Now I have a little bit of time on my hands (surprisingly little actually) I was doing a bit of retro programming.

And here's something I forgot: anyone remember when hard drives were called Winchester drives?

Now we have the Internet I can look these terms up and see how those names came about...

Sophie WIlson had no idea what hassle she was going to cause me in the future by using LF+CR instead of CR+LF

https://retrocomputing.stackexchange.com/questions/29929/why-did-acorn-use-lfcr-instead-of-crlf-as-a-line-ending

Why did Acorn use LF+CR instead of CR+LF as a line ending?

The most common conventions for indicating newline nowadays are CR+LF (most famously Windows, but also the Internet standard) and bare LF (Unix and Unix-like systems). Historically there have been ...

Retrocomputing Stack Exchange

@dpiponi asks " anyone remember when hard drives were called Winchester drives?"

I do! In 1985 in my first job in industry (while I was a university student at the same time).

It had 10MB, which was wow at that time.

@MartinEscardo @dpiponi And they were the size of a washing machine! Yep, I think the first one I saw was around the same time, maybe a year or two earlier, and it could hold 5 MB. Damn, that was a lot of storage space!
@dpiponi we had them in a little room at school next to thhe tuck shop - very mysterious - only the sixth-formers were allowed in to play with the 8" floppies
@dpiponi In Hungarian that was by far the most common term for them from the 90s onwards (not sure about nowadays). At least in spoken language “winchester” – also nicknamed “vinyó” – was quickly adopted as a common noun after we left the 8-bit era behind.
@dpiponi I remember that but then my father worked in the hard drive industry so my experience may be skewed. Also, I'm pretty sure "Winchester" was an IBM brand name rather than a generic word for a hard drive.
@11011110 @dpiponi That's what I thought too, that it was a specific product with replaceable media, like a Zip drive but larger, earlier, and clumsier.
@mjd @11011110 @dpiponi Fascinating. In Germany we referred to ye olde full-size 40 MB MFM drives as Winchester drives as well. I have no idea why.
@11011110 It started as a specific product but it caught on a generic term. TBH the Winchester drives I knew as a kid may or may not have been made by IBM but we were probably less aware of brands back then.
@dpiponi I was working in a computer room at Sydney University’s CS department when I was an undergraduate. I heard this sudden loud rhythmic grinding sound. A removable Winchester disk drive attached to a mainframe had suffered a head crash and it was peeling long strips of aluminium off the platter. Quite the sight.

@dpiponi I'm increasingly wanting to go back to the 6502. It was just more fun working within those constraints and I bet it still is. And I felt (if probably incorrectly) that I had some idea how the machine worked underneath.

I wonder if I could make any useful fragment of a dependently typed language fit in 32k. Probably not but it'd be fun to try.

@edwinb Ha! My thoughts exactly But I was thinking statically typed because I don't know how to implement dependent typing. But yes, there is surely some fragment that could be built to fit in a 16K ROM (*) that could be burnt to an EPROM and run on a real Beeb :) I actually now have a nice setup (thanks to some AI) for doing ROM development on a Linux or MacOS host focused on text rather than graphics :)

(*) Hmmm...it could be split over multiple ROMs.

@dpiponi @edwinb it's not exactly the same but for the last few years i've really enjoyed working within the #uxn ecosystem [1] for similar reasons.

a stack machine with 64k of memory is powerful enough to do interesting things but with real constraints. (it's also nice that it can easily run on modern hardware, game systems, web browsers, and even some "retro" systems)

[1] https://git.sr.ht/~rabbits/uxn

~rabbits/uxn - Stack Ordinator, written in ANSI C(SDL2) - sourcehut git

@edwinb @dpiponi at one point I went pretty far down the road of trying to think how to implement a dependent type checker as a sort of forth-feeling stack-machine-y thing, deliberately so that I could imagine implementing it on the 6502 for a sigbovik paper... I got as far as dependent *first*-order types, but couldn't ever figure out a good approach for higher-order functions.
@edwinb @dpiponi Maybe compile _to_ 6502, but run the compiler on some (VM) variant with "unlimited" address space?
@BoydStephenSmithJr @dpiponi perhaps, but the fun would really start when running it on an actual BBC B. Which seems quite a challenge...