Hide Idle (>14 d.) Chans


← 2017-07-03 | 2017-07-05 →
mircea_popescu: http://btcbase.org/log/2017-07-04#1678949 << "mentally unstable secret service transgender charlie shoots president. VIOLENT COMPUTER GAMES MAY HAVE BEEN INVOLVED!!!!"
a111: Logged on 2017-07-04 03:22 BingoBoingo: January 2019 Trump will likely be Nixon proof. Pantsuit will have to resort to Kennedy solutions.
BingoBoingo: !~ticker --market all
jhvh1: BingoBoingo: Bitstamp BTCUSD last: 2612.22, vol: 13002.43778694 | BTC-E BTCUSD last: 2503.826, vol: 6281.94379 | Bitfinex BTCUSD last: 2602.5, vol: 18638.3081568 | BTCChina BTCUSD last: 2691.995072, vol: 12614.50490000 | Kraken BTCUSD last: 2619.902, vol: 6754.02624087 | Volume-weighted last average: 2615.64317885
BingoBoingo: This news is good for OMG CRASHING!!!
deedbot: http://phuctor.nosuchlabs.com/gpgkey/E51426CFDA85B249F6D15DC1C02D7D376AC72762320584257E84D9FDE6DC1393 << Recent Phuctorings. - Phuctored: 1379...0579 divides RSA Moduli belonging to '89.250.54.58 (ssh-rsa key from 89.250.54.58 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt>; ' (Unknown FI)
deedbot: http://phuctor.nosuchlabs.com/gpgkey/E51426CFDA85B249F6D15DC1C02D7D376AC72762320584257E84D9FDE6DC1393 << Recent Phuctorings. - Phuctored: 1623...1269 divides RSA Moduli belonging to '89.250.54.58 (ssh-rsa key from 89.250.54.58 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt>; ' (Unknown FI)
deedbot: accepted: 1
deedbot: accepted: 1
ben_vulpes: BingoBoingo: was otherwise occupied; twas a teep just the teep
hanbot: any spanish speakers here have a decent translation for "match"? to place in a set, not pairing, not equaling, not anything else online dictionaries'd like to synonymize me with.
hanbot: ^ shinohai ?
ben_vulpes: not i
mod6: nice sideburns
mircea_popescu: hanbot partido.
mircea_popescu: ah im sorry, to match as in a set. conjunto fambly should suit you.
hanbot: conjuuuunto. tyty!
shinohai: !~later tell hanbot Ah sorry I missed your message, I was out at that time cojunto-ing with local girls. ;)
jhvh1: shinohai: The operation succeeded.
shinohai: !~ticker --market all
jhvh1: shinohai: Bitstamp BTCUSD last: 2621.86, vol: 11540.22746202 | BTC-E BTCUSD last: 2537.349, vol: 6658.60169 | Bitfinex BTCUSD last: 2607.0, vol: 18107.17109604 | BTCChina BTCUSD last: 2668.04853, vol: 8983.97320000 | Kraken BTCUSD last: 2619.031, vol: 6922.33937901 | Volume-weighted last average: 2613.50135031
phf: i wrote a hilariously bad common lisp version of fhf
phf: ben_vulpes: i'm pretty sure your type annotations are doing nothing useful. cursory look i think only fixnum stuff is taking effect, but that's such a small overhead
phf: i'm pretty sure biggest overhead is in using adjustable arrays, because you lose out on using sbit instead of bit. sbit expects (simple-array bit (*)) and apart from doing basic type check it has very little overhead.
phf: and the other issue with adjustable arrays is that you're doing a lot of expands (and that's what mp's fhf does), you're generating n^2 garbage
phf: *if you're doing
phf: (funny thing re sbit vs bit, sbit generates immediate assembly, where's bit, in sbcl, calls out to a function named SB-KERNEL:HAIRY-DATA-VECTOR-REF. nuff said)
phf: sbcl is the original "modern"11! lisp. it gives "style warnings" on my code, but completely ignores type errors in ben_vulpes's code (compiling same code in cmucl gives a lot of warnings about the types being all kinds of confused)
asciilifeform: now if cmucl actually worked, rather than compiling to crasholade...
phf: yeah, it would've been neat if people carefully improved existing quality system instead of modernizing, alas. i hear wayland also works!!1
asciilifeform: if it worked ( rather than tried to redefine 'work' ) ..
phf: fhf implementation http://paste.lisp.org/display/350018, this one just races against ben_vulpes's, i'll see if i can beat go version once i get to building go environment..
asciilifeform: phf: there's a couple of order-of-magnitude imho obvious optimizations, but i dun wanna ruin the fun
asciilifeform: but for some reason nobody said anything yet
asciilifeform: one hint -- look at 'screw' as a modular congruence
phf: yeah, this version is straight spec implementation (in fact if you search for (fhf mp ...) that part should directly correspond to spec wording). only optimization i did was to keep the S reallocations down
asciilifeform: other hint -- stop doing mem-access-per-bitflip omfg
phf: right, but that would "break" the spec to code correspondence
asciilifeform: phf: what does your rewind do if pos is 0 ?
phf: part of spec as far as i understand it, that the (fhf mp ...) part is a canonical, but not the only implementation. so yeah you could write optimal version of mp's fhf, but it's not going to speak the same dictionary, and you won't be necessarily able to apply same optimizations to other (fhf ...). (or i guess you can, but require more cleverness than just etc.)
phf: gtfo.
phf: i mean, there's a bunch of macros that share state through referencing same symbols. if i were to post this to #lisp they would lynch me
asciilifeform: reminds me of my 'nock' compiler
phf: oh, that's actually good point re rewind, that's in the spec "except in the case our position is already 0."
phf: man, dusseldorf airport soundtrack is like the best of the 90s that nobody listens to anymore, but used to back in the day. like they are playing bryan adams's "i do it for you" which i remember serenading to some girl. makes me heart tingle it does.
asciilifeform: phf: hm, 'woods' turns out was in de ?
phf: this is new woods
asciilifeform: dun need the <= phf
asciilifeform: = will do...
phf: that's true, and is closer to wording..
asciilifeform: btw this proggy is nostalgiatronic in that i nao return in my head to ye olde naggumwank over 'if*'
asciilifeform: ( he had an eternal and bitter feud with the author )
asciilifeform: unrelatedly, i still have not thought of a proof that mphash terminates on all inputs.
asciilifeform: ( and currently suspect that you can construct one where it doesn't )
phf: i wouldn't call it bitter, at least he didn't think of if* proponents as complete retards.
phf: i wouldn't use it in my own code, but there are occasional cases where it "makes sense", particularly when you're writing potato code
phf: hmm, never mind his later rants on the subject contain a lot more vitriol
asciilifeform: phf: potato?
phf: but i use it to refer to intentionally or unintentionally naive code. like the kind of stuff you would write in BASIC.
a111: Logged on 2017-07-04 13:52 phf: (funny thing re sbit vs bit, sbit generates immediate assembly, where's bit, in sbcl, calls out to a function named SB-KERNEL:HAIRY-DATA-VECTOR-REF. nuff said)
mircea_popescu: truth be told fhf was designed by a coder hater, and it shows. exactly typical ops a sane person would never even contemplate.
asciilifeform: how's that
mircea_popescu: "enlarge datastruct one bit at a time ? because your mother fed you artichokes whole or why!"
mircea_popescu: "just tell me how large it ends up and stop wasting my time! WHAT THE FUCK IT CAN'T BE KNOWN IN ADVANCE!"
mircea_popescu: and then there's the "screw". as phf sez, nuff said.
mircea_popescu: phf whereas is not where's, etymologically or otherwise. i dunno if you know that.
asciilifeform: the 'potentially infinite mem' and 'potentially not terminates' is the boojum, not exotic bit ops
mircea_popescu: asciilifeform mp-fhf terminates, and in finite mem. of course, other fhfs might not.
asciilifeform: mircea_popescu: picture 'i'd luvvv to verify this tx, but gotta buy some ram, and for 3rd time this week'
asciilifeform: mircea_popescu: didja prove ?
mircea_popescu: i did not prove ;/
asciilifeform: that it terminates
mircea_popescu: i am surrounded by looseleaf to the degree girl walked in said "you remind me a lot of mr happy" "whassat ?" "pet hamster i had when i was 5"
mircea_popescu: but no prove.
mircea_popescu: nevertheless, the reason it has those bojums flows from the atomic ops, not gift from god.
mircea_popescu: http://btcbase.org/log/2017-07-04#1679054 << this is a fine approach. "here's the per spec impl, here's the optimized impl. you can verify they agree wrt results ; and you can trivially verify the former is spec-accurate."
a111: Logged on 2017-07-04 14:15 phf: part of spec as far as i understand it, that the (fhf mp ...) part is a canonical, but not the only implementation. so yeah you could write optimal version of mp's fhf, but it's not going to speak the same dictionary, and you won't be necessarily able to apply same optimizations to other (fhf ...). (or i guess you can, but require more cleverness than just etc.)
mircea_popescu: in fact specified as opposed to farted code has this property, that you write two programs per spec like that. hence "agile" ie "code is the spec" ie "please shoot me" modernity.
mircea_popescu: saves on paper! envirowhiny!
jurov: http://btcbase.org/log/2017-07-04#1678979 lol what do you smoke? what about numerous dividend mispayments, some resulting in a rollback?
a111: Logged on 2017-07-04 03:43 asciilifeform: mircea_popescu: mpex afaik never reverted trades either
mircea_popescu: this is related ?
jurov: dividends are not transactions?
mircea_popescu: well depends what you mean and what you're talking about. some people call their http "transactions", and the "road ahead is clear" as they say to next propose mpex css ain't good enough to be nasdaq.
mircea_popescu: but the discussion here isn't "naive perception of item x as compared to naive perception of item y". that'll be biased by the direct interaction. the discussion is "industrial item x vs y".
jurov: "alrighty"
mircea_popescu: if every time a broker advanced cash to a customer you took out nasdaq and shot it, it'd live microseconds each iteration.
jurov: Yet it outlived bitbet.
mircea_popescu: jurov nasdaq solution to bitbet problem is to take kako out and shoot him. that i'm a civilised fellow is not something you can oppose to me.
mircea_popescu: on the contrary, in this context it is my merit and you're welcome to praise.
mircea_popescu: in the sense that if you discuss it at all, all you may say is "mp is so fucking great, he put personal freedom above corporate continuance."
mircea_popescu: yes this is a choice i've made repeatedly and threaten to keep making. it also happens to be fundamental to republican vs you know, imperial outlook.
phf: in fact http://btcbase.org/log/2017-03-09#1623811 :/ i was trying to remember recently but failed what it is that i was consistently using wrong
a111: Logged on 2017-03-09 18:06 mircea_popescu: phf you know it's whereas not where's.
a111: Logged on 2017-07-04 14:12 asciilifeform: one hint -- look at 'screw' as a modular congruence
mircea_popescu: phf what it was that i consistently ; not what it is that i consistently was :D
mircea_popescu: russki! english has backwards tenses.
phf: also articles are optional
mircea_popescu: this is a very fine example, incidentally. it's in my mind right there with the oxford comma ("i had eggs, salmon and biscuits" interpreted as the speaker telling salmon and biscuits as an item that he had eggs) as anglo-antilogic in my head.
mircea_popescu: which is why qntra style says "logical comma always". as opposed to oxfordian.
mircea_popescu: anyway, it is logical to have the error permanent "what it is", and the self-reference tensed "i was". rather than pretend the error has no further substance than its past occurence, whereas the id is eternal and so therefore is.
mircea_popescu: yet in english it's how they do it, and THIS is what they mean when they say the language it speaks will tell you a lot of a people.
mircea_popescu: ie, there is no such thing possible as "a christian english-as-single-language speaker".
a111: Logged on 2017-07-04 14:46 asciilifeform: ( and currently suspect that you can construct one where it doesn't )
mircea_popescu: http://btcbase.org/log/2017-07-04#1679079 << i thought samagon was made out of chicken feathers not potatoes.
a111: Logged on 2017-07-04 14:52 phf: but i use it to refer to intentionally or unintentionally naive code. like the kind of stuff you would write in BASIC.
phf: irishmen came up with that term, it's all about potatoes with those people
mircea_popescu: "the great potato famine'd have been lots worse were it made out of wheat instead"
asciilifeform: 'or if stalin did it and not brit crown'
mircea_popescu: http://btcbase.org/log/2017-07-04#1679092 << my current thinking is, the only way to resolve this within a lifetime is to see if a mpfhf cycle could be constructed. if it can, solved, if it provably can't then it is time&space finite, even as a statistical ceiling argument.
a111: Logged on 2017-07-04 14:57 asciilifeform: mircea_popescu: didja prove ?
asciilifeform: mircea_popescu: well yes
asciilifeform: how else.
mircea_popescu: asciilifeform there are other ways. they work better on the exact type of hash fhf isn't.
mircea_popescu: haven't yet managed to pin down what teh absolute requirement list for a cycle is though.
asciilifeform: not necessarily cycle tho. if i can construct a 'for (i=0; i<maxint; i++) {}' then just as good
asciilifeform: in terms of 'won't terminate before machine crumbles to dust'
mircea_popescu: this is a point in practice, but not relevant re the finite-ness.
asciilifeform: recall the 'primes is in p' thread
asciilifeform: ( buncha d00dz published a polynomial time primality test in early 2000s. but try and use it. )
asciilifeform: *deterministic primality test
mircea_popescu: kinda why we're even doing all the prof work on it, see that crumble to dust bit first.
a111: Logged on 2017-07-04 14:09 phf: fhf implementation http://paste.lisp.org/display/350018, this one just races against ben_vulpes's, i'll see if i can beat go version once i get to building go environment..
mircea_popescu: asciilifeform god damned. check this out : 1. only way to cycle is through the rewind branch. 2. rewind branch only rewinds one, so to cycle must be run through every time. 3. the branch with a rewind always does a S in R screw, and a R flip. for cycle to happen therefore a) message bit must be 0 and b) S and R must be in such a state that screw and bitflip produce same bit in R.
asciilifeform: well yes
asciilifeform: how else
mircea_popescu: asciilifeform BUT
BingoBoingo: In other delusionariums, guide for ES(ingle)L Chinese Miners; A FANFIC: https://archive.is/tY0Ya
BingoBoingo: Much better title than their bullshit
mircea_popescu: eh bullshit, it crumbled.
asciilifeform: oh ffs BingoBoingo , 'bitcoin magazine' ?!
BingoBoingo: Hey, I labeled it fanfic
phf: ben_vulpes: http://paste.lisp.org/display/350022 current version
BingoBoingo: "OMG HARD DATES LISTED!!!" Ought to make the Shillbert fail more sweet than the GaVERToominium failures so far
phf: i threw in a handful of type declarations (which eliminated generic-+ calls), but that as expected shaved less than half a second
asciilifeform: phf: i can see argument for 100% unoptimized 'potato'. but if yer gonna optimize, may as well do it properly neh
mircea_popescu: asciilifeform no. no. it's there. asciilifeform : "This operation consists of taking the bit count of either S or R, iterating over that value, at each step multiplying the iterator with the current position in M" << screw will always be m-position divisible. |If the bit in R found at the position equal to the remainder of the division of our position in M by the size of R is 0" << bit deciding rewind is ALWAYS touched by th
mircea_popescu: e screw part of that branch. so we need a M/S/R that takes the 1, flips it to 0, so then the rewind flips it back to 1 again and so it couldf cycle.
asciilifeform: mircea_popescu: aha
mircea_popescu: BingoBoingo they still wanking with bitpython v 3.0 or w/e it is ?
asciilifeform: mircea_popescu: i admit , i half-expected to have mircea_popescu pull the noncycling proof out of his hat, 'aha, someone finally thought to ask..'
phf: asciilifeform: there's different kinds of optimizations. you can nail down the types and get a complete type signature of your code. i don't think there's any shame in it
asciilifeform: phf: i did not say 'shame' lel
mircea_popescu: gah this is fucking frustrating
asciilifeform bbl -- meat.
BingoBoingo: mircea_popescu: Haven't fed it through pantsuit.txt parser yet, will after sleeping
mircea_popescu: ok i fucking got it!
mircea_popescu: the requirements for cycle : a M/S/R that takes the 1, flips it to 0, so then the rewind flips it back to 1 again and so it could cycle.
mircea_popescu: these requirements can never be satisfied, because the branch in question EXPANDS s, and therefore it will never be the case that the position modulo R of S will be equal to the position modulo R of S+1.
mircea_popescu: asciilifeform ^
deedbot: http://phuctor.nosuchlabs.com/gpgkey/61C5D4EF56CBBFBAFDE33BD7F870253F9F5898F82CD8BF8E6954CCDF8A70245E << Recent Phuctorings. - Phuctored: 1501...5889 divides RSA Moduli belonging to '91.90.148.18 (ssh-rsa key from 91.90.148.18 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt>; ' (server4718.mivitec.net. DE)
deedbot: http://phuctor.nosuchlabs.com/gpgkey/61C5D4EF56CBBFBAFDE33BD7F870253F9F5898F82CD8BF8E6954CCDF8A70245E << Recent Phuctorings. - Phuctored: 1764...8029 divides RSA Moduli belonging to '91.90.148.18 (ssh-rsa key from 91.90.148.18 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt>; ' (server4718.mivitec.net. DE)
deedbot: http://trilema.com/2017/no-such-labs-snsa-june-2017-statement/ << Trilema - No Such lAbs (S.NSA), June 2017 Statement
mod6: Good to hear that the lost units came back.
ben_vulpes: http://btcbase.org/log/2017-07-04#1679046 << should beat that as well yeah
a111: Logged on 2017-07-04 14:09 phf: fhf implementation http://paste.lisp.org/display/350018, this one just races against ben_vulpes's, i'll see if i can beat go version once i get to building go environment..
ben_vulpes: phf: illuminating, ty
mircea_popescu: in other usistani ideas : register "techrite" as a brand. then people can say "he's got the techrite" about your users.
BingoBoingo: !~ticker --market all
jhvh1: BingoBoingo: Bitstamp BTCUSD last: 2586.4, vol: 9738.16797447 | BTC-E BTCUSD last: 2509.146, vol: 5957.60415 | Bitfinex BTCUSD last: 2568.8, vol: 15703.98682071 | BTCChina BTCUSD last: 2672.89806, vol: 9377.74330000 | Kraken BTCUSD last: 2588.045, vol: 5887.51865475 | Volume-weighted last average: 2588.20440617
asciilifeform: happy 2015phorqday folx
BingoBoingo: Happy Phorking day to all you motherphorkers!
← 2017-07-03 | 2017-07-05 →