m3galinux

You haven't accidentally sent a prompt to a locally-running LLM using the wrong delimiters/formatting then. Noise, random parts of words, looping repetition, it all happens.

On PC the "original" (Windows 3.0 and earlier) shortcuts for cut/copy/paste were Ctrl-Del, Ctrl-Ins, and Shift-Ins respectively. I believe these were inherited from some older IBM user access standards. And for the most part they still work in Windows, especially in command lines.

Ctrl-X/C/V were originally Macintosh (and maybe Xerox before that?) and then were adopted by Windows in the 3.1 era so Word for Windows would match up better with Word for Mac. But PC not having a Command key, and the Windows key not being a thing yet, they were applied to Control instead. Which of course overlapped with the other major use of Ctrl-C as "interrupt" starting back into early UNIX machines (and adopted by DOS) but... being a graphical environment they apparently didn't consider that important at the time.

Or is it to lobby for restrictive rules on AI research that only the largest firms can hope to comply with? And "only coincidentally" ends up outlawing open source / open weight / import of overseas AI models?

Yes they do. General physical appt. 6 months ago. Doctor spent 2x more time on the wall-mounted EHR terminal slowly typing and clicking through menus than having a conversation or doing physical exams.

Just means their interface and workflow is bad and needs to be improved though, not that the doctor needs to be removed from the process altogether.

The idea is that enough users insist on non-attestation devices and platforms that governments and mega corps aren't able to require them for critical services. And loudly protest, switch to different services, etc when they DO attempt to be required. Example: already personally switched banks when they tried to require a validated Android device, and let them know directly and in reviews all over the place.

Don't fall for the trap that all of this is inevitable, you have to try and resist it first.

I've seen some great successes using HomeAssistant combined with one of these that connects your vehicle's various CANbuses (via OBD port) to Wifi/BLE. https://www.meatpi.com/products/wican

Linux intends to be left out of all this attestation garbage because it completely undermines the point of fully owning and controlling your own devices. I don't want or need to ask permission before I run a program - not from random megacorporations, and ESPECIALLY not from any of the various governments. If some third party service wants to make sure I'm not doing anything nefarious, they should do it at the border of their servers and the services they offer.

Saw the word Bullshittery. Thought someone might have written a piece on SovietWomble[1]. Actual article ended up being about economics and business, and defeatist at that. Was disappointed.

1: https://www.youtube.com/user/SovietWomble

TagConnect is too fragile. Broken off way too many of the little plastic "hooks" that go through the board.

If you developed (truly developed; including test cases and documentation and deployment strategy and change management and...) a 12 month roadmap in a couple days with Claude Code, then you seriously overestimated your time requirements in the roadmap. 2x, even 5x improvement maybe but not 300x. And you're left with a new codebase dripping in tech debt and structure nobody understands, and an agent that's likely to wipe your Prod instance and backups at some random future time.

I mostly just go with 3.5mm audio jacks; FTDI makes a prebuilt cable: https://ftdichip.com/products/ttl-232r-5v-aj/

Not so useless. In my experience LLMs are about 50/50 on making a regex that actually works and covers the cases you asked it for. Even less when you get into cases needing advanced features like backreferences and lookahead.

Can you have an "anti-"emotional connection to a brand? The iPhone for me is missing a critical feature, which is ability to run the software I choose even if it didn't come from the App Store. Which means that brand is dead to me until that situation changes.

Not particularly happy with Google for other reasons either. There are some days I want to go back to the days of Windows Mobile ROM kitchens and PalmOS. At least it wasn't such a monoculture back then.

Or SS7 attack to intercept SMS messages, no SIM swap required.

Didn't help. With AC on maximum it'd still "turn off" early. Honda claimed working as designed.

Whoever bought my old Honda Fit is asking the same question right now; I installed a button in about the same place. They'll have fun figuring that one out. Honda Fit's AC is designed a lot more for fuel efficiency than effectiveness. So I added a resistor parallel to AC temperature sensor (and the switch inline) which makes the system think it's warmer than it really is, so it cools more. But with the risk of allowing the coil to freeze up. I called it the "AC Boost Switch".

5170 (IBM AT) was a 286 design. The 5150 (IBM PC) and 5160 (IBM PC XT) were the previous 8088-based systems. Sounds like it's described as "based on" because it's heavily reverse engineered, modified to use more readily available components where possible, and then improved to 20MHz capability over the original 5170's 6/8MHz.

Lots of info on the 51xx series of machines is here: https://www.minuszerodegrees.net/index.htm

Oh that's just the start of it. I recently learned there are cheating methods that use sniffing and live modification of system RAM via PCI Express DMA transfers. As usual, physical access to hardware trumps all.

Challenge: Replace faulty RAM on an M1/M2 MacBook. Then when you discover you can't, pull the SSD and dump the data for transfer to another machine. Slight problem there too...

Wear parts shouldn't be irreplaceable. That's like having a car you have to throw away when the brake pads or tires are worn out, because they're welded in place, AND have a computer making sure they don't get replaced anyway by someone who knows how to weld.

ChatGPT when asked this question answers that its responses are probabilistic, so the responses aren't reproducible. I tested that myself, of course. Since it gave me 2 different (but overall equivalent) answers from the same prompt I'd have to agree.

I've used MythTV since something like 2004. Still great to use for Digital OTA broadcasts (and free-to-air satellite / DVB if that's available in your area). Cable DRM has unfortunately made it much less useful for recording cable broadcasts, thanks to them being allowed to encrypt all QAM signals, and now aren't required to support/provide CableCards either: https://www.nexttv.com/news/charter-cuts-off-cablecard-suppo...

I seem to remember one of the TRS-80 Basic manuals being a bit silly too. A lot more so than you'd expect from Radio Shack, anyway.

Found it: https://archive.org/details/Getting_Started_with_TRS-80_Basi...

Completing a "search" SHOULD return a 200 with an empty results set. But a search is "/api/employees?name=Bob", not /api/employees/1199. The former is an endpoint that exists but was unable to find data: it should return the correct data structure normally returned for searches, but with no results. The latter is a direct link to a particular resource, which should 404 if it doesn't exist (as if any other file at a particular path doesn't exist).

Many of your examples are changing the query string, not the resource path. That's an important designation in REST semantics. REST APIs shouldn't use query strings for accessing singular records with IDs, that's what direct resource paths are for. Anything accessed with a query string should be treated as a search.

/api/v1/employees/100 - direct path to a resource. 404 if it doesn't exist.

/some_photo.jpg - direct path to a file/resource. 404 if it doesn't exist.

/photo/100 - direct path to resource: photo ID 100. 404 if it doesn't exist.

/photo?id=100 - search for a photo with ID "100". Return a code 200 with an empty results set if that ID is not found. (With optional application-specific error feedback in the response body.) 404 if /photo doesn't exist at all.

It's not just browsers. One popular Windows-based video surveillance platform has hardware decode support for H264, but it's software only for H265; that's on both Windows clients and Android mobile. That means your H264 cameras show up instantly but H265 takes a few seconds per stream, and too many of them on one screen means you're skipping frames. At least two companies I installed these for decided to double their storage instead of taking the perceived UI performance hit.

Windows 10 Enterprise and group policy to disable telemetry? Agree it's not exactly "practical" though for non-business settings.

Any chance it was an IBM terminal? Looks like it was common for units like the 3151 to be connected to RS/6000 systems: https://commons.wikimedia.org/wiki/File:IBM_3151_terminal.jp...