Hide Idle (>14 d.) Chans
← 2022-03-21 | 2022-03-23 → ↓
crtdaydreams: !seen punkman
scoopbot: crtdaydreams: my valid commands are: src, uptime, version, help, update
crtdaydreams: !!seen punkman
deedbot: 2022/01/27 14:39:41 <punkman> I mean "bug"
verisimilitude: I recall in what seemed to be a Russian documentary: ``Asides from drinking, there's absolutely nothing to do here.''
verisimilitude: Wouldn't it be fair to assume part of the sanctions is to remove what makes life more bearable for many people, such as video games, and expect it to cause unrest?
verisimilitude: There's been a photograph circulating of a fat man who handcuffed himself to a closing McDonald's restaurant.
signpost: I wouldn't normally link twitter, but apparently this is what our government uses to communicate.
dulapbot: Logged on 2022-03-21 17:07:18 signpost: crtdaydreams: poast is a pretty bog standard IRC corruption of spelling.
mangol: "Noun poast (plural poasts) Obsolete form of post."
signpost: huh, there you go.
mangol: cannot source that anywhere apart from wiktionary :-//
mangol: potential prank
signpost: go back far enough and english spelling gets pretty loose, in either case
mangol: re: post vs article, another suggestion "farticle"?
mangol: and a section of one is a particle
dulapbot: (trilema) 2016-10-24 shinohai: s/article/farticle/
mangol: :D
mangol: press f in the chat
signpost gonna go pass out; see ya
verisimilitude: I loathe Wikimedia.
crtdaydreams: nearly finished 1984... absolutely horrified
verisimilitude: I read ``Brave New World'', and it wasn't nearly so great as I'd been led to believe.
crtdaydreams: Recently read Dune (Book 1), that was pretty good. If I don't move to more non-fiction I'll try and read the rest.
crtdaydreams: One Day in The Life of Ivan Denisovitch is pretty eye-opening.
crtdaydreams: I have been told to read Animal Farm though, any good?
verisimilitude: I don't know.
jonsykkel: but i vant dataflow chip now
jonsykkel: is there any text on this in adition to stuff in logs? emulators???? maybe have to build
mangol: jonsykkel: AFAIK dataflow is approximately the same thing as reactive
mangol: i.e. a program is essentially a dependency graph. tickle one node in the graph, and that causes an update to ripple through everything that depends on it
mangol: IOW a program is a makefile
mangol: [COSA][http://www.rebelscience.org/Cosas/COSA.htm] is reactive + synchronous
mangol: look into the "well known" synchronous languages from France. Esterel, Lustre, Lucid Synchrone (latest one is functional + synchronous)
mangol: i.e. higher-order synchronous. don't ask me what that means in detail
mangol: the Esterel line of inquiry begat some of the best stuff in CS, AFAICT. quietly, extremely successful and profitable in industry. like the L4 microkernel family.
mangol: SCADE is a GUI on top of Esterel, big success in industry
mangol: functional reactive programming (FRP) is all about dataflow. unclear whether Lucid Synchrone qualifies as FRP or something different
mangol: any spreadsheet is a dataflow program AFAICT
mangol: https://en.wikipedia.org/wiki/G%C3%A9rard_Berry <-- seems like he'd belong in any proper CS pantheon
mangol: https://www-sop.inria.fr/meije/esterel/esterel-eng.html <-- good set of links to reference material
mangol: https://misc.lassi.io/2022/foundations-of-esterel.pdf <-- must-read paper by Berry. if you only read one paper about this stuff, make it this.
mangol: Alan Kay's idea of OOP (i.e. message passing) is not far off from reactive, and neither are neural networks
mangol: basically programming has two main schools, math-inspired (algorithms and proofs) and biology-inspired (objects and networks)
mangol: the above stuff is the biology-inpired end (criminally neglected in conventional CS curricula)
mangol: the French dudes, being french, when they started studying networks made many theory arcs toward math
mangol: which you can see very clearly if you try to work thru the Lucid Synchrone primer
mangol: and Esterel compiles into a state machine
mangol: SCADE input is a visual equivalent of an esterel program, output is some autogenerated C monstrosity
jonsykkel: right, make -j9000000000001, enjoy keypres---0.000001ns----->pixels on screen
mangol: https://www.youtube.com/results?search_query=scade+tutorial <-- many videos about SCADE from Ansys, the company that bought the startup that made it
jonsykkel: good list of links, tanks! will get reading after slep
jonsykkel: this is cool stuf
jonsykkel: specialy intresting if theres something in here i can pley around with
mangol: happy to help. i should get back to this stuff too!
mangol: start from the Foundations of Estere paper
mangol: sadly, there's not much to (cheaply!) play around with. try to get an evaluation copy of SCADE or LabVIEW somehow?
mangol: or if you're into audio, Max/MSP or PureData
mangol: https://www.ansys.com/products/embedded-software/ansys-scade-suite <-- look at the applications in this brochure and you get an idea of why it isn't cheap
mangol: re: why these aren't household names, obligatory naggum "is lisp dying?" / neurosurgery poast: https://www.xach.com/naggum/articles/3140985758772364@naggum.no.html
mangol: can't resist one more link... re: microkernels, this blog is the thing to read: https://microkerneldude.org/
jonsykkel: cool, ill check out teh paper
jonsykkel: found labview torrent also, will put on my win7 machine dedicated for abuse
asciilifeform: mangol: the only dataflow system asciilifeform had used in earnest (not counting hardwarisms in verilog etc) was wolfram's. and was bog-slow, per se, and from what asciilifeform suspects are fundamental impedance mismatches with vonneumannistic irons
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087016 << had read, tho afaik not mentioned in the log. theoretically nifty
dulapbot: Logged on 2022-03-22 07:34:15 mangol: https://misc.lassi.io/2022/foundations-of-esterel.pdf <-- must-read paper by Berry. if you only read one paper about this stuff, make it this.
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087001 << some yrs ago asciilifeform set about designing one, then hit the 'no usable fpga' problem
dulapbot: Logged on 2022-03-22 06:53:32 jonsykkel: but i vant dataflow chip now
dulapbot: Logged on 2022-03-22 09:57:42 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087016 << had read, tho afaik not mentioned in the log. theoretically nifty
mangol: if there was more time, i'd explore a GUI programming language based on this synchronous / reactive paradigm
mangol: it's unconscionable that UIs still jam and show the hourglass cursor dozen times a day
mangol: a UI can't be truly comfortable without being based on response time guarantees of some ort
mangol: *sort
mangol: since general-purpose OS is unavoidably based on unknown wait times (the "interactive" paradigm in Berry's paper) the synchronous layer would have to check "mailboxes" where the interactive layer posts updates
mangol: the human body is a good model here. your eyes and ears don't randomly jam and give you an hourglass / wait tone
mangol: yet the circulation systems of the body have somewhat unpredictable pace, and give you notifications
mangol: heart pumps, etc, but it's not that precise
dulapbot: Logged on 2022-03-21 22:36:59 mats: http://paste.deedbot.org/?id=e2Oa transcript here
signpost curious how many overt acts of war russia is prepared to tolerate.
asciilifeform: signpost: a little obsolete, too, iirc the abramoviches' yachts already confiscated (mostly from repair depots). reich is still free to try its luck with actual ru navy, lol
asciilifeform a++ in favour of deoligarchization -- along with good % of ru public apparently
dulapbot: Logged on 2022-03-21 18:22:34 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-21#1086901 << the ru oligarchs fwiw were creatures of the reich, whose purpose was to steal & exfiltrate as much as possible of what 80yrs of collective orcish sweat had built. and nao that the brakes were hit on this exfiltration, naturally, 'pig aint gonna get any fatter, time for the knife'
asciilifeform: signpost: rumour has it that mr.p gave the pigs fair warning, 'take the dough and yachts home, or get fucked'. they chose to get fucked.
signpost figures the piggies mostly deserve the slaughter.
signpost: still a notable escalation that US shall be paying mercs/pirates to rob citizens of another great power.
signpost doesn't trust usg to use that judiciously.
asciilifeform: aint as if they weren't already stealing anyffin that floats, flies, etc. that dun have w/ what to shoot back
asciilifeform: for 100+y
asciilifeform: ukr gold reserve, allegedly, already flown to reich.
asciilifeform: a la iraq's
PeterL: in this terminology, are oligarchs separate from Putin?
asciilifeform: PeterL: there's 2 basic flavours of oligarchs in ru: 1) the comprador elite 2) the elite loyal to the land (i.e. no plans to miami)
dulapbot: Logged on 2022-03-21 18:22:34 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-21#1086901 << the ru oligarchs fwiw were creatures of the reich, whose purpose was to steal & exfiltrate as much as possible of what 80yrs of collective orcish sweat had built. and nao that the brakes were hit on this exfiltration, naturally, 'pig aint gonna get any fatter, time for the knife'
asciilifeform: at this pt moar or less clear to ~erryone that mr.p himself aint a (1)
asciilifeform: reich's bet is that he's surrounded helplessly by (1) and will be poisoned. ~month into the wank, not happened yet afaik
mats: even if he’s killed, it’s not clear how anyone steps back from this
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087064 << imho rather obv. at this pt that reich doesn't recognize the existence of any powers other than self. the lulz where usg talking head emitted a 'china must obey...' quite illustrative.
dulapbot: Logged on 2022-03-22 12:10:45 signpost: still a notable escalation that US shall be paying mercs/pirates to rob citizens of another great power.
mats: nobody is getting their shit back and unwinding sanctions ranges from difficult to impossible
asciilifeform: mats: indeed, even under yeltsin's 90s 'pleez dun hurt us' muppet regime, sanctions from '70s-'80s stayed in place
asciilifeform: there's no mystery re why.
dulapbot: Logged on 2022-02-28 14:43:20 asciilifeform: verisimilitude: moar interesting is the 'big pic' -- the current war in fact started from 4th crusade, and continues episodically in form of various 'drang nach osten' by impoverished western eureichs, who will not give up dream of 'lebensraum' and delish phree minerals from 'unfairly squatted by orcs' 1/6th of planet3
asciilifeform: and note that no guarantee who will sit on throne if putin poisoned will be similarly moderate (yes). so far very little in the way of countersanctions -- gas still pumped (through ukristan!) to eureich; refined u -- to usa; rockets not yet reinstalled in cuba (why not? high time) ; etc
asciilifeform: for that matter even the 'withdraw from all intellectualproperty(tm) idiocy and open a ministry of warez' thing not afaik actually happened yet
PeterL: does Putin have a designated heir apparent?
asciilifeform: not afaik
mangol: that's weird
asciilifeform: imho not so strange; nobody seems to know who is heir of e.g. soros either
asciilifeform: ru at least has a publicly known ruler, unlike reich
mangol: reich has regressed to the point that kamala harris can be vice president; not surprising that they'd mess something up
mangol: though the old guard seems to have sharp people
asciilifeform: mangol: the 'elected' muppets in usa decide nothing
mangol: ofc, but even as a public display it's embarrassing and easily avoided
mangol: and could get out of hand -- start drinking own kool aid
billymg: https://twitter.com/McFaul/status/1506087506298621954 << not sure if this is satire or not but appears to be earnest
asciilifeform: billymg: prolly not satire, d00d is former reich ambassador to ru, and currently famous for 'there are no innocent russians' and other lulz
billymg: the replies are lulzy: "great idea!"
billymg: asciilifeform (or anyone): have you tried using this for hooking up postgres to SBCL? https://github.com/marijnh/Postmodern
asciilifeform: billymg: asciilifeform has used; worx
billymg: neato
asciilifeform not tried anyffin 'fancy' with it, but does issue queries, return results, etc
billymg: yeah, don't think i'll need anything fancy. i've started reading SICP and i think when i get far enough along in that i want to re-write my crawler in lisp
billymg: not the www portion, just the actual crawler part
mangol: was some kinda consensus reached about file transfer over pest? would be nice to be able to broadcast "a file with this hash is available" and to ask the net "where can i get contents of file with this hash"
asciilifeform: mangol: iirc signpost is working on this
mangol: would this involve adding new message types?
asciilifeform: mangol: almost certainly
mangol: signpost: are you extending blatta or smalpest?
asciilifeform last knew, he was working on the algo, but will let signpost answr concretely
mangol has about half the code for a scheme client, not yet tested
asciilifeform: mangol: neato!
mangol: can add a "scheme.lisp" wrapper file for running the same code in CL implementations if there's interest. or one of the other lispers here can rewrite in CL
mangol: my personal opinion is that CL and Scheme should be merged, but that's several years away at best, and very controversial
mangol: tailor-made wrappers are not hard
PeterL: for a lisp beginner, would you suggest learning scheme or CL?
mangol: depends on your goals and personality. starting with either is fine
mangol: elisp is also fine as a 1st lisp
mangol: all of these have notable strengths and weaknesses
mangol: if you feel a strong pull toward a particular dialect that you can't rationally explain, start with that one. curiosity is always the best motivator.
mangol: otherwise, scheme is way more elegant and jewel-like than CL. CL is way more practical out of the gate.
mangol: you can write CL about as elegant as Scheme, and Scheme about as practical as CL, but takes work
mangol: CL's strength is that you get ~every way to solve a problem out of the box. weakness is that it's hard to resist mental masturbation
asciilifeform: mangol: which scheme are you using ? ( standard dun include udp or much of any i/o.. )
mangol: Scheme culture is very "get down to business", "just get it done". scheme programming doesn't really involve thinking about what tool to use.
mangol: if you use ML (SML or OCaml), the many parallels to Scheme cannot be ignored
mangol: if you go from Haskell to ML, it's the same "my god, these people just solve the problem and move on" culture shock as you get moving from CL to Scheme
mangol: the reason there are so many scheme implementations and subsets is that we schemers do our mental masturbation on this level
mangol: whereas common lispers do it within the confines of ANSI :)
mangol: the "just get it done" culture is entirely enabled by recursion (tail calls and "named let") and higher-order functions
mangol: you just use those for everything. a macro and a record type here and there, that's basically it
asciilifeform: mangol: i thought the reason why over9000 scheme impls. is that the standard is largely missing i/o
mangol: i never compare 5 different ways to solve something in scheme. in CL i did that all the time
mangol: asciilifeform: the reason is just personality. the technical reason is simply that the standard is small enough that one person can implement it.
mangol: schemers are an extremely individualistic bunch
mangol: https://github.com/FFTW/fftw3/tree/master/genfft <-- this codebase is a good example of the "just get it done" culture -- this is in OCaml
mangol: if you look at the source files, there's virtually no type wankery. just recursion and the types are mainly record types
mangol: schemers use this same style
mangol: the application needs to do something, you pop open a source file and there's straightforward recursive code for the application to do it
mangol: asciilifeform: re which scheme -- if you want to replace a shell script use gauche, to replace a c program use chicken
mangol: there are endless choices for more esoteric requirements
PeterL: whoosh, all these terms are just going over my head.
mangol: PeterL: are you familiar with the basics of recursion?
PeterL: I believe so - a algorith that can call itself?
mangol: yes. the classic "hello world": (define (factorial n) (if (<= n 1) 1 (* n (factorial (- n 1)))))
mangol: (map factorial (iota 10)) => (1 1 2 6 24 120 720 5040 40320 362880)
mangol: factorial is a recursive function, and map is a higher-order function (i.e. takes another function as an argument, in this case we give it the factorial function)
mangol: (iota 10) just gives the list (0 1 2 3 4 5 6 7 8 9)
mangol: the above is scheme code. you can do the same in CL but functional programming and recursion is somewhat less elegant there, so CL coders do less of iit
mangol: recursion is hard to learn (for non-trivial use), but once you learn it becomes second nature. and in scheme (and ML) you can do almost everything with it, so you don't have to learn much else
PeterL: why is it called iota and not range?
signpost: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087097 << ran production systems on it, worked great.
dulapbot: Logged on 2022-03-22 13:45:37 billymg: asciilifeform (or anyone): have you tried using this for hooking up postgres to SBCL? https://github.com/marijnh/Postmodern
signpost: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087104 << determined to first fart out vtronic linux, then get to this.
dulapbot: Logged on 2022-03-22 14:26:06 asciilifeform: mangol: iirc signpost is working on this
signpost: anyone who beats me to, entirely welcome, will help as I can
billymg: PeterL: i am currently reading and working through the exercises in SICP and Thinking Recursively, i'd highly recommend both
signpost: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087107 << only code I've written is the fountain code impl on my blog, which is in python. that said, I've gotten about halfway into a CL impl of same with proper parallelism
dulapbot: Logged on 2022-03-22 14:28:51 mangol: signpost: are you extending blatta or smalpest?
billymg: the first is in scheme (rather than lisp), but my understanding so far of the languages is that they're similar enough you can start with either to learn the fundamentals. the other book gives examples in Pascal, but i complete the exercises in scheme
signpost: (blatta is in py, so can potentially pick up the py impl and refactor into using multiprocessing for parallelism. apparently this is possible upon shared numpy arrays)
signpost: SICP was the item from MIT curriculum before they bent over for the industry and switched to python iirc.
signpost took a girlfriend through it once upon a time.
PeterL: mangol: I could write an almost identical function in python, is there something special that scheme brings?
signpost: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087121 << I found CL displaced arrays useful for the CL fountain code, perhaps there's similar in scheme
dulapbot: Logged on 2022-03-22 14:41:56 mangol: CL's strength is that you get ~every way to solve a problem out of the box. weakness is that it's hard to resist mental masturbation
signpost imagines so.
signpost: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087128 << I suspect this "everybody has their own pet scheme utils" thing becomes a communication barrier when reading other people's code.
dulapbot: Logged on 2022-03-22 14:47:12 mangol: the "just get it done" culture is entirely enabled by recursion (tail calls and "named let") and higher-order functions
signpost: PeterL: in re: python vs scheme, there are huge ideological differences that you may or may not care about.
signpost: python came from an attempt at "end user programming", not unlike SQL. even got some DARPA money for this early on.
signpost: from this you get the massive library of standard data structures, and the childlike rigid syntax.
signpost: (and also that routinely python acts as a scripting language over "real programming" in C)
signpost: scheme is designed from a perspective of (curious whether mangol is satisfied with this broad generalization) maximizing the leverage of expert programmers.
signpost: this means boiling the language down to the fundamentals of programming, and giving the programmer tools to design higher-level language for their problem as they see fit.
signpost: this isn't different from CL in my mind except that at a certain point folks that had done ^ independently came to some agreements at an industrial level on what much of the useful higher-level language was.
signpost ftr isn't being a snob re: childlike syntax, wrote a product in python for ~decade, worked fine as a text munger in between db and webserver.
signpost: risk of staying in a python for too long is you start saying "what's the bittorrent of pest" rather than "from first principles, how do I achieve efficient, fault-tolerant distribution of this byte-array among my friends that want it.
signpost hears a lot of x of y thinking among startup bros, i.e. "uber of dog-grooming", "airbnb of community gardens"
asciilifeform regards pythonism as an item on the awk, sed, perl, etc 'dirty hacks' continuum
asciilifeform would like to be rid of it
dulapbot: Logged on 2020-07-18 20:01:25 asciilifeform: trinque: i'd dearly like to throw out the pythons, perls, sh atrocities, etc. in favour of sumthing that makes some semblance of sense (incl. when its binary thrown into objdump -D ... )
verisimilitude: Shouldn't tabular programming be able to help there, asciilifeform, since most of the memory dump would be static data?
asciilifeform: verisimilitude: perhaps oughta elaborate. in context, 'make sense when objdumped' meaning, a la e.g. 'm', weighing next to nuffin and invoking no external libs
asciilifeform: i.e. being small enuff that someone could, if wants, choose to study it by squeezing the binary through disasmer of his choice and straight to printer
asciilifeform: think back to '80s micro culture, when proggies often enuff were published in the shape of a pg or 3 of magazine fulla hexdump
asciilifeform: reader was expected to be able & willing to make sense of it, even w/out 'source' in the usual sense being given.
asciilifeform: for that matter, verisimilitude's proggies for his emulated micro, as i understand, are in similar style
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087138 << what asciilifeform would prefer would be to 'replace errything', with a minimal core + all necessary extensions running on top of same, eventually displacing the old sad os
dulapbot: Logged on 2022-03-22 15:00:36 mangol: asciilifeform: re which scheme -- if you want to replace a shell script use gauche, to replace a c program use chicken
dulapbot: Logged on 2022-03-17 14:55:09 asciilifeform: upstack, even w/out any prospective of 'moon missions', could start by 'localized' minimalism of defining a usable prog lang that can actually be implemented in <32kB x64 asm.
asciilifeform: and to fughet about c entirely
mangol: asciilifeform: your choices are forth and scheme, more or less
asciilifeform: imho entirely feasible to have grand total of 32kB of executable x86 coad on the machine.
mangol: or basic / lua / other flavor of the month
asciilifeform: mangol: scheme (or a scheme-like statically scoped lisp) would be the ticket.
mangol: for all intents and purposes, the whole lisp family is statically scoped in present times
mangol: there are a few teenage impls that use dynamic scope, but no serious lispers use those
asciilifeform: moore already did this exercise w/ forth. did not result in particularly usable system imho (not in fact a high level language, demands much of same 'sewing under magnifying glass' as simply writing in asm, in practice, to do anyffin nontrivial)
mangol: they usually also bear other marks of teenage engineering, like randomly renaming all the operators and using diff. num. of parentheses for `let` and the like
mangol: yes, moore's colorforth site and fireside chats were a goldmine of thought provoking stuff
mangol: and yes, he tried "sourceless programming", was too much even for him
asciilifeform: mangol: asciilifeform has nuffin against renaming operators, so long as there's some kinda logic behind it (e.g. asciilifeform himself used nonstandard ops in his forth-like 'peh')
mangol: also re: asm, the core of scheme and ML is basically an assembly language for the lambda calculus
asciilifeform: mangol: as showin in 'secd' etc, yes
asciilifeform: getting the x86 nonsense (incl. toolchains) 100% outta the way in an absolute minimum of mass aint unimportant, however.
asciilifeform: and to date, not seriously attempted afaik.
mangol: if you want to end up with some subset of scheme in the end, i've done a lot of exploration on that
mangol: as far as prog langs go, scheme is just about the best foundation for subsetting you can find
dulapbot: Logged on 2022-03-22 15:39:42 signpost: hears a lot of x of y thinking among startup bros, i.e. "uber of dog-grooming", "airbnb of community gardens"
dulapbot: Logged on 2020-12-31 14:24:36 asciilifeform: verisimilitude: yarvin was evidently trying to emulate various 'giants of industry' in picking up public materials, giving'em a coat of paint, and then claiming 'inventions'
asciilifeform: 'let's pound the square rod through the round hole with an over9000x bigger sledge.' 'and let's paint it purple' 'and use cloud disk' etc
mangol: in fairness to the startup bros, the goal of a startup is to make money. if being "the uber of dog grooming" makes money, it's fair game.
mangol: the business world doesn't look kindly upon original people anyway, unless they sacrifice a lot and get very lucky
asciilifeform: it's the 3yo's 'creativity', and most dun grow past this stage.
asciilifeform: ftr, pasta: 'It is useful, however, to distinguish between tinkering and creativity. Tinkering consists of exploring relatively minor variations on known themes, or subjecting new stimuli to an array of already known techniques. Thomas Kinkade rarely creates and mostly tinkers. Babies tinker constantly. They put every new object in their mouth. Eventually they figure out that most things are not good to eat. When they
asciilifeform: develop motor control, they throw things. Serious curiosity consists of actively seeking new kinds of stimuli. Creativity consists of juxtaposing objects and ideas in new ways, and having a sound intuition for separating the significant result from the trivial… Now we can address the contention that children are innately curious. They are not in the sense used here – they are tinkerers. The commonplace observation
asciilifeform: that children have short attention spans is direct refutation of the notion that they are creative and curious in any deep sense. The tragedy of our society is not that so many people outgrow their childlike curiosity, but that so few do. The adult equivalent of childlike curiosity is channel surfing and the ten-second sound bite.'
mangol: open source / hacker culture wears "tinkerer" as a badge of pride. i never understood what's so awesome about tinkering. your quote finally puts it in the right perspective
mangol: i always thought tinkering is boring and yearned to generalize. iirc as long as i've had access to a computer
mangol: i.e. even as a kid, tinkering was boring
asciilifeform: whole piece worth reading imho.
mangol: yeah, great stuff
mangol: thanks for linking it
mangol: + always funny to see smart academics tiptoeing around politically sensitive topics where the layman's perspective is that the "heretic" in question is stating something pretty obvious
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087212 << very rarely is their objective to 'make money' in the superficially-claimed sense. normally rather is to pawn the thing an agent of the reich's printing press and walk.
dulapbot: Logged on 2022-03-22 17:07:02 mangol: in fairness to the startup bros, the goal of a startup is to make money. if being "the uber of dog grooming" makes money, it's fair game.
mangol: "There is no persuasive evidence that any societies have ever had a high proportion of people who were deeply curious in a systematic, disciplined way."
mangol: >> deeply curious and creative people are generalists, i.e. curious and creative about every topic. one of those topics is society itself. too many such people, everyone wants to shape society in his image
mangol: over 9000 different scheme implementations ^W^W societies
verisimilitude: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087186 Yes; most of the games I've documented, but didn't write, were only available in the form of sexadecimal dumps beforehand.
dulapbot: Logged on 2022-03-22 16:50:07 asciilifeform: for that matter, verisimilitude's proggies for his emulated micro, as i understand, are in similar style
verisimilitude: Important, mangol, is the pattern of Insidious Optimizations.
mangol: verisimilitude: sex dumps on tape. sex tapes.
verisimilitude: Base six is a poor choice.
mangol: there are 36-bit architectures
verisimilitude: Yes, but those were sexier times.
mangol: those were the days. and nights.
asciilifeform: howthefuck can it be 'free' ?
verisimilitude: It could be implicit from something else, say.
asciilifeform: lulzily, current-day irons in fact do all kindsa ~entirely undocumented 'register renaming' and similar attempts to silently turn 1 instr into entirely other, behind the scenes
asciilifeform: and as result of it happening in undocumented iron, programmer has 0 control over it, and often enuff makes ~impossible to actually optimize as a consequence
verisimilitude: That's not a real stack, mangol.
verisimilitude: That architecture is still gross, mangol.
verisimilitude: What relevance has it to me, mangol?
mangol: you were talking about registers in the post you linked
verisimilitude: Alright.
verisimilitude: The latter of the pair is the better read.
mangol: the Elision article. you linked it a few days back
dulapbot: Logged on 2022-03-22 15:15:23 PeterL: why is it called iota and not range?
mangol: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087162 << guaranteed tail calls so you can tail-recurse over a huge data structure or number range and not blow the stack
dulapbot: Logged on 2022-03-22 15:25:29 PeterL: mangol: I could write an almost identical function in python, is there something special that scheme brings?
mangol: however, (define (factorial n) (if (<= n 1) 1 (* n (factorial (- n 1)))) is not tail recursive
mangol: (define (factorial n) (let loop ((product 1) (n n)) (if (<= n 1) product (loop (* product n) (- n 1))))) is a tail recursive version
mangol: kinda hard to read on irc
mangol: but the idea is, the last thing that `loop` does is call itself
mangol: hence "tail"
mangol: a compiler or interpreter can turn the call into a jump. no need to allocate a new stack frame, can overwrite existing frame
mangol: a Python, Common Lisp, or C compiler/interpreter can do that, but Scheme coding style is centered entirely around tail calls
mangol: having it guaranteed in the spec means you can safely use recursion anywhere you'd use iteration in other langs
mangol: which means people end up using recursion all over the place, which in turn means people don't bother to add iteration constructs to the language
mangol: so you end up with a drastically different coding style. kinda like comparing infix vs postfix calculators. you can compute the same stuff, but completely different feel
mangol: re:minimalism, recursion is universal enough that you can make a really compact langauge (R5RS spec is 50 pages) that also yields really compact programs
mangol: usually compact programs require a sophisticated, complex language implementation. conversely, a compact lang impl is too wimpy to support compact programs
mangol: basing the language around recursion lets you have both
mangol: i'm an essentialist when it comes to languages. every prog lang has essential features that lead to a characteristic coding style, and the rest of the well-known features of the language are just historical accidents.
verisimilitude: The solution to a recursive algorithm in APL is to seek a closed form.
mangol: tail recursion elimination is the essential feature that leads directly to scheme's characteristic coding style (in stark contrast to CL and other lisps)
mangol: Lisp = code is lists; Emacs Lisp = Lisp + closures; Scheme = Lisp + closures + tail calls; CL = Lisp + closures +generic functions
mangol: verisimilitude: "closed form" approx. = convergence?
verisimilitude: I believe so.
verisimilitude: The closed form of (SUM (RANGE 1 N)) is (* (/ N 2) (+ 1 N)).
mangol: neat. does that technique work mainly for numbers or more generally, e.g. for trees?
verisimilitude: Rephrase that.
mangol: take e.g. a HTML document represented as a tree of elements, and recurse over that -- filtering, adding, or removing nodes, or computing some kind of summary of them
mangol: in principle, any tree can be represented as an array
verisimilitude: I guess so.
mangol: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087171 << python's main achievement was to show that most of what people thought is "real programming" is basically just scripting. a laudable feat
dulapbot: Logged on 2022-03-22 15:32:39 signpost: (and also that routinely python acts as a scripting language over "real programming" in C)
mangol: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087172 << quite close - maximizing leverage with a minimal core, afaict.
dulapbot: Logged on 2022-03-22 15:34:03 signpost: scheme is designed from a perspective of (curious whether mangol is satisfied with this broad generalization) maximizing the leverage of expert programmers.
mangol: lua later designed using the same principle, to good effect
mangol: and earlier, forth
mangol: and apl and prolog
verisimilitude: I'm to understand Prolog is a weakened PLANNER.
dulapbot: Logged on 2022-03-22 18:43:55 mangol: http://logs.nosuchlabs.com/log/asciilifeform/2022-03-22#1087171 << python's main achievement was to show that most of what people thought is "real programming" is basically just scripting. a laudable feat
← 2022-03-21 | 2022-03-23 → ↑
Random(asciilifeform) | Download hourly DB snapshot | Get Source Code