There’s a joke about whitespace here somewhere, I just know it.

xscreensaver of course! Note that this is not an option on Windows—jwz hates Microsoft, and any xscreensaver port to Windows is against his wishes.

I use yabai and sketchybar for a tiling WM feel. It’s nowhere as nice as my preferred i3, but it’s ok. Unfortunately it often breaks with major OS updates, so I’m sure to hold back updating my system until yabai is working.

IIRC sshfs will work on macOS but it’s more work to install. Worth it if allowed by your IT policies and your work can benefit from it.

Vim, tmux, and the usual *NIX stuff you might want.

The coreutils are not the GNU coreutils you typically find on a Linux system, so you may find a few differences. I believe sed is slightly different, and the flags for ls must be before the filename arguments, but I’ve found it’s mostly silly stuff like that (I used zsh before using macOS, so no problem there).

Regarding DNS servers, what router do you have? Some routers have simple enough DNS capabilities — I have a MikroTik, and have it set up with DNS entries for internal services (including wildcard). Publicly accessible services just use my registrar’s DNS (namecheap — no complaints).

Oracle Free tier, amd64. Only use it because it’s free—limited bandwidth, but given I have slow upload at home it’s never really been a bottleneck. Hate to admit it given it’s Oracle, but I’ve been completely happy with it.

If I switch to a paid VPS I will probably go with racknerd (suggestions welcome though if you have thoughts).

Especially after adding in all the power draw of the automation requires…

What exactly is the incremental power draw for automation? My network gear and server (a little nuc) are sunk power costs as I self host other services.

Idling, my home uses around 100W with the fridge off. One 10W light is an additional 10% of my power budget, and I have a lot more than one light in my house. I also pay about $0.40/kWh.

I can be a bit neurotic about turning off lights when I leave a room, so Home Assistant was a nice way to free up brain space for me. A few motion sensors here and there + some simple automations, and the lights mostly handle themselves. Zigbee sensors and Zigbee or Matter-over-WiFi bulbs, so everything is local. A free VPS+WireGuard setup means I can access them remotely should I need to, with TailScale as a backup.

Cloud failures mean I can’t access remotely, but local control is unaffected—if my smart devices stop working it’s almost certainly my fault :)

Adjusted for inflation, or better yet something like median salary, would probably be more meaningful.

Seems this will preferentially screw folks in low cost of living areas. If you’re in a HCOL/VHCOL area and making ends meet, then a new car is probably affordable. If you’re making ends meet in a LCOL area, then this is likely a huge expense.

Good point, edited to add comment.

Edit: as pointed out below, these numbers are for type 1 and 2, so the population is requiring insulin is much lower than this.

Among the U.S. population overall, crude estimates for 2021 were:

• 38.4 million people of all ages—or 11.6% of the U.S. population—had diabetes.

• 38.1 million adults aged 18 years or older—or 14.7% of all U.S. adults—had diabetes (Table 1a; Table 1b).

https://www.cdc.gov/diabetes/php/data-research/index.html

Sure, the majority of folks don’t have diabetes, but come on, this affects a huge number of people, and I would bet that a vast, vast majority of people at least know someone with diabetes.

And yes, those are national whereas this is California—but it’s also about changing hearts and minds. When someone from Texas, struggling to pay for their kid’s insulin, learns about this, they might just question some things.

I’ve been really impressed with Immich, can’t recommend it enough.

I’d put substitute first, but yours sounds better :)

(I’m a big Immich fan, and I’m taking and sharing photos more than ever before, in part because Immich is awesome, self hosted, and open source [the other part is that I have kids now so I’m taking way more photos that grandparents want to see].)

On low end CPUs you can max out the CPU before maxing out network—if you want to get fancy, you can use rsync over an unencrypted remote shell like rsh, but I would only do this if the computers were directly connected to each other by one Ethernet cable.

Not sure I agree.

First, stocks tend to be highly correlated with “the market” (see financial “β”/“beta coefficient”). For example, look at, say, The Home Depot or Ford Motors. From January 2000 to January 2003 (spanning the dot com bubble) they each lost about a third of their value, yet these are not “dot com”-centric companies.

Second, the promise of AI is that it will help every company that has desk jobs. So every company has this expectation now priced into their stock, and if the bottom falls out, well…

Not an analyst/I don’t pick stocks, but just my 2¢.

If you’re running it via docker compose it’s trivial to upgrade, and there are no breaking changes. Pull, down, up, you’re done.

I think “boring” (making a hole with a revolving but) in the title is a pun on this…

Prescriptive vs. descriptive is different in colloquial language than in science.

If my data logger captures 1kB/km, how many bytes/meter is that? In every other quantitative unit I can think of, the k should cancel out; but if you want computers to be special, that’s your preference.

Metric sucks. Powers of ten are arbitrary, a fluke of biology. Powers of two are the only sensible way to make a system of measurements.

Then why are you trying to shoehorn binary into decimal? As in: why are you using decimal prefixes in the first place? Answer is probably that most people have intuition behind powers of ten. You can easily express in log2-bytes instead (a GiB is 30, a TiB is 33…etc.). Be the change you want to see!

I’m born and raised in the USA, and while imperial units can be handy for a few every day tasks, there’s a reason why the sciences in the US tend to use metric.

Regarding cooking, I’ll stick to metric, measured by weight. I can double, halve, or multiply my recipe by pi, and all I have to do is look for a different number on my scale.

Giga, Mega, Kilo…these are all SI prefixes; they differ by a factor of one thousand, which is very clean in base ten.

Ten in binary systems isn’t special, but two is; and two to the ten is very nearly a thousand, and a thousand separates the major SI prefixes. This is a neat coincidence, but should not IMHO change the meaning of the prefix.

Metric units are awesome in large part because of the use of prefixes; messing up the meaning of prefixes is a disservice to the SI/metric system. Giga == billion independent of the context. A light-year is close to 10 petameters, but no one would claim it’s exactly 10Pm.

Now, if you want to call it an “imperial gigabyte,” by all means…

Frigate is pretty good, too. I’ve only been running it for a few months but I’m very happy with it.

Though, technically that leaves you more at risk of ransomeware or something that overwrites your data.

I rsync as well, but use snapshotting on the remote drives. So, a bad rsync would suck but shouldn’t really result in data loss. Ransomware on my local+remote server would of course be very bad…

I do something similar — I have a raspberry pi and a HD, with daily rsync and snapshots (monthly retained indefinitely, weekly retained for a month, daily retained for a week). It’s at family’s house, connected to my home via WireGuard via a VPS. Tailscale (or anything really) would also work here.

It’s a great setup! Just have some watchdog reboot if it can’t talk to home (a simple cronjob with ping -c1 home.lan || reboot or similar).

Even our “slow” 35Mbps upload speed is way more than enough for incremental rsyncs of my Immich library. The initial sync was done in person, though.