2026-08-27 00:21:01 <lisbeths    > xentrac are you online?
2026-08-27 00:22:03 <xentrac     > yes, why?
2026-08-27 00:22:37 <lisbeths    > yesterday everyone was debating about the performance of lambda calculus in haskells ghc, and you had proposed to look at the compiler output of ghc
2026-08-27 00:22:51 <lisbeths    > we in #haskell finally managed to get some compiler output and then you had responded to what I sent you
2026-08-27 00:23:07 <lisbeths    > then I sent your responses to them and they asked me to tell you to join #haskell rather than me passing messages back and forth
2026-08-27 00:23:15 <lisbeths    > are you willing to join #haskell to help settle the debate?
2026-08-27 00:31:19 <xentrac     > my level of interest is not *that* high
2026-08-27 00:31:51 <xentrac     > also I don't see it as a debate
2026-08-27 00:32:29 <lisbeths    > we just need someone to read the assembly
2026-08-27 00:32:48 <lisbeths    > finding out the truth could help future forth coders if for whatever reason they want a tiny abstract machine
2026-08-27 00:33:06 <xentrac     > I told you some things I thought, some of which I'm sure enough about that I don't really want to talk about them with someone who thinks it's a debate
2026-08-27 00:33:30 <xentrac     > hopefully they're helpful to you
2026-08-27 00:33:38 <lisbeths    > well theres no debate left really. haskells compiler only works one way. so reading the compiler output would be decisive
2026-08-27 00:34:02 <lisbeths    > it would also really help me in my own research
2026-08-27 00:34:16 <lisbeths    > if you dont feel like doing it today I could schedule another day with you in the future. I can wait months or years
2026-08-27 00:34:46 <lisbeths    > all we need to know is the latency for doing a single beta reduction in haskells runtime lamdba claculus which should be a handful of instructions
2026-08-27 00:34:48 <bjorkintosh > lisbeths: do you have to wait for anyone?
2026-08-27 00:35:06 <lisbeths    > currently I am not waiting on anyone for anything why
2026-08-27 00:36:37 <bjorkintosh > [previously in libera.Chat##forth]<lisbeths> if you dont feel like doing it today I could schedule another day with you in the future. I can wait months or years <--
2026-08-27 00:36:46 <bjorkintosh > <lisbeths> currently I am not waiting on anyone for anything why
2026-08-27 00:36:49 <bjorkintosh > inconclusive.
2026-08-27 00:37:00 <lisbeths    > I am currently not waiting on xentrac for anything xentract hasnt agreed
2026-08-27 00:38:57 <bjorkintosh > lisbeths: may I ask, what is the purpose of pursuing purity in your implementation?
2026-08-27 00:39:16 <lisbeths    > I am not worried about purity just total abstraction
2026-08-27 00:39:30 <bjorkintosh > what does that look like? I don't think I know.
2026-08-27 00:39:37 <lisbeths    > this debate started when some guys in here were making claims about really high performance of the lambda calculus that ghc outputs
2026-08-27 00:40:02 <lisbeths    > so abstract machines are very similar to the idea of a virtual machines only abstract machines are totally sandboxed
2026-08-27 00:40:32 <lisbeths    > abstract machnes know nothing of hte outside world and have no affects on the outside world except its outputs, and abstract machines take no affects from the outside world except their inputs
2026-08-27 00:40:45 <lisbeths    > and abstract machines are 100% portable between two computers as long as they both have enough ram
2026-08-27 00:41:00 <lisbeths    > so its like a special kind of virutal machine that comes with important mathematical qualities
2026-08-27 00:41:17 <lisbeths    > the consequence i san abstract machine is usually slower than native compiled code and takes up more system resources
2026-08-27 00:41:34 <lisbeths    > an example of a very small abstract machine that would fit on a microcontroller is subleq
2026-08-27 00:41:35 <xentrac     > you may want to read SPJ's paper on the STG-machine; code compiled with GHC can do several β-reductions at once
2026-08-27 00:42:10 <lisbeths    > I am still trying to determine where the current research is on optimizing away one or more beta reductions
2026-08-27 00:42:19 <lisbeths    > belive me its on my list
2026-08-27 00:43:23 <lisbeths    > normally I wouldnt care about people makin claims about some language like lambda calculus, but lambda calculus in particular is very core to my very long research project
2026-08-27 00:43:23 <bjorkintosh > SP-J's books on FP implementation are free online as well.
2026-08-27 00:43:42 <lisbeths    > im guessing thats a no from xentrac then
2026-08-27 00:43:58 <lisbeths    > whoever said ghc lambdas were extremely performant will have to agree to disagree with me then
2026-08-27 00:44:26 <bjorkintosh > who was it?
2026-08-27 00:44:37 <bjorkintosh > speak up, and free lisbeths once and for all!
2026-08-27 00:44:55 <lisbeths    > huh it was 2 nihts ago I would have to do alot of scrolling to find who it was
2026-08-27 00:44:58 <lisbeths    > two or three guys in here
2026-08-27 00:45:31 <xentrac     > it wasn't me; I said GHC-compiled code was typically 4–6× slower than C, on part with Forth, but much faster than SUBLEQ
2026-08-27 00:45:37 <xentrac     > *on par with
2026-08-27 00:45:37 <lisbeths    > it doesnt matter who said it the important thing is determining the true performance of a highly performant lambda calculus system
2026-08-27 00:45:53 <lisbeths    > yes It wasnt xentract it was like two or three dudes 2 or 3 nights ago
2026-08-27 00:46:06 <lisbeths    > xentrac merely offered that we should read the compiler output of ghc
2026-08-27 00:46:29 <xentrac     > and that the actual λ-calculus without primitives like arithmetic is going to be an order of magnitude worse than that
2026-08-27 00:46:32 <bjorkintosh > <nmz-> lisbeths: when haskell compiles, it compiles to lambda calculus, does a reduce algorithm, and then converts to c- which then compiles to your machine <--- this?
2026-08-27 00:46:57 <lisbeths    > well as soon as you add x86_64 style arithmetic you are no longer talking about lambda calculus per se
2026-08-27 00:47:17 <xentrac     > yes, and I think that's the big difference between GHC and the λ-calculus
2026-08-27 00:47:20 <lisbeths    > they are saying theyve taken away the c phase but yes
2026-08-27 00:47:35 <lisbeths    > that being said not all lambdas get reduced away at runtime in haskell
2026-08-27 00:47:43 <bjorkintosh > GHC is System-F anyway, is it not?
2026-08-27 00:47:57 <lisbeths    > there ar some lambda expressiosn that apparently survive to runtime and ive been told those lambdas pass through a lambda evaluator which is present during runtime
2026-08-27 00:48:15 <lisbeths    > system f is very similar to lambda calculus but with types. it should have a very similar performance
2026-08-27 00:48:27 <bjorkintosh > hmm. I see.
2026-08-27 00:48:48 <lisbeths    > furthermore system f exists in haskell during compile time, but afaik if you compile a lambda expression in ghc as code, it will be run as an untyped lambda at runtime
2026-08-27 00:49:18 <lisbeths    > of that last statement I made I am not totally sure
2026-08-27 00:50:09 <bjorkintosh > have you run any experiments?
2026-08-27 00:52:54 <bjorkintosh > ladies, set your lambdas.
2026-08-27 00:56:50 <lisbeths    > i have not developed the skill of benchmarking yet
2026-08-27 01:00:31 <bjorkintosh > https://archive.org/details/PerformanceAndEvaluationOfLispSystems  Performance And Evaluation Of Lisp Systems
2026-08-27 02:59:04 <crc         > lisbeths: the assembly bit you shared for that lambda expression is basically loading three registers and a tail jump
2026-08-27 02:59:27 <lisbeths    > this seems to have been already established yes
2026-08-27 02:59:45 <lisbeths    > the haskell coders have assured me the lambda expression from the source would not be optimized out, but I have my doubts
2026-08-27 03:00:25 <tabemann    > one difference between a Haskell lambda IIRC and, say, a Scheme lambda is that a Haskell lambda does not have to worry about mutation
2026-08-27 03:00:35 <tabemann    > which enables certain optimizations
2026-08-27 03:00:58 <tabemann    > whereas to make a Scheme lambda you have to do things like boxing locals because they may be modified later
2026-08-27 03:01:05 <crc         > What's the actual purpose?
2026-08-27 03:02:09 <lisbeths    > so the purpose of reading the compiler output of ghc was originally to settle the debate over how performant ghc lambdas are
2026-08-27 03:02:27 <lisbeths    > and that is becuase there was a disagreement about how performant lambda calculus can even reasonably be at runtime
2026-08-27 03:03:20 <tabemann    > one question is how does the Haskell allocator work, because a lot of FP allocators involve simply advancing a pointer and filling a few fields
2026-08-27 03:03:20 <lisbeths    > the debate about the performance of lambda calculus was started when I claimed subleq is a faster simpler abstract machine for embedded systems than lambda calculus was
2026-08-27 03:04:26 <tabemann    > the main thing is that lambda calculus implies a GC in the naive case
2026-08-27 03:04:44 <tabemann    > and the problem with GC in embedded systems is that they make timing unpredictable, not that they are slow
2026-08-27 03:05:27 <tabemann    > (you can make a reasonably fast single-threaded GC with ease, provided you give up assumptions like that addresses don't change)
2026-08-27 03:06:08 <lisbeths    > yeah thats one of the reasons why I think something like subleq or subleq+ would be more suited for mcu coding
2026-08-27 03:06:33 <tabemann    > oh, and also things like copying GC's often require you to give up half your RAM at any time as unused
2026-08-27 03:07:08 <tabemann    > the problem with SUBLEQ on the other hand is that it's like programming in brainfuck
2026-08-27 03:07:41 <tabemann    > you can easily implement brainfuck, but to *do* anything with it...
2026-08-27 03:09:10 <lisbeths    > well I mean if you are using compilation to generate subleq code you dont really have to worry about the brainfuck nature of it
2026-08-27 03:09:33 <tabemann    > you can say your magical compiler can do anything
2026-08-27 03:09:40 <lisbeths    > coding in subleq would be like coding solely in the core primatives of a forth without ever defining any new words
2026-08-27 03:09:52 <tabemann    > you can say your magical compiler will generate Conway's Game of Life-executables
2026-08-27 03:09:58 <crc         > If you're writing a compiler you certainly need to worry about it
2026-08-27 03:10:06 <tabemann    > yeah
2026-08-27 03:10:21 <lisbeths    > well no only the uahtor of the compiler would have to worry about how subleq works
2026-08-27 03:10:27 <lisbeths    > s/auhtor/author
2026-08-27 03:10:30 <tabemann    > and anyone who thinks compilers can magically do anything obviously haven't learned from the lessons of Itanium
2026-08-27 03:10:49 <lisbeths    > I have been using compilers for years and have still never learned a true asm language
2026-08-27 03:11:52 <crc         > It's worth learning my least some assembly
2026-08-27 03:11:59 <tabemann    > you need to learn how to write a compiler before you can remark on the practicality of compiling to arbitrary instruction sets
2026-08-27 03:12:04 <lisbeths    > what I am trying to express is that as forth coders what you'd likely prefer to do is treat subleq as if it were an ISA and write a forth that compiles to subleq
2026-08-27 03:12:07 <tabemann    > you can't simply handwave it away
2026-08-27 03:12:10 <lisbeths    > and then you would have a true abstract machine on forth
2026-08-27 03:12:27 <tabemann    > that feels awfully handwavy
2026-08-27 03:12:33 <lisbeths    > well I mean thats not true we can talk about the limitations of subleq without having to commit to a certain compiler design
2026-08-27 03:12:42 <lisbeths    > there is just less that we can know for sure without that commitment
2026-08-27 03:13:16 <tabemann    > but saying that "we can simply write a Forth compiler that targets SUBLEQ" without explaining *how* to do it is handwaving
2026-08-27 03:13:31 <tabemann    > it's the "assume a can opener" of computer science
2026-08-27 03:13:40 <lisbeths    > well I would imagine most proper jones-forth style forths written to subleq would look very similar to each other
2026-08-27 03:13:51 <tabemann    > it's the mistake Intel and HP made when they designed the Itanium
2026-08-27 03:14:07 <lisbeths    > you are talking about a finished project which in that instnace would require concrete designs
2026-08-27 03:14:13 <lisbeths    > I am just doing research
2026-08-27 03:15:24 <lisbeths    > the wrong time to commit to a design is when you are at the blackboard playing with different coding paradigms and trying to see what happens when you combine them
2026-08-27 03:15:30 <tabemann    > you have so many questions that need answering -- how do you maintain the stacks, how do you store scratchpad values, how do you encode control constructs, how do you encode indirect calls, how do you return from calls, and so on
2026-08-27 03:16:01 <lisbeths    > I mean that depends on your engineering style. I personally dont process the problem that way at all
2026-08-27 03:16:12 <lisbeths    > first I imagine the design of jonesforth which is written in x86 asm
2026-08-27 03:16:21 <tabemann    > targeting SUBLEQ is targeting brainfuck -- theoretically possible as in its turing complete so it can do anything, but beyond that...
2026-08-27 03:16:26 <lisbeths    > then I imagine a seasoned forth coder hand compiling that jones froth from x86 asm to subleq
2026-08-27 03:16:34 <lisbeths    > I mean thats objectively false
2026-08-27 03:16:41 <lisbeths    > they have run javascript in subleq+
2026-08-27 03:16:46 <lisbeths    > would you say javascript is like brainfuck?
2026-08-27 03:16:54 <tabemann    > what I'd do is write an emulator for a sane system in SUBLEQ, target that with my Forth, and forget SUBLEQ from that point on because it's insane
2026-08-27 03:17:11 <crc         > 
ACTION would find translating x86 assembly to subleq to be frustrating 

2026-08-27 03:17:11 <lisbeths    > a full linux distribution iwth desktop linux runs in subleq+
2026-08-27 03:17:43 <tabemann    > how likely is it that someone wrote an emulator for a normal instruction set in SUBLEQ+ and actually targeted that?
2026-08-27 03:18:02 <xentrac     > so I just found out that comp.lang.forth exists and is actually pretty active.  I'm using eternal-september.org
2026-08-27 03:18:06 <lisbeths    > I am not sure how they pulled it off you can read about it from the eternal computing initiative project
2026-08-27 03:18:34 <crc         > if I had to use a subleq machine, I'd likely implement a nicer vm and use that
2026-08-27 03:18:38 <lisbeths    > but emulating a nomral instruction set atop of subleq+ is the exact type of "compiler magic" that yo uare saying is handwavy
2026-08-27 03:18:51 <tabemann    > what I'm trying to tell you is that for these turing tarpits the only way to do anything reasonable with them is to write a VM on top of it, and use that VM as the actual target
2026-08-27 03:19:05 <lisbeths    > indeed thats the etnire discussoin that started this is what is the best "abstract" vm for embedded systems like forths
2026-08-27 03:19:25 <xentrac     > tabemann: I think subleq probably supports abstraction better than brainfuck
2026-08-27 03:19:39 <lisbeths    > calling a coding a nasty name like a turing tarpit is not an argument for why that system has no practical use
2026-08-27 03:19:53 <tabemann    > the key thing is that whenever you write an emulator you're usually giving up at least an order of magnitude of performance in the very least
2026-08-27 03:20:15 <tabemann    > the key words here are "in the very least"
2026-08-27 03:20:17 <lisbeths    > well we havent established that the subleq+ machine running linux does indeed use an emulation layer
2026-08-27 03:20:36 <lisbeths    > but yes there is always a performance tradeoff
2026-08-27 03:20:48 <lisbeths    > with vms and abstract machines you are trading performance for portability and isolation
2026-08-27 03:20:50 <tabemann    > lisbeths: unless they made gcc target SUBLEQ+...
2026-08-27 03:21:03 <lisbeths    > you may very well be correct
2026-08-27 03:21:11 <xentrac     > I think a factor of 10 is pretty normal.  you can do worse or better than that
2026-08-27 03:21:39 <lisbeths    > https://github.com/adriancable/eternal
2026-08-27 03:21:51 <lisbeths    > according to their page the way they did it seems ot be that they wrote a compiler from llvm to subleq
2026-08-27 03:22:19 <lisbeths    > or rather to subleq+
2026-08-27 03:25:17 <tabemann    > the question I have, though, is how long did Eternal take to boot to a login prompt?
2026-08-27 03:27:42 <tabemann    > about compiling from LLVM to SUBLEQ+, the obvious way to that is to define an abstract machine model on top of SUBLEQ+, and compile LLVM to that
2026-08-27 03:28:17 <tabemann    > so you're still thinking in terms of things like registers and sane instructions
2026-08-27 03:31:05 <lisbeths    > you can test it on your macihne by going here https://eternal-software.org/ and then clicking boot linuxin your browser
2026-08-27 03:31:25 <lisbeths    > took me 9 seconds
2026-08-27 03:31:39 <lisbeths    > the shell it puts me in is rather sluggish
2026-08-27 03:31:56 <lisbeths    > I think though that we arent debating about subleq being slow. subleq is ineed slow this isnt in contestion
2026-08-27 03:32:10 <lisbeths    > what is contestion is whether or not coding in subleq would be exactly like brainfuck
2026-08-27 03:32:15 <lisbeths    > an my answer is only for the compiler writer
2026-08-27 03:33:10 <tabemann    > and my point is that saying "only for the compiler writer" trivializes the act of writing a compiler
2026-08-27 03:34:10 <tabemann    > because you have two choices on a difficult platform -- either have a hard time implementing a reasonable compiler, *or* generate slow code by actually targeting an abstract machine model implemented on top of the substrate
2026-08-27 03:44:17 <tabemann    > and the difficulty of writing a compiler than generates *performant* code can be important -- that's part of what killed the Itanium
2026-08-27 03:47:33 <xentrac     > amusing that https://eternal-software.org/ and https://www.eternal-september.org/ showed up here within a few minutes of each other.  Not the same!
2026-08-27 03:49:06 <tabemann    > the problem with SUBLEQ to make software eternal is you need a fast machine to emulate it on to make it work
2026-08-27 03:49:31 <tabemann    > if you really want software to be eternal, I'd say target something like the 6502
2026-08-27 03:49:58 <xentrac     > writing a usable emulator for a 6502 is a great deal more difficult than implementing SUBLEQ
2026-08-27 03:50:03 <tabemann    > where even if nukes fry all electronics, we could re-create the electronics for it from little within a few decades
2026-08-27 03:50:22 <tabemann    > xentrac: I mean in hardware
2026-08-27 03:50:32 <xentrac     > implementing SUBLEQ in hardware is also very easy
2026-08-27 03:50:53 <bjorkintosh > writing _anything_ these days is just a prompt away.
2026-08-27 03:50:55 <tabemann    > xentrac: but can you do anything with it efficiently?
2026-08-27 03:51:26 <xentrac     > not *efficiently*
2026-08-27 03:51:32 <xentrac     > but possibly *fast*
2026-08-27 03:51:35 <tabemann    > bjorkintosh: I'm talking about if we nuke each other to ashes and have to recreate electronics from scratch
2026-08-27 03:51:46 <bjorkintosh > we'd have bigger issues.
2026-08-27 03:51:51 <bjorkintosh > food :-)
2026-08-27 03:51:57 <xentrac     > I'm not sure that's a bigger issue
2026-08-27 03:52:03 <bjorkintosh > *blink*
2026-08-27 03:52:27 <bjorkintosh > feeding ourselves after a nuclear disaster would not be a priority??
2026-08-27 03:52:35 <xentrac     > a priority, yes, but I mean people can eat a pretty large fraction of the life forms
2026-08-27 03:52:36 <tabemann    > because that's the main scenario I see where modern electronics as a whole would actually be lost
2026-08-27 03:52:42 <bjorkintosh > we'd be spending our time and resources rebuilding chineseum instead?
2026-08-27 03:52:53 <xentrac     > one certainly hopes so
2026-08-27 03:53:02 <bjorkintosh > most things WOULD be too contaminated!
2026-08-27 03:53:08 <tabemann    > I'm talking on a scale of decades *after* the nuclear apocalypse
2026-08-27 03:53:15 <xentrac     > maybe not where there were surviving humans
2026-08-27 03:53:15 <bjorkintosh > tabemann: oh in that case.
2026-08-27 03:53:49 <xentrac     > I mean the people who didn't have enough to eat would die, so their food would stop being an issue
2026-08-27 03:54:08 <bjorkintosh > xentrac: please watch this if you haven't: https://www.imdb.com/title/tt37041021/ Emergência Radioativa
2026-08-27 03:54:54 <xentrac     > no thanks
2026-08-27 03:55:27 <xentrac     > I prefer to get my information about radiological disasters from reliable sources
2026-08-27 03:56:07 <bjorkintosh > as you like.
2026-08-27 03:56:25 <tabemann    > the question of whether we'd die from radiation really depends on whether people decide to use cobalt bombs
2026-08-27 03:57:03 <bjorkintosh > tabemann: please watch this if you haven't: https://www.imdb.com/title/tt37041021/ Radioactive Emergency.
2026-08-27 03:57:10 <tabemann    > if they don't use cobalt bombs, then the main threat is either dying directly from nuclear blasts or starving to death or all of technological civilization breaking down due to EMP's
2026-08-27 03:57:57 <xentrac     > dunno
2026-08-27 03:58:15 <tabemann    > people will die from radioactivity, e.g. from radioiodine-induced cancer, but those won't be a majority of deaths
2026-08-27 03:58:49 <xentrac     > it might depend a lot on the exact circumstances
2026-08-27 03:59:06 <tabemann    > the big worry would be starvation
2026-08-27 03:59:08 <bjorkintosh > as long as the survivors knew how to repair electronics and create electronic devices ...
2026-08-27 03:59:29 <bjorkintosh > 'cause that knowledge does not naturally arise.
2026-08-27 03:59:36 <tabemann    > bjorkintosh: I'm talking about everything being fried, and having to recreate things from the tech of the early 1900's
2026-08-27 03:59:44 <bjorkintosh > WAT?
2026-08-27 03:59:58 <bjorkintosh > tabemann: put it in sci fi form let's see. I cannot imagine it.
2026-08-27 04:00:21 <bjorkintosh > you mean, having forgotten everything including transistors and crooke's tube and all that??
2026-08-27 04:00:24 <tabemann    > of course, any surviving paper records of how things like transistors work mean that we wouldn't have to wait as long as it took for transistors to be developed the first time
2026-08-27 04:01:10 <tabemann    > but we'd still have to do things like essentially recreate lithography machines from scratch
2026-08-27 04:01:32 <bjorkintosh > I figured mechanical skills would take priority over electronic skills.
2026-08-27 04:02:07 <tabemann    > the problem is most mechanical things rely on electronics today, and all electronics beyond the most basic analogue electronics would be lost
2026-08-27 04:02:22 <bjorkintosh > I say if all this were obliterated, mankind wouldn't attempt to recreate it again from scratch.
2026-08-27 04:02:37 <bjorkintosh > we don't make pyramids anymore, for instance.
2026-08-27 04:02:42 <bjorkintosh > we build sky scrapers instead.
2026-08-27 04:03:14 <bjorkintosh > tabemann: not ... really. start from blacksmithing.
2026-08-27 04:03:31 <bjorkintosh > it's just fire, hammers, tongs, and a good anvil.
2026-08-27 04:03:39 <tabemann    > no, but we'd be set back to the tech of the 1950's at the latest
2026-08-27 04:04:14 <bjorkintosh > you're highly optimistic. ever been to a third world country like out in the country, Arkansas?
2026-08-27 04:04:27 <vdupras     > also, the first time around, we had access to cheap energy and a bountiful holocene, which we won't have the second time around, adding to the difficulty level
2026-08-27 04:05:01 <bjorkintosh > modern civilization irreversibly consumes material
2026-08-27 04:05:13 <bjorkintosh > including fresh water, it turns out.
2026-08-27 04:05:26 <bjorkintosh > vdupras understands.
2026-08-27 04:05:48 <tabemann    > the key thing is loss of knowledge, because most of our modern records are electronic, and anything electronic would be lost due to EMP's
2026-08-27 04:05:48 <bjorkintosh > we'd mostly be scavenging. as a matter of fact, there was a scene from samsara...
2026-08-27 04:06:09 <vdupras     > not only knowledge, knowhow
2026-08-27 04:06:17 <tabemann    > that too
2026-08-27 04:07:10 <tabemann    > we would be limited to what is written on paper
2026-08-27 04:07:20 <vdupras     > wasn't there an anecdote somewhere that we discovered that from the (plentiful) documentation from one of the Appolo mission, reproducing it was actually impossible because of some knowhow that had gone with the retired people?
2026-08-27 04:07:40 <vdupras     > *Apollo
2026-08-27 04:07:40 <tabemann    > not surprised
2026-08-27 04:08:10 <bjorkintosh > vdupras: yes. if they'd waited 2 more decades they would have had to create it from scratch, instead of from experience.
2026-08-27 04:08:49 <vdupras     > We've thrown the ladder leading to where we are
2026-08-27 04:10:25 <tabemann    > we'd have to recreate computer technology from discrete transistors and diodes and like simply because lithography would almost certainly be lost
2026-08-27 04:10:46 <tabemann    > and then we'd have to bootstrap modern computers again on top of that, a process that'd take decades
2026-08-27 04:11:04 <bjorkintosh > Or.
2026-08-27 04:11:10 <bjorkintosh > we could live simple, nomadic lives.
2026-08-27 04:11:29 <vdupras     > We're lucky that the ideas around Forth are so compact and clear. It fits well on a few sheets of paper
2026-08-27 04:12:00 <tabemann    > if anything from modern computing survived, Forth would probably be one of the few things to survive
2026-08-27 04:12:11 <vdupras     > bjorkintosh: yeah, but controlling electron is a massive, massive enabler
2026-08-27 04:12:20 <bjorkintosh > lisp first, tabemann. it's origins are clearly documented.
2026-08-27 04:12:45 <bjorkintosh > vdupras: heh yes indeed. but knowledge of such is very unevenly spread. think about it for a minute.
2026-08-27 04:12:58 <tabemann    > bjorkintosh: the problem is that writing a Lisp compiler is much more esoteric than writing a Forth compiler
2026-08-27 04:13:06 <bjorkintosh > nah...
2026-08-27 04:13:17 <tabemann    > everyone and their dog writes a Forth compiler. few people write Lisp compilers.
2026-08-27 04:13:18 <bjorkintosh > ask lisbeths, they do it on a regular basis, if not every hour.
2026-08-27 04:13:38 <vdupras     > if we're thinking about computer from discrete component, the lightest system will win and Forth is lighter
2026-08-27 04:13:57 <tabemann    > Lisp early on required heavy-duty systems like PDP-10's
2026-08-27 04:14:07 <bjorkintosh > IBM 704.
2026-08-27 04:14:21 <bjorkintosh > that was the first machine it was implemented on. before then, it solely existed on paper.
2026-08-27 04:14:22 <tabemann    > Forth can run on just about anything that supports more than raw machine code
2026-08-27 04:14:49 <tabemann    > bjorkintosh: in theory. early Lisp's were too limited to be useful before about the PDP-6 came along
2026-08-27 04:15:09 <bjorkintosh > that's a bit of a stretch there, tabemann.
2026-08-27 04:16:28 <bjorkintosh > the time of the early lisps corresponded with a time of VERY few computers in the entire world, in the first place.
2026-08-27 04:16:43 <tabemann    > also, there's far, far more documentation out there today of how to write a Forth compiler than how to write a Lisp compiler, so there's more chance of it surviving a nuclear apocalypse
2026-08-27 04:17:39 <bjorkintosh > that would be true if there weren't so many lisp books published over the years, showing exactly how to implement lisp, vs Forth.
2026-08-27 04:18:04 <bjorkintosh > the challenge would be garbage collection.
2026-08-27 04:18:35 <bjorkintosh > besides that, I'd say they're on par for implementation documents.
2026-08-27 04:18:55 <tabemann    > simple garbage collectors are actually trivial to write. the problem is making them performant for realtime operation
2026-08-27 04:19:19 <vdupras     > With come luck, the "theme" around Collapse OS makes it more likely to be shrink-wrapped into a "in case of emergency, open this" label, making Forth more discoverable :)
2026-08-27 04:19:19 <bjorkintosh > at that point in your nuclear holocaust, performance would not be a priority.
2026-08-27 04:19:28 <bjorkintosh > just turning on a light bulb would be a damned miracle.
2026-08-27 04:19:50 <tabemann    > I'm talking on a scale of decades later
2026-08-27 04:20:08 <tabemann    > we'd be starting off with recreating the lightbulb, that would take about a decade
2026-08-27 04:20:16 <bjorkintosh > so, half a century AFTER the event?
2026-08-27 04:21:07 <tabemann    > the time frame would be shortened relative to the first time we developed these things, as there would be people who still remembered how things basically worked and some surviving paper records
2026-08-27 04:21:46 <tabemann    > but we'd first probably be focusing on recreating things like generators, electric motors, lightbulbs, and radiotelegraphy
2026-08-27 04:22:39 <vdupras     > let's not forget scavenging. It's not because we've lost the ability to re-create things that we can't scavenge for tools helping us too bootstrap ourselves. This might include computers
2026-08-27 04:23:56 <tabemann    > the problem is that all microelectronics would be lost
2026-08-27 04:24:06 <tabemann    > they'd be fried by the EMP's
2026-08-27 04:24:31 <tabemann    > the only surviving electronics would be non-micro-electronics
2026-08-27 04:24:35 <vdupras     > But would they really all be fried?
2026-08-27 04:25:20 <tabemann    > a typical scenario in nuclear war would be the use of high altitude blasts to generate EMP's to destroy the enemy's electronics en masse
2026-08-27 04:25:39 <vdupras     > Just today, I had a friend tell me how it seemed to them that their house act as a kind of faraday cage because the cell signal is really bad
2026-08-27 04:25:57 <vdupras     > maybe that in such house, electronics wouldn't be fried
2026-08-27 04:26:24 <tabemann    > the thing is that EMP's are much higher-intensity than cell phone signals
2026-08-27 04:26:30 <vdupras     > maybe some people like to put electronics in potato chips bags
2026-08-27 04:27:15 <vdupras     > Yes, it's stronger, but also, cell signal doesn't destroy electronics
2026-08-27 04:27:44 <tabemann    > okay, I give that there may be some surviving microelectronics, but the key thing is knowledge of how to create more microelectronics
2026-08-27 04:27:57 <vdupras     > What I'm saying is that maybe that this faraday cage effect would be enough to be the difference between "fried" and "not fried" for some kind of electronics
2026-08-27 04:28:42 <vdupras     > yes, it is, but it's a much, much better headstart to have a few than to have none
2026-08-27 04:30:46 <vdupras     > like it's a much better headstart to have a single knife than to have none, survival-wise, even if ultimately, what's important is the ability to create new ones
2026-08-27 04:32:13 <tabemann    > the key thing here is restarting civilization
2026-08-27 04:32:22 <tabemann    > beyond merely surviving
2026-08-27 04:33:02 <tabemann    > a large percentage of the human population is likely to starve to death, but those who don't are left with recreating modern life from very little
2026-08-27 04:33:42 <vdupras     > I hope they won't try to recreate what we had
2026-08-27 04:35:58 <tabemann    > well there are certain things that they could do differently than us, such as focusing more on recreating solar and wind power rather than redeveloping fossil fuel power
2026-08-27 04:47:30 <lisbeths    > so theyve got backups of github that are deep underground in georedundant secure caves safe from radiation and sealed against emp and emi
2026-08-27 04:47:40 <lisbeths    > similarly theyve got certain military spec chips backed up
2026-08-27 04:48:12 <lisbeths    > you would have to cursh up and deofrm the mantle of the earth severely to get at all of those underground backups
2026-08-27 04:48:29 <lisbeths    > and still somebody will probably find an old broken iphone in the sand somewhere that has enough of a processor to reverse engineer the technology
2026-08-27 04:56:58 <bjorkintosh > 'military spec' rarely ever means high quality.
2026-08-27 04:57:53 <lisbeths    > true
2026-08-27 04:58:11 <lisbeths    > I mean to say that they have chosen chips that would be useful for rebuildling the tech infrastructure
2026-08-27 04:58:35 <lisbeths    > the military has spent alot of money on doomsday prep
2026-08-27 05:01:15 <bjorkintosh > the military is DIRECTLY responsible for creating doomsday.
2026-08-27 05:01:33 <bjorkintosh > they have no other use.
2026-08-27 05:01:58 <lisbeths    > I mean it depends on what faction/compartment of hte military you are talking about
2026-08-27 05:02:15 <lisbeths    > the airforce parachuting/rescue service certainly isnt tasked with doomsday they strictly save lives
2026-08-27 05:02:32 <bjorkintosh > I think any salvation from a doomsday we have, will depend on advancing nanotech today.
2026-08-27 05:02:49 <bjorkintosh > the components would be small and easily stored for later use.
2026-08-27 05:03:07 <lisbeths    > My suspicion is cellular life on earth is floating though space and latching onto astroids and landing on to planets like earth
2026-08-27 05:03:17 <lisbeths    > and my suspiciouns is that cellular life is the result of nanotech from previous worlds
2026-08-27 05:03:35 <bjorkintosh > we're instead wasting money on building energy and fresh water guzzling datacenters so that we can ask an LLM how to spell strawberry.
2026-08-27 05:04:03 <lisbeths    > well the military has their own internal datacenters they are not directly responsible for these ai datacenters
2026-08-27 05:04:10 <lisbeths    > the ai datacenters are partially a money laundering scheme
2026-08-27 05:04:44 <bjorkintosh > that's okay. they take the lion's share of our annual budget.
2026-08-27 05:05:14 <lisbeths    > and purchased global dominance with it
2026-08-27 05:05:19 <bjorkintosh > for what?
2026-08-27 05:05:19 <lisbeths    > I would say that is a priceless trade
2026-08-27 05:05:28 <bjorkintosh > a bunch of pedos to mock us?
2026-08-27 05:05:42 <lisbeths    > there are factions of the military attempting ot mass arrest the pedos
2026-08-27 05:06:06 <bjorkintosh > the entire establishment is corrupt to the damned core, and they have no excuse for _anything_. nothing useful comes from having such absolute power.
2026-08-27 05:06:20 <lisbeths    > there ar ea few primary reasons for that
2026-08-27 05:06:20 <bjorkintosh > lisbeths: their commander in chief remains at large.
2026-08-27 05:06:33 <lisbeths    > first of all jeffry epsteins job was to get blackmail material on every single powerful person on the planet and he was very good at it
2026-08-27 05:06:46 <bjorkintosh > hasn't made a single difference.
2026-08-27 05:06:52 <lisbeths    > second of all the legacy system of putting royalty into power is still somewhat in place when you trace the lineages of our worl dleaders
2026-08-27 05:07:00 <bjorkintosh > but I digress. much of this will just piss me off.
2026-08-27 05:07:16 <bjorkintosh > oh look, a combinator!
2026-08-27 05:07:29 <lisbeths    > i agree ther are alot of negative things about the military I am not trying to discount that. Im merely pointing out that there are some good parts too
2026-08-27 05:07:43 <lisbeths    > one of those good parts is that the computer technology is safe from doomsday up to a certain degree
2026-08-27 05:08:25 <lisbeths    > on the sites that have the physical backups of github there is a plaquard that says "open source won"
2026-08-27 05:08:55 <bjorkintosh > really?
2026-08-27 05:09:10 <bjorkintosh > isn't github owned by an evil corporation in Seattle?
2026-08-27 05:09:22 <lisbeths    > yes but the code isnt licensed to them
2026-08-27 05:33:45 <xentrac     > tabemann: starvation would be a huge worry for individuals, but it's not a worry for the survival of the species or civilization
2026-08-27 05:34:06 <xentrac     > 02:01 < bjorkintosh> I figured mechanical skills would take priority over electronic skills.
2026-08-27 05:34:21 <xentrac     > in most societies, at most times, mechanical skills have not been highly valued
2026-08-27 05:34:58 <xentrac     > 02:05 < bjorkintosh> modern civilization irreversibly consumes material
2026-08-27 05:34:58 <xentrac     > 02:05 < bjorkintosh> including fresh water, it turns out.
2026-08-27 05:35:25 <xentrac     > this is nonsense; freshwater loss is the farthest possible thing from irreversible.  it literally falls from the sky
2026-08-27 05:35:39 <xentrac     > it's unlikely that EMPs would destroy all electronics
2026-08-27 05:35:51 <lisbeths    > there are emp and emi resistant technologies such as aluminum foil
2026-08-27 05:36:03 <xentrac     > 02:11 < bjorkintosh> we could live simple, nomadic lives.
2026-08-27 05:36:14 <xentrac     > nomads live very complex lives; it's the only way they can survive!
2026-08-27 05:36:40 <xentrac     > 02:13 < tabemann> Lisp early on required heavy-duty systems like PDP-10's
2026-08-27 05:36:51 <xentrac     > PDP-10s didn't exist until Lisp was 15 years old
2026-08-27 05:38:06 <xentrac     > the IBM 704 had 4096 words of RAM and executed 12000 floating-point addition instructions per second
2026-08-27 05:38:32 <xentrac     > which is about what you need for a minimal interactive Forth, too
2026-08-27 05:42:22 <bjorkintosh > all the best in the post apocalypse, xentrac.
2026-08-27 05:42:48 <lisbeths    > heres the best apocalypse food scenario
2026-08-27 05:42:53 <lisbeths    > in rome they had something called the cura anona
2026-08-27 05:43:02 <bjorkintosh > xentrac> in most societies, at most times, mechanical skills have not been highly valued <-- good luck making anything.
2026-08-27 05:43:08 <lisbeths    > for every roman "citizen", they would get a stipend of raw pork, grain, and olive oil
2026-08-27 05:43:20 <bjorkintosh > their salary, perhaps?
2026-08-27 05:43:28 <lisbeths    > so for a modern day versoin we can feed all of our scrap food to pigs, and then take the pork and make it into canned pork
2026-08-27 05:43:43 <lisbeths    > and then we have a shelf stable version of cura anona that will las thundreds of years on the shelf
2026-08-27 05:44:15 <bjorkintosh > lisbeths: not everyone eats pork, or meat, for that matter.
2026-08-27 05:45:13 <bjorkintosh > but I get what you're saying.
2026-08-27 05:46:42 <lisbeths    > well there are only two animals that can eat almost any scraps you give them
2026-08-27 05:46:44 <lisbeths    > pigs and chickens
2026-08-27 05:46:59 <lisbeths    > if you really wanted to make it more kosher I would recommend switching the pork for powdered eggs
2026-08-27 05:47:05 <lisbeths    > thats what the navy does apparently is stockpiles powdered eggs
2026-08-27 05:47:15 <bjorkintosh > talking about vegans.
2026-08-27 05:47:29 <bjorkintosh > & vegetarians.
2026-08-27 05:48:03 <lisbeths    > if you want to go that route you probably want rajma
2026-08-27 05:48:15 <lisbeths    > specifically the kind that they cook in batches for like 10 thousnad person crowds in india
2026-08-27 05:48:32 <lisbeths    > the reason why you would still want to have pork in your stockpiles though is pigs can eat almost any scraps
2026-08-27 05:48:41 <lisbeths    > and 50% of the food tha tgets produced in america gets wasted
2026-08-27 05:49:01 <lisbeths    > honestly youd just want ot stock pile a big variety of anything and everything that is extremely shelf stable
2026-08-27 05:49:07 <lisbeths    > and the military solved that like in the 50s
2026-08-27 05:49:39 <bjorkintosh > if I had to do it, I'd do beans, and lentils, and cowpeas and other such.
2026-08-27 05:50:27 <lisbeths    > I recommend looking up sihk ramja temples
2026-08-27 05:50:41 <lisbeths    > its a very good recipe that is halal and the only animal product is clarified butter
2026-08-27 05:50:48 <lisbeths    > you can eat it every day and not get sick
2026-08-27 05:51:07 <bjorkintosh > I think there's one in Dallas.
2026-08-27 05:51:11 <bjorkintosh > very tasty food.
2026-08-27 05:51:37 <bjorkintosh > wait. no that's a Krishna temple.
2026-08-27 05:51:40 <bjorkintosh > n/m.
2026-08-27 05:57:31 <lisbeths    > not exactly
2026-08-27 05:57:41 <lisbeths    > so in hundism people can merge with god when they die (in a nutshell)
2026-08-27 05:57:49 <lisbeths    > and krsna is one of the gurus who were said to have done this
2026-08-27 05:57:59 <lisbeths    > the sihks look at krishna as a historical figure who existed
2026-08-27 05:58:12 <lisbeths    > but they dont exactly worship krishna as their primary god or messiah
2026-08-27 06:14:41 <xentrac     > bjorkintosh: yes, indeed, we live in a society that is very much better at making things than most societies at most times
2026-08-27 06:15:13 <xentrac     > I'm not familiar with rajma
2026-08-27 06:16:04 <xentrac     > is rajma just kidney beans?
2026-08-27 06:16:27 <lisbeths    > they cook a differnet beans recipe every single day, as well as flatbread, and sweet rice pudding for desert
2026-08-27 06:16:36 <xentrac     > what's rajma?
2026-08-27 06:16:39 <lisbeths    > they cook it in giant vats that might feed 1000-2000 people per vat
2026-08-27 06:16:47 <lisbeths    > I think rajma translates to a specific kind of beans and rice
2026-08-27 06:16:54 <xentrac     > the temple in Dallas with very tasty food is a Hare Krishna temple, which is Vaishnavi
2026-08-27 06:16:58 <lisbeths    > it doesnt directly translate to "beans and rice" but thats kind of the idea
2026-08-27 06:17:10 <lisbeths    > yeah so sihks believe in krishna
2026-08-27 06:17:27 <lisbeths    > sihks are a offshoot of hinduism similarly to how southern baptist stems from christianity
2026-08-27 06:17:39 <xentrac     > rajma is kidney beans: https://en.wiktionary.org/wiki/%E0%A4%B0%E0%A4%BE%E0%A4%9C%E0%A4%AE%E0%A4%BE#Hindi
2026-08-27 06:18:00 <xentrac     > I think it's close to how Mormons or Muslims stem from Christianity
2026-08-27 06:18:21 <lisbeths    > muslims stem from judaism not christianity
2026-08-27 06:18:32 <xentrac     > they don't think so
2026-08-27 06:18:49 <lisbeths    > the concept of the messiah is a jewish concept
2026-08-27 06:18:53 <xentrac     > yes
2026-08-27 06:18:58 <xentrac     > and they reject it
2026-08-27 06:19:12 <xentrac     > historically speaking, though, they were much more strongly influenced by Judaism than by Christianity
2026-08-27 06:19:23 <lisbeths    > in islam muhammed is the messiah
2026-08-27 06:19:27 <xentrac     > no
2026-08-27 06:19:41 <xentrac     > he is the seal of the prophets
2026-08-27 06:19:44 <xentrac     > that is not the same thing
2026-08-27 06:19:48 <lisbeths    > to be fair I know more about hundiusm htan I do about islam or the judaic faiths
2026-08-27 06:20:44 <xentrac     > Islam calls Jesus "al-Masih", which means "the Messiah"
2026-08-27 06:21:05 <lisbeths    > you may be correct I dont study islam
2026-08-27 06:22:10 <xentrac     > anyway, the relationship between Sikhism and Hinduism is more like the relationship between Islam and Christianity than the relationship between Southern Baptism and Christianity
2026-08-27 06:22:33 <xentrac     > although Hinduism as a polytheistic faith is a bit of a retcon
2026-08-27 06:23:21 <lisbeths    > are you familiar with hindu ascention?
2026-08-27 06:25:27 <xentrac     > that's not a word
2026-08-27 06:25:47 <lisbeths    > ok?
2026-08-27 06:26:00 <lisbeths    > if there was not a word for fish would fish still exist?
2026-08-27 06:26:54 <xentrac     > we're debating ontology now?
2026-08-27 06:28:39 <xentrac     > there's probably a word for the concept you meant but "ascention" isn't it: https://en.wiktionary.org/w/index.php?title=ascention
2026-08-27 06:32:41 <lisbeths    > oh sorry I missspelled it
2026-08-27 06:32:46 <lisbeths    > it means the process of ascending
2026-08-27 06:32:47 <xentrac     > what word did you mean?
2026-08-27 06:32:52 <xentrac     > ah, ascension
2026-08-27 06:33:02 <lisbeths    > are you familiar with the process of hindu ascension?
2026-08-27 06:33:22 <xentrac     > do you mean moksha?
2026-08-27 06:33:45 <lisbeths    > that is one form of it
2026-08-27 06:39:35 <xentrac     > go on
2026-08-27 06:45:06 <lisbeths    > I was just curious if you knew about it
2026-08-27 06:45:26 <lisbeths    > I think explaining why I was curious would be too far off topic
2026-08-27 08:57:57 <veltas      > xentrac: That's basically what Christ means right?  Comes from 'anointed' i.e. 'chosen'.  I think Messiah is Hebrew for annointed one.
2026-08-27 09:17:13 <xentrac     > so it seems: https://en.wiktionary.org/wiki/Christ#Etymology
2026-08-27 09:24:57 <veltas      > Etymology is really cool, I like when we get similar words appearing in stuff like Hindu and English
2026-08-27 09:26:12 <veltas      > Like 'charisma' which is a favourite of the kids today shortened as 'rizz', it is similar in English and Hindu although they have quite different usages but the same root I think
2026-08-27 09:26:36 <veltas      > Hindi* I mean
2026-08-27 09:27:33 <xentrac     > funniest version of this is a friend of mine whose surname is Datta; she's American but her parents were Bengali
2026-08-27 09:27:42 <xentrac     > so her family home is called the Dattabase
2026-08-27 09:27:49 <veltas      > lol
2026-08-27 09:28:00 <veltas      > It's like karishme in Hindi, I know an Indian woman called Karishma
2026-08-27 09:28:57 <xentrac     > "data" in "database" is the past participle of the Latin verb meaning "to give"; it means "given", because originally "data" were the numbers provided in a math exercise to start your computation from
2026-08-27 09:29:29 <veltas      > So data is similar to input?
2026-08-27 09:29:32 <xentrac     > yes
2026-08-27 09:29:36 <veltas      > Interesting
2026-08-27 09:29:56 <xentrac     > the Latin verb is "dare" https://www.etymonline.com/word/data, from proto-Indo-European "*do-", "give"
2026-08-27 09:30:05 <xentrac     > people have been giving each other things for a long time
2026-08-27 09:30:14 <veltas      > Turns out
2026-08-27 09:30:38 <veltas      > Giving poor quality copper
2026-08-27 09:30:56 <xentrac     > no, that was sold!
2026-08-27 09:33:58 <xentrac     > Bengali is derived from proto-Gauda-Bangla, which is derived from proto-Gauda-Kamrupa, which is derived from proto-Magadhan, a dialect of Prakrit, which is derived from Old Indo-Aryan, which is derived from proto-Indo-European
2026-08-27 09:34:14 <xentrac     > and it turns out that her surname "Datta" also means "given" in Bengali today
2026-08-27 09:35:40 <xentrac     > the word happened to develop in the same way in Latin and in Prakrit; it's literally the same word as "data"
2026-08-27 09:37:19 <xentrac     > in Sanskrit "gives" became "dadati"
2026-08-27 09:38:09 <xentrac     > I'm not sure what it was in Prakrit
2026-08-27 09:53:03 <veltas      > Is 'give' always gratis then?
2026-08-27 09:54:28 <xentrac     > not always, but it carries that implication.  "Datta" generally means "gift", and "gift" (from our "give") *is* always gratis
2026-08-27 10:10:15 <veltas      > Of course
2026-08-27 10:28:43 <xentrac     > maybe "connotation" is the right word rather than "implication", which is maybe too strong
2026-08-27 10:32:04 <veltas      > That's more nuance than I would understand anyway
2026-08-27 15:35:06 <cleobuline  > the HC thing https://github.com/cleobuline/hc/releases/tag/HC-0.5.5
2026-08-27 15:36:45 <cleobuline  > https://ibb.co/Swf8rXsb
2026-08-27 16:01:52 <hc          > what hc thing
2026-08-27 16:03:40 <crc         > they're making something  hypercard inspired iirc
2026-08-27 16:07:18 <cleobuline  > hc thing is a personal remake of Hypercard :)
2026-08-27 16:07:47 <cleobuline  > full script compatible
2026-08-27 19:24:46 <crc         > I have a mostly functional ilo in APL now. An excerpt from it: https://charles.childe.rs/temp/ilo-core.apl.utf8
2026-08-27 19:27:16 <xentrac     > interesting
2026-08-27 19:28:05 <xentrac     > my first thought is that you could →optable[op] instead of doing 29 conditional jumps?
2026-08-27 19:28:13 <xentrac     > 30
2026-08-27 19:29:10 <xentrac     > maybe you're using an APL that special-cases →(x=y)ρz?
2026-08-27 19:30:15 <crc         > not sure. Using GNU APL. I'll see if I can improve that
2026-08-27 19:30:43 <xentrac     > I suspect that will make the code about 30% shorter and about 10× faster
2026-08-27 19:30:44 <crc         > for now, I'm happy to have it getting far enough that I can run it for a bit without it crashing immediately :)
2026-08-27 19:30:56 <xentrac     > congratulations!
2026-08-27 19:48:39 <xentrac     > well, probably not 10× faster.  maybe 2×
2026-08-27 19:54:45 <crc         > I didn't benchmark it, but it's nicely shorter; thanks for the suggestion
2026-08-27 19:54:57 <crc         > (I updated the excerpt)
2026-08-27 21:32:09 <veltas      > Bill Gates' love of APL
2026-08-27 21:32:30 <lisbeths    > I think alot of coders love APL
2026-08-27 21:32:35 <lisbeths    > many lovable things about it
2026-08-27 21:32:41 <lisbeths    > ive always wanted to learn it
2026-08-27 21:33:38 <veltas      > Good thing IBM came along and distracted him from whatever APL plan he had
2026-08-27 21:34:10 <veltas      > I love the concept but I've not tried it, and would want to try one of the dialects where you use ASCII
2026-08-27 21:41:21 <crc         > veltas: I'm also working on a J implementation ilo ilo, but that's not as far along.
2026-08-27 21:44:12 <veltas      > I guess it's a good exercise for learning a language to implement ilo in it
2026-08-27 21:48:31 <lisbeths    > fortran is honestly superior to c
2026-08-27 21:48:40 <lisbeths    > you're talking about an ordinary engineer versus the department of defence and ibm
2026-08-27 21:49:02 <lisbeths    > but the gcc compiler is so optimized that it can come close for most common coding tasks
2026-08-27 21:49:19 <lisbeths    > I have a mathematician friend who knows some special way to use fortran that calculates actual mathematics functions really fast comapred to c
2026-08-27 21:49:32 <lisbeths    > I might have been talking about on some special hardware tho
2026-08-27 21:52:10 <hc          > APL? I thought these days they do Lean?
2026-08-27 21:52:31 <hc          > lisbeths: I think the problem with
2026-08-27 21:52:57 <crc         > I'd rather write in C than Fortran personally
2026-08-27 21:53:00 <hc          > lisbeths: I think the problem with modern hardware is that its instruction sets are developed so quickly that it's hard to keep track and using an optimizing C compiler could be the safer bet to use the latest instructions?
2026-08-27 21:53:13 <lisbeths    > c is much nicer to use than fortran and is sufficiently performant on most tasks
2026-08-27 21:53:40 <lisbeths    > yeah instruction set writing is its own special type of hell if you are doing it for desktop processors
2026-08-27 22:00:15 <veltas      > I wish I knew classic Fortran
2026-08-27 22:03:11 <lisbeths    > c was originally meant to be a fortran-like language so you would probably be right at home
2026-08-27 22:03:23 <lisbeths    > fortran is very clean, for example all function definitions must go at the top of the file
2026-08-27 22:03:39 <lisbeths    > the type system is very similar to c
2026-08-27 22:03:51 <lisbeths    > or rather the memory layout of types
2026-08-27 22:05:09 <lisbeths    > fortran was so good it makes me suspicious how in the hell the military made it as one of their first compiled languages
2026-08-27 22:05:58 <lisbeths    > the guys in the #fortran irc channel are retarded. they only allow discussion of support questions, and for any support question they just send you a link to the fortran specificaiton
2026-08-27 22:06:06 <lisbeths    > its like why sit in the channnel all day for that
2026-08-27 22:08:50 <veltas      > So basically there's no discussion?
2026-08-27 22:08:57 <veltas      > Sounds like bliss
2026-08-27 22:09:36 <veltas      > It's also very forward-thinking with the way IRC is heading
2026-08-27 22:11:05 <veltas      > I'm writing a C program right now, a server
2026-08-27 22:15:30 <lisbeths    > we should just shut down all of the irc servers and replace it with a single url to mit website
2026-08-27 22:18:56 <crc         > 
ACTION still likes irc

2026-08-27 22:19:21 <X-Scale     > nothing beats irc
2026-08-27 22:20:08 <crc         > it's changed a lot over the ~25 years I've been using it, but I enjoy reading and (less frequently now) discussing things
2026-08-27 22:20:55 <veltas      > The activity doesn't stop dropping off, although this channel is still active and has interesting conversation
2026-08-27 22:22:18 <veltas      > Since the freenode / libera.chat drama a LOT of people just quit forever
2026-08-27 22:25:36 <lisbeths    > the thing about irc is that experienced coders prefer irc to discuss code. and its because irc is both open source and works in the terminal
2026-08-27 22:40:58 <X-Scale     > and irc is not related to the web cancer