is maintained (my instinct is the nextcloud devs because… have you seen nextcloud? but also, most apps don’t have this recurring problem). But at this point, the benefits I get out of it are largely gone. And when so many issues boil down to “We need more people and resources to maintain this”, it kind of feels like getting off the train BEFORE it crashes rather than after.

I’m on the alpha and it still won’t update any of my feeds. And going through the github issues it is basically summed up as “We will do another stable release once we have a frontend developer” which is basically never. So, at best, it will work until it doesn’t and then I have to fix it myself yet again and… yeah.

And if my choice is to run an older version of nextcloud to support one app? Hell no.

To be fair, that is true of all miniatures and model kits. It is generally less noticeable if you are dealing with injection molded sprues as a good pair of nippers can get you most of the way. But it is still good to have some sandpaper to touch things up.

Hell, I use an old pair of nippers to remove supports and overhangs from a lot of my “normal” prints.

… and now I am really curious what a 3d printed runner would be. Might do a proof of concept/experiment in onshape.

Yeah. Although I guess now I don’t really know how resin would handle those kinds of support either.

And yeah, it is pretty funny that a resin printer and curer and ventillation setup comes out cheaper than most 40k armies. But the answer probably is to scale up models (there is an artist on mastodon I keep meaning to throw money at).

Thanks

Or they heard a noise and are investigating their territory.

But yeah. I definitely get the warm fuzzies when my little goblin wanders up from her living room tree to her office tree to hang out with me.

Basiaclly all DRM models have had variations of that problem. It, again, boils down to what the check is, when they do it, and how often they do it.

For example:

• Back in the day, Splinter Cell Conviction (and a few other ubi games) actually connected to a remote server for game logic. If you were running a cracked version and a blocker (I think peerguardian is what we used? Been a minute) then you would actually notice your game just completely hang when you went through certain doors and Sam wouldn’t start talking until you turned PG off.
• Similarly, quite a few securom and even starforce games would add the DRM check as part of the fundamental gameplay loop so you were potentially checking dozens of times per SECOND. This was a rapid checksum or a value in memory but it was still very noticeable

And Denuvo is kind of the worst of all worlds since it is an activation model which, potentially, involves phoning home to a server.

To my knowledge, every single case of “Denuvo killed performance in mah gerhms!!” was either

• Complete noise. Like, less than 5% difference which could just as easily be a case of having a different tab open in your browser
• A case of a poor implementation where the checks were way too frequent

I am not aware of anything that was fundamentally denuvo itself. I would love to know more if you can point to a documented example but everything I have seen that actually has numbers ends up being one of the above.

I am not going to say that I think Denuvo is good for gaming. I fully accept the importance of DRM for week one sales (which make a huge difference to publishers) and understand that activation models are incredibly useful for that but I also think activation model DRM is fundamentally shite because it renders games unplayable in order “Why is this random ass server plugged in in this closet?”.

But I do think people overly attribute negative performance to denuvo. Implemented correctly, there are MAYBE a few checks per hour and that is system noise. The problem is that, for whatever reason, so many games end up adding the denuvo checks to critical path operations that either completely delay the loading of a new area or tank performance completely because it is checking a dozen times per minute. And that is 100% on Denuvo for not working properly with the studios they license their tools to.

But for the ones who DO implement it sanely? It is barely noticeable to the end user… from a performance standpoint.

Remember kids: Hate mother fuckers for what they actually do. Rather than going the “bitch eating crackers” route.

So… Windows?

I vaguely recall an extra worry a few min because. But my expert is that windows inevitably clobbers the boot loader eventually. So unless it is a laptop (where you are better off with a VM or wsl anyway) just give each os their own disc and use the BIOS for boot order.

I have nothing against the wolders but the admins of that instance raise more than a few eyebrows. Particularly when they rolled back specific anti hate speech policies in favor of vague common sense ones to stay ahead of the anti-woke crowd.

Kind of feels like they have been trying to take over Lemmy. Which… Could be a lot worse but still rubs me the wrong way.

And they change nothing? It still boils down to human beings being liable.

Hosting costs money. Theoretically (some) cryptocurrency can obfuscate that but… there is a reason graph problems got really popular again for a little bit.

But domain names also cost money. They also need to be registered to a person/company.

That is why a lot of torrent sites end up getting their domain stolen.

Also, all of this ignores the actual developers. But that is par for the course when it comes to discussing liability with open source projects.

Yeah. It is why I really like that Proton basically say "We will turn on you in an instant if we get a legal order. But here is what we’ll actually turn over and here is how you can minimize your vulnerability to that.

C&Ds and legal threats still go to a person who isn’t going to go to jail or have their wages garnished by Mario until the end of their days.

At best you have a single gitlab instance running in a country that is known to not cooperate with those kinds of demands. But… just ask the various torrent sites how well that works after actively pissing off an army of lawyers who don’t mind slipping a hosting company a couple hundred bucks to get the identity and address of the person paying for that.

Plenty of hosts and providers over the years have claimed to be super privacy oriented and blah blah blah. Once they get a legal request, they roll over. Because they aren’t going to prison for the customer anymore than they are going to prison for you.

Nope.

C&Ds and legal threats still go to a person who isn’t going to go to jail or have their wages garnished by Mario until the end of their days.

At best you have a single gitlab instance running in a country that is known to not cooperate with those kinds of demands. But… just ask the various torrent sites how well that works after actively pissing off an army of lawyers who don’t mind slipping a hosting company a couple hundred bucks to get the identity and address of the person paying for that.

What is the alternative?

Closed source “I swear this executable that will let you play all the games you want is legit” hosted on sourceforge or random websites that get hugged to death every day that ends in a y?

Also: it isn’t like that would stop this kind of legal action. MS/Github are barely a factor. It is the devs themselves getting the “hey, stop it or we’ll turn you into indentured servants” letters.

Part of it is the same “human speech” aspects that have plagued NLP work over the past few years. Nobody (except the poor postdoctoral bastard who is running the paper farm for their boss) actually speaks in the same way that scholarly articles are written because… that should be obvious.

This combines with the decades of work by right wing fascists to vilify intellectuals and academia. If you have ever seen (or written) a comment that boils down to “This youtuber sounds smug” or “They are presenting their opinion as fact” then you see why people prefer “natural human speech” over actual authoritatively researched and tested statements.

And… while not all pay to publish journals are trash, I feel confident saying that most are. And filtering those can be shockingly hard by design.

But the big one? Most of the owners of the various journals are REALLY fucking litigious and will go scorched earth on anyone who is using their work (because Elsevier et al own your work) to train a model.

We live in a world where everything you do or say on the internet is potentially part of someone else’s training data. Lemmy is already a shitshow for that but you can bet that “ooh, piracy board” is fair game for everyone because nobody would be dumb enough to sue about that.

If someone wants to delete a comment for ANY reason, they are more than welcome to.

Yeah. Keychrons are kind of a clusterfuck because they release like five limited runs every month. But… basically all of those are the same feature sets but might have slightly different internals based on whatever they bought in bulk the previous month. It DOES make figuring out what to buy kind of annoying but… logitech and razer have been doing that for years anyway.

Keyboard wise? At this point, prices have dropped enough that there is no real reason to go to one of the major manufacturers for anything that isn’t disposable. And basically “all” of the smaller batch mechanical keyboards are dependent on QMK or VIA to some degree which means you can customize them on any machine that can run chrome.

For the logitech price point/build quality? Unless you know why you don’t want one, you can’t go wrong with a Keychron (https://www.keychron.com/). The price and build quality isn’t “the best” but it is very much on par with the logitechs and razers of the world and they are perfect for someone who just wants “a keyboard that works” or someone who wants to learn what they ACTUALLY want out of a keyboard.

Mouse wise? There are an increasing number of “third parties” but… they basically all suck unless you are going to go crazy and mod them. And while I think the firmware matters less in these cases, there are an increasing number of qmk/via mice but… they mostly feel “cheap” or like they are just proving the viability. I have a friend with a ploopy but even he doesn’t really recommend it. So… you are still more or less suck with logitech and razer and the like for that. But hopefully as those companies lock their hardware down more it will lead to something in between “here is a cheap no name ergonomic mouse” and “here is a five hundred dollar mouse”.

Are there any good resources for qmk/via mice? Mostly can only find a few that feel like they are, at best, proof of concepts and more likely just outright predatory scams to cash in on people who want to spend money.