The late Luiz Barroso used to say that the x86 ISA is the most successful ISA ever designed, by sheer numbers and longevity. I guess time will tell, but so far, I think he has been right. The instruction set is still out there, likely with more binaries compiled for one version or another than for any other ISA. Intel's recent attempts to remove backward compatibility for 16-bit Real Mode (X86-S, which failed) are a case in point. On Sun, Jun 14, 2026 at 9:30 AM Johan Helsingius via cctalk < cctalk(a)classiccmp.org&gt; wrote:

That's fair, though I think in the context Luiz used to make this point, it remains remarkable. He used to argue that the success of an architecture (be it software or hardware) is only loosely related to how well thought out and "elegant" it is, and is much more influenced by successfully building a long-lived ecosystem (which, arguably, the S/360 example demonstrates as well). Despite x86 being a stop-gap hodgepodge, it succeeded against much more elegant and modern designs because of extraneous factors unrelated to the design itself. He made that point in writing (to some extent, given the limited context of a one-pager) in his "Innovate Within" manifesto: https://datacenter-book.org/essays/innovate-within On Sun, Jun 14, 2026 at 11:17 AM Paul Koning &lt;paulkoning(a)comcast.net&gt; wrote:

On Sun, 14 Jun 2026, Johan Helsingius via cctalk wrote: The sole advantage of Intel's approach, is by making each one a hodgepodge of kludges from its predecessor, rather that a thorough redesign from scratch, is that less work was required to modify programs for the next one. Obviously starting over with a complete redesign makes for a much better final product, but longer delays to port software. 8086/8088 binary required VERY few changes to run up through Pentium. Micropro (later Wordstar International) ported Wordstar from 8080 to 8088 in about two weeks! It then took them much longer than that to revise the manuals. (Would using a different word processor have sped that up? :-) -- Grumpy Ol' Fred cisin(a)xenosoft.com

The 80XXX series was far from the best.  It was successful solely because the largest computer company in the world chose it. The saying back then was "You can never go wrong buying IBM" The original PC was really not a very good machine, especially not for more then $5,000.  It was three programs that made is useful. Wordstar, Visicalc (later Lotus 123) and DBASE II. IBM chose the 8088 because the 68000 was nearly equivalent to the 360 at the time and the NS32000 wasn't in silicon yet. On 6/14/2026 8:00 PM, Murray McCullough via cctalk wrote:

The uVAX-II CPU (KA630) ran at 5MHz (200ns machine cycle). On Sun, Jun 14, 2026 at 7:03 PM Jon Elson via cctalk &lt;cctalk(a)classiccmp.org&gt; wrote:

X86 is a zombie. It's dead but it won't lie down. On Tue, Jun 16, 2026 at 2:53 PM Ken Seefried via cctalk < cctalk(a)classiccmp.org&gt; wrote:

On Tue, 16 Jun 2026, ben via cctalk wrote: Is that the problem with the modern machines also?

On Tue, 16 Jun 2026, Ken Seefried wrote: Libraries are dead books are dead USA manufacturing command line mainframes serial/parallel composite video Floppy disks CRTs wired headphones manual transmission soldering (especially lead solder) DIY hand sanding hand drill non-power tools non cordless power tools shoe repair ANY repair of anything ("just buy a new one") using anything for as long as it works sewing clothesline non-cellular, landline phones music/movies and distribution other than streaming Over the air TV (except: "amazing new device lets you get TV without cable!" ("digital" rabbit ears)) non-microwave cooking cooking anything from scratch: spaghetti sauce, guacamole, chili, soup, cookies, . . . my entire world?

On Tue, 16 Jun 2026, Mike Katz via cctalk wrote: chemical photography, both B&W and color. 4x5 and 8x10 sheet film are still available, but where? Kodak has discontinued making photographic print paper. Kodak has discontinued roll film, other than 35mm and 120. Kodak has discontinued Kodachrome!! ("Don't take my kodachrome away.") Polaroid

Someone said assembler was dead, I have to disagree.  I still program in 6809, 68000, ARM, HP-41 and HP71 assembler from time to time.  I've used ARM assembler in my work recently. On 6/16/2026 8:39 PM, Ken Seefried via cctalk wrote:

I would gladly pay somebody to maintain code in each of these languages and more besides if it would get the “$x is dead!” forum trolls to sit on their hands for twenty minutes. We could start a foundation or something. ok bear.

For a languageto get CALLED "dead" does not require any of those objective criteria. It gets called dead if somebody hasn't heard anything about it in a long time. Somebody who hasn't heard anything about mainframes in years thinks that mainframes are dead. Somebody who hasn't heard anything about COBOL in years thinks that COBOL is dead. Most stuff being called "dead" is not by objective criteria; it just hasn't been talked about much lately. Among THAT person's social circle. I agree with most of your objective criteria, except lack of recent update. While it is true that most active languages get frequent updates, but that isn't requisite for not being dead. Should we say that a person is dead if they haven't been to the doctor in 10 (25, 50) years? It would call for a wellness check, but hardly an assumption of demise. Recently some AI generated facebook post recounted Cliff Stoll's famous tracking down of intruders. The same article said that Cliff died in May 2024. Cliff says that the news of his death is slightly exaggerated. -- Grumpy Ol' Fred cisin(a)xenosoft.com On Tue, 16 Jun 2026, Mike Katz via cctalk wrote: > > I'm sure there are more "dead" computer languages than there are > living ones. > > However,  what is the definition of a dead computer language.  Here > are a few potential definitions (choose one or more or add your own): > > * If the language is not running, as an interpreter or compiler, on > any currently manufactured computer.  Running on antique > computers > or simulators doesn't count. > * If no one is being paid to program in that language or maintain > code > in that language. > * If the standards for the language have not been updated in more > than > 10 (25, 50)  years. > * If the language is no longer being used in a production/commercial > environment. > * If the language is only being used in the > hobbyist/historian/antique/simulation environments. > > Here is an example question:  There is an in production add on to an > antique computer written and being supported in Forth.  The Forth > interpreter/compiler is running on a modern ARM based micro.  Even > though the target of the product is an antique computer since it is > using a current technology micro with a supported forth, I would say > that Forth is not a dead language. > > Here is a list of languages from my past, how many of them are > officially dead by one or more of the above definitions? > > APL > Forth > Lisp > Algol > Dibol > Focal > Occam > Prolog > Watfor & Watfive > Ratfor > Flap > Ralf > Teco (editor and macro language) > Pilot > DB2 > Foxbase > Any of the Hp Calculator languages (RPL, HP-41 User code) > I'm sure their are dead dialects of BASIC but BASIC is currently > supported as Visual Basic and Dartmouth Basic. > > Please update this list as to whether any of these languages are dead > (by the current definition above) or alive.  Also, please add new > definitions and languages that are dead or nearly dead. > > Note:  Dead dialects of a living language don't count.

On Tue, 2026-06-16 at 21:39 -0400, Ken Seefried via cctalk wrote: Honeywell provided a COBOL to Ada translator more than forty years ago. I don't know whether anybody actually used it.

On Tue, 2026-06-16 at 20:38 -0400, Bill Gunshannon via cctalk wrote: I remember reading about a naive young bride who asked the supermarket clerk to help her find a box of scratch because that's what her husband said his mother made everything from. Billions of lines of COBOL still in use. New standard was published in 2023. Fortran standard committees are meeting in Coventry this week putting the final touches on Fortran 2028.

COBOL was recently added to the GCC compiler suite (as was Modula-2 and Algol 68). And gfortran sees lots and lots of development; it seems that various fields use it extensively still. Weather forecasting seems to be one example. And ADA has been and remains an active part of GCC, too. Not to mention that it's the foundation of VHDL (and the open source GHDL simulator is written in Ada). As for assembly language, that certainly is still alive. A FORTH package I use extensively on the Raspberry Pico is written mostly in Forth, but the rest in assembler -- no C to be found anywhere at all. For that matter, you can't program the Pico's programmable state machine engines in anything other than assembler, for the simple reason that nothing else will serve when you only have 32 words of program memory. paul

On Tue, Jun 16, 2026 at 9:30 PM Paul Koning via cctalk < cctalk(a)classiccmp.org&gt; wrote: FFMpeg has a crazy amount of asm. This is the code that is used by virtually everything that streams video over the Internet (like YouTube, Netflix, TikTok). And from an asm perspective, that's pretending the entire embedded world doesn't exist. KJ

On Tue, 2026-06-16 at 21:11 -0500, Mike Katz via cctalk wrote: When the Cassini-Huygens mission to Saturn was being planned at JPL, they were hoping to use NS16032 or NS32032. Sandia kept promising to deliver a rad-hard version, but never did. Then, out of the blue, DoD and CIA etc suddenly declassified about a dozen quite good rad-hard processors. Their reasoning was "everybody knows about these anyway" — except apparently JPL. The JPL engineers looked at the specs, bought processors, made computers, and threw them over the wall to the programmers — who discovered the only programming environment available was Ada. They were hopping mad because they had their teeth all set to write all the flight software in C++. Then the Ada compiler started finding bugs that a C++ compiler would never have caught, and most of them would be mission-killers. As with most religions, the most recently converted adherents become the most vocal evangelists.

On Tue, 2026-06-16 at 21:29 -0400, Paul Koning via cctalk wrote: I wrote a program for the IBM 1401 that prints powers of 2, with their exponents, until the print line overflows. It fits on two cards. Yeah, 160 characters only 80 in memory all at once (the program runs entirely in the read area — 1-80). I gave it to CHM. The two-card deck is labeled "The Power of Two." They occasionally run it to entertain visitors, to demonstrate the power of a two-card program. They usually run a fancier program that also prints the last one, 128 digits, in English as well as decimal.

On Sun, 14 Jun 2026, Mike Katz via cctalk wrote: Those good software items are what made it successful. Similarly, some say that Apple's success was reliant on VsiCalc. It was only $5,000 if you bought a ridiculous bundle from IBM with seriously inflated prices for commodity hardware, and all of the software that you did NOT want. I hope that nobody other than college administrators paid THAT. Without unwanted software bundle, it was more like $2500 My first was $1320? plus about $600 for CGA and disk controller boards, and then I put in my own 4116 RAM and floppy drives (same as TRS80), and used my own CCTV composite monitor. Google Gemini says $1565, including CGA (~$300) True, IBM wanted about $3000 more for FDC and one drive, 48K (three rows of 9 4116s), CGA board, and more than $600 for a color monitor! Floppy controller board, and CGA were only available through IBM initially, but the rest was same as TRS-80, etc. at about 1/10 the price. IBM sold VisiCalc; Wordstar was independently from MicroPro; dBase II was from Ashton Tate. For a word processor, IBM tried selling "Easy Writer" by John Draper! (included in the $5,000 bundle?) Supercalc came out soon later, and was much cheaper than IBM's price for VisiCalc, Immediately after I started teaching in the college, my department chair had me come in on the weekend, and he and I put RAM, FDC, CGA, and drives into a couple dozen bare PCs. -- Grumpy Ol' Fred cisin(a)xenosoft.com

[usual long rambling tangents] Microsoft was not a big player in any of that until 1981. They had no major input in OEMs decisions of processor until 1980? Although their earliest product was a BASIC for 8080, they tried to branch out into more platforms of BASIC. They wrote the BASIC for the TRS80 (Z80, which was based on 8080), they also wrote Applesoft BASIC (6502). They did eventually make some hardware, most notably the Z80 Softcard for the ApplelII, and later, mice. Tim Paterson was developing computers based on 8086 at Seattle Computer Products. He wrote a "placeholder" OS called "Quick and Dirty OS" to use while waiting for the overdue CP/M-86. When IBM decided to "take over" home/personal computers, they looked at the 8086. They chose 8088, which I liken to an "8086 cut back, to make an 8 bit version of 8086", because the computer could be designed and built cheaply, as if it were 8 bit. [insert whole long CP/M - MS-DOS/PC-DOS history] Microsoft providing the OS created a long-term working association. The popularity of the 5150 PC created an unbelievable market for clones, imitations, similar, etc. running MS-DOS. And caused Microsoft to become what it is. In contrast, Jobs locked in the Mac, so that hardly anybody could build an imitation. IBM used the "8 bit version of 8086". But not an 8086. They used an 8086 in the Displaywriter (1980?) Later, they went to 80286 (5170 PC/AT). Around 1985-1986?, IBM came out with a "whole new line" of computers ("PS/2") In 1987? They made the PS/2 model 25 and the PS/2 model 30 which used 8086. I've heard that the PC/JX (Japanese market) had an 8086. I don't know of any other IBM devices with 8086, although 8088 and 8086 overlapped in design so much that that conflation is excusable. Intel was the biggest in processors; IBM dominated PC design; Microsoft had a virtual monopoly on Operating system. UCSD Pascal never really caught on, for several reasons. DRI's CP/M-86, "Concurrent CP/M-86", and later "Concurrent PC-DOS" and DR-DOS were welcome alternative choices, although never big enough to seriously challenge Microsoft. Microsoft trademarked "MS-DOS", IBM did NOT trademark "PC" (and it would have been absurd to try), and IBM DID NOT trademark "PC-DOS", declaring "PC/Personal Computer" and "PC-DOS to be a "desctiptive terms, not really names"! So, although it created confusion, particularly assumed affiliation with PC-DOS (IBM specific version of MS-DOS), DRI, IBM, and Microsoft did not end up in court. Over "Comcurrent PC-DOS". Another factor was that MS-DOS was based on CP/M (not copied nor plagiarized, but indisputably based on it), so IBM and Microsoft were loathe to fight with DRI over the trademark. so, Intel, IBM, and Microsoft did a lot of business together and relied on each other, but they never managed to create a cartel. So, IBM's dominance of the market made the 8088 extremely poplular, and many companies eventually branched out creating variants, some of which used 8086. A small number of 80186, then lots of 80286, 80386, 80486, and then other chip manufacturers got in on the action. Nec made the V20, V30; AMD, Cyrix, etc. After the 80486, Cyrix came out with a fancy 80486 that they called the 80586, . . . The consensus was that you "can't trademark a number" (?? How about the Oldsmobile 88, Oldsmobile 442, etc?) So Intel chose to go to a name, instead. They came up with "Pentium". I heard that that was the winner of an in-house contest; what were the losers? Even so, there were possible problems that they didn't think of: Asahi Pentax was acquired by Honeywell. Honeywell, who made some computers, could have come out with a "Pentaxium" processor (and contracted with a chip maker) with enough prior usage claim to be protected. But, it also gave the other chip makers impetus to come out with their own variation, rather trying to exactly copy Intel Some people think of Intel, IBM, and Microsoft as being a cartel, but they never got it together enough to organize anything like that. -- Grumpy Ol' Fred cisin(a)xenosoft.com On Sun, 14 Jun 2026, Murray McCullough via cctalk wrote: > What role does the duopoly, Intel/Microsoft, play in the success of the > earlier years of the 8086? When choice is restricted sometimes the best > isn't the most successful! > > Happy computing, > > Murray 🙂

Intel has shipped approximately 12 Billion CPUs since it's introduction in 1978. There have been more than 350 Billion ARM based CPUs sold since it's introduction in 1985. That means Intel is selling, on average, 250 Million per year in the 48 years since it's introduction. Whereas, ARM, on average, there have been 8.5 billion ARM based CPUs per year in the 41 years since it's introduction. ARM wins in sheer numbers and wins even more in terms of architecture refinement and structure. On 6/14/2026 12:56 PM, Fred Cisin via cctalk wrote:

Fred Cisin wrote: I have seen it claimed that Intel made the 8086 as a stop-gap until the iAPX 432 was finished.