Wait 'til he hears it was invented in New York

I don’t make Android apps because the environment is a treadmill - constantly requiring app updates to run on the newer OSs. I have not (yet) experienced the same forced obsolescence in any of the Linux flavors - apps I wrote 20 years ago still compile and run without mods, just relink to the latest libraries and go.

an AOSP (Android Open Source Project) phone already is a Linux phone

Until it has ssh access and a command line package manager ala Debian APT, RedHat YUM, Arch Pacman, Alpine APK… it’s not “Linux” enough for me.

No: PMS doesn’t count. Compared to ssh -> shell, ADB is a PITA. If ssh is too much attack surface for you, securely switch it off until you need it.

Superposition is for quantum computing, C++ does not apply.

I bought them a long time ago, and always refused to pay scalper prices. My Zero 2Ws were all $15, most of the others are Pi4 and Pi5 from before the RAM inflation.

Power is nice and stable in this arrangement, a bunch of the RasPi foundation supply bricks on a power strip mounted to the breeze-tunnel.

I’ve been a NUC fan since forever, but last year I had two NUCs that I bought in 2020 die within a month of each other, one had a power supply issue and I think the other one had a worn out SSD - it was working fine when I unplugged it, but when I plugged it back in 6 weeks later it wouldn’t boot.

Like I said, one old / low power NUC could do this, but when it dies far too much of our smart home goes dumb all at once.

Another thing LLM agents are good at is reading log files and giving you clues as to what just went sideways, when that includes a system restart it’s nice to be able to look back acrosss the logs of the past several cycles.

They (fs2ram and similar) are annoyingly, frustratingly fiddly and not worth the effort IMO… until I got Claude to do the fiddly bits for me… It isn’t just the initial setup, it’s also the additional hoops you have to go through whenever you want to change something on the “protected” partition. Fiddly is the word for it, not hard, but time and brain-power consuming.

I’ve got a box I made with a air filter and 200mm fan on one end, then it blows through a “tunnel” that has a network switch and 4 Raspberry Pis inside, one runs Frigate with an AI hat, another runs home assistant, another is running my Pi Hole, another is basically a NAS with a 2TB external USB drive, and they do a whole bunch of other miscellaneous ad-hoc tasks. The clean airflow keeps them cool just sitting there with open PCBs and the normal little heat sinks, they’re all wired to the switch so only 1 ethernet cable comes out.

On the one hand, a single Core i3 PC could probably handle ALL the tasks easily enough, but… what about when Frigate goes weird? I don’t want my whole home DNS/DHCP and home automation to go with it. Same for everything else, and if one of the microSD cards goes bad (none have yet after over 1000 days of 24 hour continuous operation except during power outages) - that’s just one thing going bad, not EVERYTHING.

I must admit the 2TB external drive is cheap junk, and those things do “wear out” - even internal SSDs go bad over time, I’ve had it happen at work more than once within about 5 years of normal use. I try to get the better microSD cards for the Pis and so far they’re not having problems for me.

One thing you can look into if you like messing around with system configurations is partitioning your microSD with “immutable” partitions where any changes are written in RAM (yes, RAM heavy Pis are even more expensive than they used to be, but it all depends on your application…) I do this for my Pis that run around in the world and get their power cut off without warning - failures are so rare in the first place it’s hard to know if all the partition work is worth the effort, but so far I’ve never had a corrupt file system with the immutable partitions carefully configured, and I have had several before I started doing that - not failed SD cards, just corrupt filesystems that needed to be re-formatted.

of course, on Lemmy you’ll get downvoted because “ai bad”

There was a time that was 100% true, I’m seeing a bit more from different points of view lately.

24 months ago I wasn’t saying “ai bad” I was saying “ai nearly useless.”

Difference being: the subsidized copper was actually installed and put to use.

The subsidized fiber is mostly vapor.

Honestly, passing on any code review is a pretty thick headed thing to do.

Do you let Claude write your commit messages?

The reason I do is because it makes the agents more capable of working with the code in the future, they put more info into the commit messages than I do and when they look back the commit messages help them navigate the change history better.

Also, if you see a commit message that just looks whacked, completely missing the actual objective - that might be a Claude hallucination.

Congrats, we all had AI free apps 2 years ago, and for the 50 years before that… Keep calm and carry on, what worked then, still works now.

Or, freak out - you just might consider having an agent like Opus 5.5 review your code and write you a report of its findings… if you don’t, others will - that’s the new landscape. If you do get “findings” in the report, I’d be curious to hear if any of them were both valid and concerning. I’m sure you won’t care about a lot of what it finds, but you might just be surprised by some things it does find.

Claude has established precedent that Claude credits itself on commits unless you direct it not to. I don’t fight it, I would rather people looking back know… Yes, I used this tool - no, it’s not “responsible” for the code, I am.

I really liked zed but didn’t like having to install it everywhere I go… If I’m going for a heavyweight install, it used to be Qt Creator, and lightweight editing in nano (because it’s always there…) I’m not loving Qt Creator so much anymore, but haven’t really replaced it yet either…

The actual form we use at work is “kick the can down the road” usually to make people feel bad about a lazy decision: “Yeah, you can kick that can down the road now, but it’s not going away until you deal with it.”

One I don’t think I’ve heard yet is “that can of worms is already open, if you kick it dwn the road it will only be a bigger mess later…”

“It’s a dog-eat-dog world out there, and I’m wearing Milk-Bone underwear.”

“People who live in glass houses can count their chickens in the rain.”

“We’ll cross that water when it’s under the bridge.”

I’ll jump on your unpopular side. With a decent IDE you should be able to point/click and decode what any operator is doing in any context. If I have two vectors, or quaternions, and I want to operate on them, why should I be forced to pass them to a standard function instead of clearly seeing what I’m doing. If you can’t be bothered to know the types of your variables in a strongly typed language… IDK.

You can beat a dead horse but you can’t make it drink.

We’ll goose that cross when we come on it.