🇺🇦 haxadecimal 🚫👑

@lowqualityfacts@mstdn.social nor one about the reddest big dog, or about the doggest big red

@FluentInFinance@mastodon.social Nice computer ecosystem you've got here. It'd be a real shame if an AI cartel "bought" all the RAM and SSDs to drive up the prices... – RAM and SSD manufacturers

@RueNahcMohr@infosec.exchange delay line?

@CommonSparrow@mindly.social @ai6yr@m.ai6yr.org Pack my box with five dozen liquor jugs.

Writing a FORTRAN-66 parser in C++. As one does.

I just read William Gibson's _Neuromancer_ for the fourth time, or thereabouts. It's still effing amazing!

I saw a toot to which I really wanted to reply, as a smartass, "pics or it didn't happen," but in this specific case, in doing so, I would have been a major asshole. I managed to restrain myself and not reply. I try not to give in to my asshole tendencies. I'm not always successful. Sometimes my inner asshole manages to do something before my brain recognizes it for what it is.

@mralancooper@hachyderm.io Well, sure, but then some people might get something they don't "deserve". – every Republican politician Which is actually a somewhat odd position for them to take, because they generally seem to think that only the rich deserve anything.

@lritter@mastodon.gamedev.place No, no, NO! The whole modern economy is based on vendors selling you the most cheaply made and dangerous products, at the highest prices, with no liability! You wouldn't want the whole economy to collapse, would you? Would you?

@chadmccullough@mastodon.bsd.cafe I used to joke with my friends who were hardware engineers, that they had it easy. They just had to wire together the pins with the same names.

I have a non-work meeting once a week using Jitsi, and while it's not as featureful as Zoom or Teams, it's more reliable.

Four new-old-stock Amp 916557-3 ZIF PGA sockets, "Socket 7", for Pentium 75 MHz and up, Pentium MMX, AMD K5, K6, and others. One of the four, lower right in photo, is missing a portion of the top plastic, but I don't think that will affect its usage.

@flyingsaceur@ioc.exchange or maybe Plan 9, on anything that runs it

@foone@digipres.club So much this! Me: "Wait, why didn't that change to the code affect it? At all?" Me, later: "Wait, how did that even compile?" Me, later later: "I'm such a ****ing moron!"

I might eventually try to get FreeBSD 1.1.5 or a tiny Linux running on it. The first iteration will likely have only about 4 MiB of fast (10 ns) static RAM.

@fesshole@mastodon.social "You're not getting my Bud Light."

@muddle@infosec.exchange @emeb@society.oftrolls.com It amazed me the flames Scalzi ignited with his post about how being a white male in the USA was playing life on the easiest skill setting. He wasn't claiming that white males in the US "had it made," or anything like that. But change any part of that, and it clearly could only make life more challenging. A bunch of white male US crybabies objected. I'm a white male in the US, and my life isn't the best, but I fully agree with Scalzi's statement.

@darkuncle@infosec.exchange I just registered a few domains in .dev last night. It was new to me, but I haven't tracked new TLDs in a long time. I didn't know that it actually new.

@RueNahcMohr@infosec.exchange You're going to eat stuff that grew in dirt?!

@niconiconi@mk.absturztau.be Wow, news to me. I guess I see why some people might write instructions like that for non-experts, but I dislike it. I'd write instructions that talked them through the steps they need. :-(

@jwz@mastodon.social @jef@mastodon.social Agreed, #3 would only be relevant to programming really mathy stuff, like error coding/correction, which typically uses linear algebra over finite Galois fields. I understand that stuff just well enough to know that I should leave it to the experts.

It wasn't until later x86 slots/sockets that enough pins were devoted to voltage requirement signalling for a reasonably wide range to support both initial and future requirements.

@muddle@infosec.exchange @aral@mastodon.ar.al Variously translated as fruits or works.

@revk@toot.me.uk I knew GCC had the nested functions extension, and I've sometimes been tempted to use it, but I've resisted in the interest of portability. I used nested functions a fairly bit in Pascal and Ada. I am slightly curious as to whether the GCC implementation uses a display vector, or a static link in the stack frame. But not curious enough to actually try it myself. Good luck with your GCC on Mac issues. Sorry I don't have any advice on that.

@ryanc@infosec.exchange Mind if I quote that at the top of my source files?

@muddle@infosec.exchange The cake is misleading.

@VeroniqueB99@mastodon.social Yeah, right.

If the record is long (say 64KiB), and the tape is low density (800 bpi), and the EOT happens at the start of the record write, that can take 7 feet of the 14 feet of post-EOT-marker tape. Hopefully the remaining 7 feet are sufficient to prevent the tape from unspooling from the supply reel. Records that long, or longer, were uncommon back then, though the IBM hardware did not actually put a limit on record length. 7/

@niconiconi@mk.absturztau.be Marginally less awful, then. Maybe.

Now I'm changing my code to deal with these early entry-symbol items, by deferring them until the module name is found. For now, I'll still consider finding anything else before module-name to be an error. If later I discover that I'm wrong about that, more code changes will be needed. 3/

@hanshuebner@mastodon.social @me_@sueden.social Without looking, probably to interface the J11 bus to the entirely dissimilar Apple II bus, do address decode, and perhaps to support on-card memory. At least, that's what I've done for my own Apple II projects, and my own J11 projects. Otherwise it would likely take too many chips to fit on an Apple II card.

@nixCraft@mastodon.social Why should we do anything whatsoever "for the benefits of AI?" Instead, and for the benefit of human beings, let's avoid so-called "AI."

That got me as far as parsing FORLIB.REL successfully. 4/

Next hurdle: FORLIB.REL contains not one, but TWO modules named NEG. I'm not surprised that LIB80 doesn't actively prevent duplicate module names in an output library, but I'm surprised that whoever set up the FORTRAN-80 release process did that, presumably unintentionally. I haven't yet checked whether the two NEG modules are identical, but I expect that they would be. 5/

One detail about tape labels I was unsure about: In addition to header labels (start of volume, start of dataset), there were trailer labels for the corresponding end, and end of reel. I wondered what happened when a data record was being written and the EOT sensor tripped (reflective marker on the back of the tape, at least 14 feet before the physical end). Some drives I've dealt with immediately give an error and prevent further writing. 5/

The "business" computers also used 7-track, but of the opposite parity, and different character encodings. The 705 and its compatible replacement, the 7080, used tape labels, but I don't think other 700- and 7000- series machines did. 4/

@xxv@mastodon.social We like the moon!

Most systems I used either did not use tape labels, or could optionally use them with special utility software. In IBM operating systems, they were used by default, but could be disabled if desired.. Previously I'd seen some technical details about the format of labels, but didn't have a solid understanding of how they were used. Now I'm surprised that they were less common in non-IBM environments. 2/

Socket 7 and the Pentium MMX (P55C) define one pin, Vcc2DET to indicate whether the CPU uses a single supply (P54C), or separate supplies Vcc2 for core and Vcc3 for pins (P55C). Unfortunately, that doesn't tell the motherboard what the correct two voltages are, even when a VRM module is used, so some confuguration is still needed. AMD defined a second pin, Vcc2H/L# to distinguish between two Vcc2 choices, but it still wasn't really sufficient.

That behavior is incompatible with tape labeling, because the write of the last rebored has to be completed, and a trailer label written. IBM's tape controller senses the EOT and reports it to software, but does NOT interfere with writing. The write of the record completes, and the trailer label can be written. 6/

@nixCraft@mastodon.social Oh, no, not again.

One of the example .REL files I test with is the Microsoft FORTRAN-80 library, FOR LIB.REL, and it contains many cases of entry-symbol items before the module-name item. I haven't seen anything else in a module preceding the name. In all examples I've seen, one of those early entry-symbol items has the same name as the module, though I'm not going to assume that to be a requirement. 2/

An earlier tape label system existed on 7-track tape used on the 709/7090/7094 mainframes, handled by IBSYS IOCS. I'm not certain whether it was used on the earlier 704 mainframe, or the similar 7040/7044. Those, and the earlier 701, were all "scientific" computers. 3/