I think you won't regret it. If the container startup installs stuff, you might lock yourself out when the remote server has issues, your network has issues, or if the package you install changes due to an update.
With it baked into an image, you have reproducible results. If you build a new image and it doesn't work anymore, you can immediately switch back to the old one and figure out the issue without pressure.
So you would expect the devs to include a filterlist for known bad packages in different potential source stores that they have no influence over? How would you distribute that? Bundled with Discover, in which case the package maintainers of the different distributions have to roll out new versions with the updated list? Or as a list maintained on some server the KDE team has to provide, which gets updated by Discover automatically on startup? What if you don't condone their decision to block something? What if the list gets abused? What should companies do that want that list customized?
it doesn't matter
Hehe.
Anyway, I am also completely on Zigbee. While I like the concept of Matter over Thread, I wouldn't want to switch, because it will start with a too small network to cover a good distance and if I start replacing Zigbee devices, I effectively sabotage that network as well. So my only move would be to replace all Zigbee with Matter/Thread devices. And that seems insane. So I hope I keep getting new Zigbee devices for a while.
The idiomatic way would be to build your own image. That's exactly the strength of the layering of container images.
thisthis
"No sex!!!" .... "You don't even give us grand kids 😭😭"
That's the "pedantic" part that also gets me, but realistically it doesn't really work out anyway, because the "servings" are individual sizes. I can't really calculate exactly how much everyone is going to eat. So even if the recipe turns out the exact amount it intended, it could still be too much or not enough, simple because someone is more or less hungry than usual / expected.
I like having reproducible results, but practically with food it just doesn't happen perfectly, even if I actually measure everything perfectly (amounts, time, etc.)
I think EA was still worse. At least in my perception.
I think EA actually bought studios just to get the IP and immediately get rid of the employees. I also think they tried to milk a few of the IPs before letting it go downhill.
MS, from what I can tell, gave studios quite a lot of freedom to do what they do best. I don't think they intentionally wanted to fuck over studios, but they rather sacrificed them.
Don't get me wrong: that's still bad. But there's a difference between fucking studios over with intent and reacting badly to changed circumstances.
I am probably too pedantic for that. If the recipe says 500g of this with 250g of that, it's typically a good 2:1 mix and the packaging sizes often aligned. Now you have shit like 400g and 220g and you can't easily align them anymore. Realistically it probably won't matter even if it's not a nearly perfect 2:1 mix either. But .... I can't help myself :D
I also like that they no longer align with typically required measurements for recipes. Nothing gets me off more than having to calculate fractions for amounts of ingredients.
The thread is about snap and why it's worse than flatpak.
For me the desire to put up with the effort to cook something came, when I bought a Ninja Speedi... because the time reduces to pretty much throwing the ingredients together. Pick something to cook (potatoes, vegetables, pasta, rice,...) and throw it in the bottom. Put the divider in and put the thing to fry at the top (meat, fries, veggy pattie, whatever). A bit of water in the bottom, timer to 12 mins, temp to 180°C and hit start. 16 or so minutes later you have your meal. It starts to heat the water to produce steam and then turns on the recirculating heat for the configured time, so your food gets steamed and fried at the same time. Not having to juggle different pots and pans at the same time made cooking much more pleasant.