log☇︎
257900+ entries in 0.159s
mircea_popescu: yes but you can definitely play to it.
asciilifeform: they share the bottleneck.
mircea_popescu: asciilifeform ah but they don't have to!
mircea_popescu: asciilifeform so good for you! none of this is any sort of genius, just systematic refusal of dumbassery.
asciilifeform: mircea_popescu: it dun work because, again, pc arch is retarded. the processors fight over the bus.
mircea_popescu: phf imagine if you will, the beauty of a system where "threads" make no sense in the first place.
asciilifeform: mircea_popescu: believe or not, i thought of this.
mircea_popescu: ie, "this our lisp os only works on processors with no less than 8 cores. because we keep all these things pre-loaded and ready to go, and then every time you try to add we do 16 checks in one tick. one op mujlti data ftw!!1"
phf: but yes, cmucl in this channel started because i was thinking of porting it to hardware. it's obviously what every lisper wants/tries to at some point. in order to achieve it though, i kind of have to read and grok and read again the code the cmucl code, which is what i've been doing..
asciilifeform: things that the silicon ought to be doing. and ~WAS~ doing in 1976.
asciilifeform: mircea_popescu: understand, addition there is no longer 'ADD rax, rbx' but a pound of type bit checks and possible code paths on failure or type-promotion etc.
mircea_popescu: intrinsic addable-value far exceeds, yes, which is why we're even talking.
asciilifeform: pointing out that the intrinsic complexity far exceeds linux, bsd, whichever c os, kernel.
phf: which would be consistent with the v threads
mircea_popescu: but because you can read it, you get to see vendor magic numbers.
mircea_popescu: asciilifeform those are not disadvantages but advantages.
asciilifeform: and if you use ANY 'c code in the drivers' you get gabriel_laddel.
mircea_popescu: what can i tell ye.
asciilifeform: mircea_popescu: the two are really not comparable.
phf: mircea_popescu: people basically come to same conclusion, and then when it comes to putting in "2 years of living from inside an os without youtube" balk
asciilifeform: mircea_popescu: try to understand that linux is barely a thing at runtime. whereas a lisp os has to make up for 10,001 idiocies of the hardware, e.g., lack of garbage collector, lack of type bits in every (yes) machine word that get tested on every (yes) arithmetic operation, etc.
mircea_popescu: incidentally, the tasks are here in my vierws overstated. linux had the tasks as hard as described, and it did break through, on shittier internet with fewer people milling about. lisp already has c code it can read for many of the drivers etc. this is major advantage. easier job to come 2nd.
asciilifeform: but mircea_popescu has it, the sane approach involves 0 c and 0 unix, and i came to this conclusion in '07.
phf: asciilifeform: that wasn't a challenge, i was just disappointed
asciilifeform: phf: why take my word for it, thing was published, eons ago, read.
phf: asciilifeform: oh, i thought for some reason they had civilized VOPs and such
mircea_popescu: make a third, but in this direction, not in that, if you will.
asciilifeform: mircea_popescu: they're dead ends.
mircea_popescu: "emulated" toys for chitlines, dear god.
mircea_popescu: be all this as it may, time to grow up and stop pretending adult lisp is anything but these two.
phf: well, only in a sense that it's written for it? i'm not sure married is the right word
asciilifeform: movitz is worse yet because it is ~married~ to 32-bit x86
mircea_popescu: you apparently don't recall early linux worth twopence.
asciilifeform: and the hardware ~is~ shit. boot up one of these (if you can actually get it to boot.) and say hello to 1 fps graphics, disks without dma (you don't know what these feel like until trying personally), nic that works when the moon is full strictly, etc. ☟︎
phf: i think one could take either of those and turn them into real systems, the way we did with trb, but it's the same class of work
mircea_popescu: so then it seems to be that it is an offense before the gods and an insult to man when in any discussion of lisp anyone proposes any other solution than "pick either movitz or mezzano"
asciilifeform: no fundamental reason. just as railway bridge could, in principle, be glued from toothpicks.
mircea_popescu: i mean, is there some fundamental reason such can't be written ? that's a problem. otherwise...
mircea_popescu: how is this a drawback ?
phf: movitz and mezzano are common lisp compilers, they are just shitty ones. they are good enough to boot an os on x86, but they are missing bits (large chunks of standard) because nobody's written those
asciilifeform: building railroad bridge out of toothpicks.
mircea_popescu: i can't parse that. say again ?
phf: the problem is that then you're stuck redoing many man years of rewriting cmucl/sbcl in them
mircea_popescu: what killed these ?
phf: the linux approach actually's been done twice, i.e. write something shitty, but it boots and then improve it until it's somewhat ok. there's https://common-lisp.net/project/movitz/ and https://github.com/froggey/Mezzano
a111: Logged on 2016-09-17 06:16 ben_vulpes: some people go crazy, some people have crazy high tolerances, some people even eat a shitload over a long time and then one day break, the variance is wild and statistical claims cannot be made beyond "oh fuck might hose your brain under entirely unknowable circumstances but hey have fun with the tradeoff analysis kid"
BingoBoingo: http://btcbase.org/log/2016-09-17#1543758 << Stereotypical last drunk seems to be X beers and a blackout where X<4 ☝︎
asciilifeform: the other bit is, just as in the fpga thread, by the time anyone ~does~ spend the man-years, the requisite hardware is unavailable. see, e.g., 'movitz'.
mircea_popescu: too fucking smart for their god damned own good.
mircea_popescu: that's also the reason steam engine wasn't done by greeks.
phf: mircea_popescu: the real reason it's not been done is because it's fucking hard and nobody cared enough to spend the necessary man years
a111: Logged on 2016-09-17 05:58 adlai is incidentally now sober for the longest he's been in weeks - and he has a loaded vape within arm's reach - so i'd say step one is not that relevant >> http://log.mkj.lt/trilema/20160826/#219
BingoBoingo: http://btcbase.org/log/2016-09-17#1543731 << All we've got is today, it works when you work it (tm)(r) ☝︎
asciilifeform: mircea_popescu: the hardware is very literally shit. e.g., ever heard of acpi ?
mircea_popescu: tell you what : the hardware's the hardware and the software's the shit.
mircea_popescu: asciilifeform yes, but you also realised the hardware was shit that time we went to olympus and venus asked you to fuck her.
asciilifeform: i blew years on it and then realized that the hardware is shit .
phf: the thing that ascii mentions he'd pay for "cmucl on hardware" explicitly involves ripping out C code and then doing necessary work to support some minimum set of x86 drivers like keyboards etc.
mircea_popescu: linux happened in a few years, need i remind you, and those were earlier, drier years.
mircea_popescu: so then for shame, all these dickheads going around holding their heads in their hands and talking about dumb shit while doing no useful work.
phf: well, you don't need C for that and there are no excuses
mircea_popescu: now, x86 i can already buy, which of tll, lelnel don't exist and what's the excuse.
mircea_popescu: there's the following parts : hardware, including memory, cpu, disk etc. we shall call this x86 for short, even if it's a longer story ; tll, which is a compiler. does exactly same as the gcc does in linux : takes lisp code, spits out bytecode ; and finally the lelnel, which does what the kernel does in linux.
mircea_popescu: lisp is the system.
phf: are you running it on a linux system? because ~if and only if~ that you are you have to bring something extra into the picture
phf: it doesn't do anything more than that
mircea_popescu: so then let it do this and no more.
phf: we agree on that point, and lisp is capable of doing this
mircea_popescu: holy hell! the processor takes opcodes. you produce the opcodes. that's it!
mircea_popescu: now you're going to import dos into this discussion ?
phf: you know dos "com", which was just a flat sequence of bytecodes and when you did foo.com it would just start executing foo.com with the first byte
phf: programs that have to run on linux don't behave like dos COM files
phf: but it's the whole point!
mircea_popescu: suppose for this discussion it's already started ; else yes, you need a running machine to compile your image as we said.
mircea_popescu: it produces asm for the x86, it runs on it.
mircea_popescu: i am trying to run it on an x86.
phf: well, but you're trying to run it on your linux correct? or we're talking pure hardware layer
mircea_popescu: so then ? i didn't import any linux anything.
mircea_popescu: you just said, lisp compiles to native bytecode.
mircea_popescu: phf there is no unix.
phf: mircea_popescu: problem is that when you do a execution like ./foo on unix, ~unix~ expects "foo" to contain certain bits and do certain kind of work to appease unix upfront. that work is specific not to "x86" but to "linux v 1.1.2, compiled on full moon, by three blind monks"
asciilifeform: i'ma let phf finish before i return to the iron bit.
mircea_popescu: phf so then what the fuck is the problem ?
phf: sbcl/cmucl already compiles whatever code you ask it to to native, architecture specific assembly, that's not the C layer's role
mircea_popescu: looky : either something's fundamentally broken with lisp, or else it has no "architecture" ; much in the way purple has no shape.
asciilifeform: mircea_popescu: this is gonna be the fpga thread all over again, isnnit. fact is, the 10,001 man-years are not available, and certainly not six times every morning before breakfast.
mircea_popescu: phf i'm not so sure that's what i said though.
asciilifeform: mircea_popescu: the difficult part is 'which skank.' it isn't 1988 and there is no standard pc arch.
phf: mircea_popescu: that's not what the layer between os and lisp does in the case of sbcl/cmucl. it explicitly doesn't do primitives, because those are, like you said, written in lisp, that's compiled into native bytecode (using VOB's, i.e. chunks for assembly)
asciilifeform: mircea_popescu: that's the easy part
mircea_popescu: lisp is supposed to exist out of what, six primitives or some such ? write them in asm. once you're done the job's ~done.
mircea_popescu: your application for thick condoms has been denied. rawdog that skank.
phf: vm in this case means something closer to x86 bios
asciilifeform: mircea_popescu: in this context, folks often say 'vm' when they are really describing emulator of a sane cpu arch.
phf: yeah, sorry that was ambigious
mircea_popescu: you keep going back to this "my code runs in javascript via php" approach to life. it's nonsense. there's 0 need for c.
asciilifeform: aaaaaand this is how mircea_popescu arrives on the island where i sat in 2010.
mircea_popescu: there is no need to have c specific issues in lisp. seriously. write the asm.
phf: write whole thing in c?
asciilifeform: pretty much every c proggy >2k or so lines is similarly unrecoverable 'under the weight.'