don-code

> I hacked the program to NOP out the loop and just immediately capture.

This is a level of freedom that I wish we still had today. I'm not talking so much about the open source, "free as in speech" freedom, but more so about the "free as in simplicity" type. Nowadays it feels like I need valid TLS and a PGP signature just to host some service in my own home, the ecosystems around them having decided that the ability to tinker was (perhaps justly) a far second to security and correctness.

(For reference, I tried building and inheriting my own Terraform providers a few months ago. It's not possible to host them off plaintext HTTP, nor is it possible to skip TLS validity checks when downloading them, nor is it possible to install them without checking for a valid signature.)

I'm somewhat surprised, disk being so many orders of magnitudes slower than compute (and to a lesser extent, memory), that there's such a wide gulf. My expectation would have been that the difference would be negligible, as `tar` writes roughly double the bytes (per the article) back to the disk.

There's a machine described in here which is compatible with both the Apple ][ and CP/M systems, at a comparable price point to either one. As a naive buyer, why would I have _not_ purchased this system?

I was somewhat disappointed that there was only one "need" for holistic systems presented: the fact that correctable DIMM errors were getting eaten by underlying firmware. While I don't necessarily disagree with the speaker's premise, I'd have at least liked to see more than one example of why this is important - I can't say that I've walked away with a good idea of why that is. The fact that correctable versus uncorrectable DIMM errors were differentiable tells me that there already _is_ an interface that allows propagating both up to the OS, and that the firmware of the system just wasn't fully implementing it.

I was writing a response to this, the point of which was going to be that there's a gradient with regard to the amount of respect shown (or not shown) to people as they climb the social ladder, but at the same time, there are some forms of surveillance which by nature _can't_ be so selective in who's targeted. Now I'm not sure that's true.

My argument was going to center around common places where surveillance happens - engagement tracking at grocery stores, license plate readers on traffic light cameras, and the like. It's hard to opt out of roads or grocery stores. But then I realized that even these reflect the gradient. Maybe big box stores like Walmart are deploying these technologies - are white-collar workers actually shopping there, or are they instead going to smaller, boutique-y shops who don't install them? Are traffic light cameras going into upper-class suburban settings, or are they first being deployed to major roads and arteries?

So... I've more or less proved my initial thought wrong - this gradient does indeed persist, because people at higher rungs of the ladder are self-selecting for experiences which don't feature these kinds of surveillance (even if that is not their primary reason for doing so).

Mine was one of the later 6800-series - an 8-bit CISC CPU with enough 16-bit extensions to not feel too constrained. It's interesting that everything which made it a great learning platform (orthogonal instruction set, instructions for common but complex operations, ability to access lower bits of registers directly) make x64 a horrible instruction set to learn as your first. In that regard, x64 seems more RISC-like: it's intended for compilers to generate that, not humans.

I once worked at a greentech firm. Someone's hackathon project one year used energy usage data to predict customers' stock prices.

Will it work on the micro level of an individual home? Maybe, maybe not, but the potential for that data to be used to model behavior is definitely there.

While you're definitely not wrong, there are probably better vehicles for home energy users to meaningfully affect emissions. I make ~7 SRECs (7MWh) per year, and assuming I'm average, it'd take close to 100 homes not selling their SRECs for one year to offset an average coal plant for just one hour (http://www.energyjustice.net/coal/igcc/factsheet-long.pdf says the average coal plant puts out 667MW).

My state doesn't have time-of-use based pricing, but I do see this as one particularly effective way of shifting load (not installed capacity!) towards more emissions-friendly sources. The sun shines on home solar panels while we're not home; then absent any other incentives, we come home in the dark and do all sorts of energy-intense activities, like heating rooms and cooking dinner.

Are you still able to sell your SRECs with the system disconnected, or are you foregoing that as a cost of being offline?

I'd probably lose out on $500 per year from my system if I didn't sell them. At this point, though, it might _actually_ be asking the question of how much I value my privacy.

This. I'd not expect to become advertisement fodder for a product that I pay regularly for. The inverter, on the other hand, is something I paid for once, so for years two through 25 (the expected lifetime), I'm essentially a cost to them as they operate whatever service that it talks back to. There's an incentive there to try and recover that cost, and they can't do it by increasing my monthly bill.

Some of these features probably lead actual usage. If you'd asked in 2009 how many people wanted to adjust their air conditioner temperature remotely, command their vacuum to clean, or see video of their front door when the delivery driver drops off a package, I'd bet those numbers would be low, but we today have IoT poster children like Nest, Roomba, and Ring making those desirable features.

> the manufacturer states that not having it connected to the internet voids their warranty

I was aware that my inverter could connect to the Internet, and probably would be connected to the Internet during setup, but my full intention was to disconnect it following the install. I was pretty happy that the one that the vendor gave me was poll-able from an internal Web server, so I could scrape the metrics myself.

It turns out, none of this was to be. I found out about the Internet connectivity requirement for the warranty _after_ it was on my wall and producing power - and when you've spent five figures on a system with a 25-year warranty, unfortunately I had to suck up my pride and let it phone home to the manufacturer's cloud. It now sits on a separate VLAN for "untrusted" devices.

What's more, I also found out that the manufacturer had intentionally removed the local web server feature, instead making it totally reliant on the cloud service for metrics. It has a laughable rate limit of 300 calls per day, which gives me roughly 5-minute granularity (I could get better granularity if I made my metrics-gathering scripts poll differently overnight, but still), and I really dislike having to reach out to a vendor for data that's produced in my own home.

I still have a 90s LaserJet printer, back from when they were made like tanks. The college I went to got rid of a bunch of them in the late 2000s, so I ended up pulling the toner cartridges out of the discard pile. I haven't bought toner since 2005 (?), and I have three more cartridges to go. The cartridges also include the drum, so there's no separate drum to fail / replace.

It's a parallel printer, so it's connected over the network with a JetDirect. Not sure what the story for using it on Windows / Mac would be, but cups and sane support it like a charm.

It's also been upgraded to a whopping 10MB of memory, and has a built-in scanner that sees relatively much more frequent use (weekly scanning - I print a few times a year).

Unless the printer mechanically fails, I'm set for awhile.

I figured Gericom was an off-brand until reading it here. A friend of mine in high school had a Gericom Hummer - desktop P4 chip with some insane heat pipes and a large fan, and a single desktop DIMM slot for memory. The bottom came off in two pieces: one for the mainboard, one for the "other stuff" like the hard drive and battery. I'd never seen anything quite like it.

Functional - I'm leaning into the "analog horror" genre with some fake newsreels of things like the COVID-23 pandemic, $6 per gallon gasoline, and (of course - I live in Boston) the Red Sox losing the World Series to the Yankees.

I've found it pretty rare to _not_ find something I'm looking for on Boston Craigslist. As something of an "out there" example, I found ten old tube TVs for my Halloween decorations just a few weeks ago. Also: I got an absolutely great deal on a car through Craigslist about five years ago.

That said, I've definitely had less success on the sales side than I did in years prior, but I can't tell if that's what I'm selling being worth less used these days (desktop computers?), or if Facebook Marketplace is the de-facto place to check.

My home lab sounds pretty similar to the author's - three compute nodes running Debian, and a single storage node (single point of failure, yes!) running TrueNAS Core.

I was initially pretty apprehensive about running Kubernetes on the compute nodes, my workloads all being special snowflakes and all. I looked at off-the-shelf apps like mailu for hosting my mail system, for instance, but I have some really bizarre postfix rules that it wouldn't support. So I was worried that I'd have to maintain Dockerfiles, and a registry, and lots of config files in Git, and all that.

And guess what? I do maintain Dockerfiles, and a registry, and lots of config files in Git, but the world didn't end. Once I got over the "this is different" hump, I actually found that the ability to pull an entire node out of service (or have fan failures do it for me), more than makes up for the difference. I no longer have awkward downtime when I need to reboot (or have to worry that the machines will reboot), or little bits of storage spread across lots of machines.

Slack probably doesn't care about this, but I do find it slightly disingenuous that Signal offers a "delete message" feature, given its focus on security and not misleading the user that they're doing something that protects them. What's saying that I couldn't similarly patch Signal, and read all of my friend's deleted messages?

Tangentially: this answered a long-standing question I've had, why the long S disappeared, which also incidentally served as a great example of how the wrong learning environment can completely kill a child's curiosity.

I came across the long S in the fifth grade, if I recall, during some line of learning about colonial American history. The conversation between the teacher and me, in the middle of class, went something like this:

> Why is the letter S shaped differently?

> If you'd read the sidebar, it says that's the letter S.

> But why is it different?

> It's just an S. We need to move on.

I've never really been curious enough to go look up the Wikipedia article in the ensuing years, but that discourse really stuck with me. Luckily, now that it's been linked here, I _have_ read the Wikipedia article, and have my answer (typesetting). Even if Wikipedia didn't exist in those days, though, it would've been nice to be pointed at an encyclopedia instead of having my question dismissed.

There are a couple of tricks I've figured out.

There's room in your setup for asymmetric screens. I currently have three 24" monitors set up side by side, with a much smaller 7" screen underneath, showing Grafana dashboards. At one point, I actually had two of the 7" screens. I also briefly had a "horse blinder" style 17" monitor at the side of my setup, because I worked adjacent to the hallway and didn't like seeing people coming towards me.

Bread tabs make great cable holders. PCI slot covers make great headphone rests.

Throwing a single switch to shut the whole desk off when you're done is immensely satisfying.

In the olden days, I didn't let the two big, boxy CRT monitors on my desk run me out of space. CRTs make great tables. At one point, I strapped an FM radio, TV tuner, and a small digital clock on top of one.

Pure coincidence, perhaps not (see the other comments on this thread), but this did come across a few times as being a hack.

For instance, on the OS ABI field in ELF headers:

> Fortunately we can get around this by just setting it to the FreeBSD ABI. This is because FreeBSD is the only UNIX operating system that checks this field.

Or on loading a zero-byte segment into memory:

> However if the size in memory is zero too, then OpenBSD will refuse to run the progarm, whereas the other kernels just don't care.

"Make money" in many energy markets is a misnomer, because not all markets pay out. For instance, I have home solar in Massachusetts, where I get a billing credit at the end of the month for excess generation. I can't cash that in (so I'll never be paid), but it does give me a "free" way to use electricity in the future. (Yes, I know it's not really free, because there's no return - I effectively lose money to inflation.)

I could do something economically productive with the billing credit, like charge EVs or mine crypto, but I consider that more a knock-on effect than a moneymaker.

I was personally surprised that maintainers would "be brave" at all with a codebase as widely used as the Linux kernel - just as much as to hear Linus suggest that this sort of thing happens time and time again.

While I certainly appreciate what both the author of this project and the WINE community have put together, how far _have_ we come? My last copy of Photoshop (albeit Photoshop 7, pre-Creative Suite) in the late 2000s worked just fine under WINE, installer and all. I honestly don't know what drift has happened since then - maybe Adobe has made this purposefully difficult in some way?

For what it's worth, I use a Zenfone 8 and regularly get 2-3 days of battery life off it. I came from older, sanely-sized HTC phones, and am actually pretty happy with it - I was lucky to get a day off any of the prior three HTC phones (Sensation, M8, 10) that I had.

This actually gave me some new context on why Web technologies can be slow: fast as the runtime might have become, it spends useless cycles converting hex to ASCII, thrashing the stack, and so on. I wonder if it wouldn't be faster to encrypt the JavaScript code, and use something like the existing Widevine DRM to distribute the keys.

Coffee ended up being much more than just the beverage for me. I've never been the type of person to pay for coffee at a coffee shop each morning, which took most of the social aspect of it away. And at the time, I was buying, yes, bad coffee - specifically, the second-cheapest espresso roast that turned out to be made by Folgers, in spite of what the front of the tin said.

It was actually a local radio station (WERS Boston) that got me into a similarly local coffee roaster (Atomic Coffee). I got a bag of their coffee with a donation to the station, and I was hooked. Enough so that I now take friends to their cafe to hang out, and work with them to cater the coffee for an event I help run every year.

What's more, a number of my coworkers are coffee drinkers - sometimes, we'll bring in coffee to share; sometimes, we'll discuss the merits and demerits of such-and-such single origin blend; sometimes, there will be holy wars about the use of a moka pot versus a Chemex, or whole-bean with a burr grinder versus pre-ground. But all in good fun.

While not cooking, per se, Grocy (https://grocy.info/) is able to store recipes organized by ingredients, and do interesting things like tell you what you can cook without running to the store.

While I can't speak to either the iPhone or the Walkman, the quality of the digital-to-analog converter is significantly more important than raw horsepower or other capabilities. Where they're analog devices, there's much less to be said about the "it's just zeroes and ones" argument.

Some older devices I've owned had good-quality DACs - a Palm Treo being the best. My current ThinkPad has a downright awful DAC, to the point where music sounds _better_ over my Bluetooth headphones (Audio-Technica ATH-M50xBT) than when plugged straight in. The quality issues are even more egregious for the cheap dock I use for my work MacBook Pro, which puts out so much electrical noise over the headphone jack as to be annoying even in in Zoom calls.

> Slack prioritises whoever is first to reply and no one wants to read a 50+ message thread to figure out if they have relevant input.

My personal belief is that Slack provides a perverse incentive to reply quickly, rather than thoughtfully. There's a time and place for quick communications, but the majority of discussions I have over Slack feel that they'd be better served by thoughtful requests with equally (if not more) thoughtful responses.

E-mail was great at this: the extra overhead (if it could be called that, but I digress) of writing an e-mail encouraged information density in that e-mail - I want to convey my point in the least number of round trips possible, and likewise, have my question answered in the least number of round trips possible. Slack does away with this; round trips so easy as to be common, and in fact we have sites like nohello.com showing that the trend towards round-trips isn't helpful.