ben_vulpes: you could be fielding questions from clients past about why you billed them nine hundred grand for misc. support over the month.
ben_vulpes: look at the upside! think of all the tacks you don't have to chew!
deedbot: zgrep voiced for 30 minutes.
ben_vulpes: unrelatedly, datomic is a *very* spiffy data abstraction layer.
ben_vulpes: proprietary proggy that sits atop your $db of choice
ben_vulpes: effectively a distributed prolog-alike db, to the extent that i can be trusted to call something 'prolog-alike'
ben_vulpes: the answer to 'ever play with symbolics *' is always 'no', dude.
ben_vulpes: not an expert system, not ai, not anything special.
ben_vulpes: simple an EAVT data store with a single threaded writer. caches recent images of the db in peer memory (yes, 'immutable' db).
ben_vulpes: but i don't know enough about elephant to say.
ben_vulpes: seemed very specifically a CLOS data store.
ben_vulpes: mostly, except as trinque discovered when one goes to do joins.
ben_vulpes: because if it cannot support multiple web hosts, it is not a db with which i will fuck.
trinque: tbh I'm willing to consider that if you wrote too many layers of joins and views, you chose the wrong data structure in the beginning
☟︎ ben_vulpes: asciilifeform: yes, that. or your actual logic layer of choice.
ben_vulpes: whatever it is that returns the JSON that sends mircea_popescu scrambling and screaming for the "programmer's view"
phf: Datomic is piece of shit for non-trivial datasets. I don't understand how they could steal so much old lisp tech and still manage to fuck it up
ben_vulpes: i wish i had the historical perspective to even know.
ben_vulpes: mostly, i find rocks of various sharpness and heaviness and use them to beat my local reality into shape.
ben_vulpes: phf: i would like to hear about what kinds of misery you've found in datomic. not to doubt! never to doubt the misery of technology. but out of curiosity, and limits-finding by proxy.
☟︎ ben_vulpes: (and also as previously discussed in logs, the scraps on my table are rarely of "web scale" magnitude)
phf: Sorry I'm on phone and just needed to pipe in. But I picked up datomic on a project as "clojure take on allegro cache" and I was expecting similar performance (tens of millions of records on a reasonable machine)
phf: Needless to say it's nowhere near there. Be prepared to grok both datomic and the underlying store.
ben_vulpes: yeah, i only run on postgres for that precise reason.
phf: Basically the game becomes "how does query translate into PostgreSQL so I can index all the things"
phf: And those proverbial millions of records require really beefy machines. 8gb of ram, fast flash drives, etc
☟︎ ben_vulpes: it is the classic "db that looks good but needs tender loving care from its authors to work well" software trap.
phf: Don't worry you can hire them to consult :)
ben_vulpes: were i not desperately trying to eke my own living out i would most likely be camping outside of the cognitect offices until they took pity upon me and let me sweep the floor.
ben_vulpes: > allows programmers to work directly with objects as if they were in memory while in fact the object data is always stored persistently << stinks of ORM shit-dippery
ben_vulpes: Foo.objects.get(pk=20) or Foo.objects.get(name="Stan")
ben_vulpes: 'tis the root of the success of Rails and friends
trinque: b-but y'know sql is hard and so caked shit atop
ben_vulpes: "you mean i don't have to think about the db!?!?"
gribble: Error: "up" is not a valid command.
deedbot: ascii_deadfiber voiced for 30 minutes.
BingoBoingo: And ascii_deadfiber thinks the middle west is horribru
BingoBoingo: Seriously, you get fiber in the middle west, that connects to hubs in different directions. We discovered redundancy 3 years ago and nao everything is grand.
BingoBoingo: Not all middle west are equal and which node is Bingo's?
ascii_deadfiber: if it's urban centres, i need those like i need second arse drilled.
BingoBoingo: Turned out box rented back then was crap still running node at home and not advertising, programable version string for reason
BingoBoingo: And no, the good fiber in the midwest is in exurbs. Usually connects to nearest urban center and a further urban center because
deedbot: dotblank voiced for 30 minutes.
deedbot: fromphuctor__ voiced for 30 minutes.
hanbot: not surprisingly the top referer is trilema at ~~1000, followed by qntra at ~~600 and phuctor at ~~80. << wait, qntra's doing 60% of trilema's load *without* tits? very impressive.
deedbot: fromphuctor__ voiced for 30 minutes.
mircea_popescu: he even has a gavinesque retarded face, also. what's with all these mongoloid cheeks everywhere in the dumbass "computing co-manatee" ?
mircea_popescu: and holy shit does twitter load a lot of spurious junk.
mircea_popescu: actually, lettuce make that url trend for "Hanno Böck caught lying", it'll no doubt do wonders for his career.
a111: Logged on 2016-05-02 23:56 asciilifeform: incidentally, didn't the derp already have a 'we did ALL the keyz' post LAST may ?
mircea_popescu: poor hanno boeck & assorted shitbag parade, they don't even run systems with native bignum support, wtf are they gonna do now! 6723834688378347131962599764946917095897099 dun fit in javascript calculator mang, wat do!
a111: Logged on 2016-05-03 00:31 asciilifeform: finally the ancient prophecy coming true, a genuine merger of software and conventional voodoo !
mircea_popescu: historically, stonemasons had a better time of it working for temples than gaols.
a111: Logged on 2016-05-03 00:45 asciilifeform: like 'elephant' ?
a111: Logged on 2016-05-03 00:47 trinque: tbh I'm willing to consider that if you wrote too many layers of joins and views, you chose the wrong data structure in the beginning
mircea_popescu:
http://btcbase.org/log/2016-05-03#1461614 << ianacs, but : fwis the way all these "modern" computing stacks work, is they are made by people with absolutely no system design training, competence or understanding. as a result, they end up hand-crafting the tail. the ready equivalent would be a "new, modern" game of chess, where as the game progresses more and more moves have to be verified against a rulebook ; and then against
☝︎☟︎ a111: Logged on 2016-05-03 00:52 ben_vulpes: phf: i would like to hear about what kinds of misery you've found in datomic. not to doubt! never to doubt the misery of technology. but out of curiosity, and limits-finding by proxy.
mircea_popescu: an expert rabbi, and finally verification be damned, the MEANING of moves can't even be extracted anymore. all this because instead of getting very clearly fundamental principles and sticking to them, they followed the "crazy shit stick on all sides of rifle" approach, with the predictable (and predicted) result that "you fuck it, it gets heavy, and you still don't hit large side of barn". so that's exactly what happens : ea
mircea_popescu: rly game looks spiffy (on all of them!), as the data moves from kb to mb and the node count goes up and on and on, the thing (all of them!) collapses in a pile of uncaught, "nobodycouldhavepredicted" complexity explosions. and all this because idiot/voodoo expert came up with the brilliant idea of "hey, we make robot for your nuclear plant, and if it drops the pu brick you just pick it up by hand and put it back in the hopper
mircea_popescu: ". like it's fucking 1700 all over again or some shit, and they're refugees from the Alchemics & Perpetuum Mobile society.
☟︎ a111: Logged on 2016-05-03 01:01 phf: And those proverbial millions of records require really beefy machines. 8gb of ram, fast flash drives, etc
mircea_popescu: lmao "my vodka identifies as water - therefore it is good for me!"
a111: Logged on 2016-05-03 02:20 ascii_deadfiber: snow, in motherfucking may.
a111: Logged on 2016-05-03 02:26 ascii_deadfiber: you can't even ~get~ real net in midwest.
mircea_popescu: Apocalyptic ^ i propose as best candidate for m-r testing.
mircea_popescu: incidentally, anyone have a general number field sieve implementation in anything ?
mircea_popescu: apparently
http://cado-nfs.gforge.inria.fr/ is what bernstein uses. except "Required software tools [...] GNU make and CMake (2.6.3 or later) for building (CMake is installed on the fly if missing. This feature requires an Internet connection.)" and then Connecting to cmake.org|66.194.253.19|:443... connected. ERROR: certificate common name `*.kitware.com' doesn't match requested host name `cmake.org'.
mircea_popescu: herpitty derp, the world works like it's on fucking fire. anyway, that exhausts my patience.
deedbot: ascii_deadfiber voiced for 30 minutes.
mircea_popescu: dun worry bout it, he won't be able to outlive it no matter what else he does. from experience half a decade later or w/e he'll start whining about how it should be deleted.
ascii_deadfiber: just crapping out the oodles of small primes a typical int is divisible by
mircea_popescu: $v C7EC2E90D0FCDCE3BF39AC7C617BC576AAB197A30624EFE862987F72F45BD8A7
deedbot: mircea_popescu rated hcb 1 << Euloran
a111: Logged on 2016-05-03 11:58 ascii_deadfiber: he's the 'replacement story' tho
BingoBoingo: <ascii_deadfiber> he's the 'replacement story' tho << No, hoaxtoshi and nao petrocheese stockpiles are
ascii_deadfiber: in that the shitcocoon they spray around everything we do is quite opaque, even to thinking folk
ascii_deadfiber: witness that ~nobody walks in asking after trb, for instance.
ascii_deadfiber: shit can be drilled through, but how does one know where to drill..
BingoBoingo: I dunno, the cheese is roughly as equal as some "Americans", but fxing
a111: Logged on 2016-05-03 08:30 mircea_popescu:
http://btcbase.org/log/2016-05-03#1461614 << ianacs, but : fwis the way all these "modern" computing stacks work, is they are made by people with absolutely no system design training, competence or understanding. as a result, they end up hand-crafting the tail. the ready equivalent would be a "new, modern" game of chess, where as the game progresses more and more moves have to be verified against a rulebook ; and then against
BingoBoingo: The one at the top that isn't struck through is right, I the struck through I forgot to replace a value from last month.
a111: Logged on 2016-05-03 08:30 mircea_popescu: ". like it's fucking 1700 all over again or some shit, and they're refugees from the Alchemics & Perpetuum Mobile society.
a111: Logged on 2016-05-03 10:28 mircea_popescu: in vaguely related lulz,
http://www.javascripter.net/math/primes/millerrabinbug-bigint54.htm as well as "However, if you try to pass an odd number n greater than 2^53 the test will not work because the argument would actually turn out to be some other number approximately equal to the desired odd number: JavaScript/IEEE754 cannot exactly represent odd numbers that large!"
ascii_deadfiber: if the only implementation of something were in js, it hasn't actually been implemented yet.
ascii_deadfiber: i do suspect that the enemy knows some factoid re prime distribution that makes factoring randints easier
mircea_popescu: ascii_deadfiber and while on this topic : it beggars belief that bignum-as-linked-list-of-any-size-up-to-avail-ram is not a fucking standard in gcc in 2016.
mircea_popescu: seriously, i gotta say ANYTHING more than "bigint" to get this ?
ascii_deadfiber: mircea_popescu: there is no easy answer to ~how to represent~ them
mircea_popescu: who the fuck things these shitboxes are even computers when i get n digits eval to 0, NaN and whatever other bullshit ?!
ascii_deadfiber: if you want anything like fast arithmetic, you're stuck with fixints, and even float barf, aha
mircea_popescu: mp lemma : any computing environment where bigints are not native is ipso facto parsed, not compiled.
mircea_popescu: the compiler is more than happy to make a linked list of fixints and do the carryovers as required.
mircea_popescu: which is how longhand multiplication even fucking works, on the basis of strings-and-regexp pretty much.
mircea_popescu: there is absolutely no reason, nor can there be any reason, c compiler can't take two strings, such as the decameron and the collected works of shakespeare, and multiply them together, as fucking numbers, and spit out the result. none.
mircea_popescu: and if it won't do this, the fact that it graciously doesn't require me to write my own asm jumps or whatnot is irrelevant already.
mircea_popescu: this shit is not a compiler. it's a low effort parser with delusions and pretensions.
mircea_popescu: !v 8DD5AEEC6512709A00BA6BD58A95158D7E9499903F24C5C129C6FD7AE049D7A6
gribble: Error: "v" is not a valid command.
mircea_popescu: $v 8DD5AEEC6512709A00BA6BD58A95158D7E9499903F24C5C129C6FD7AE049D7A6
deedbot: fromphuctor voiced for 30 minutes.
fromphuctor: yesterday i suggested to play with ssh-keyscan
fromphuctor: today i am wondering if you ever compared the keys to see if any clone is found
mircea_popescu: you seriously should make a proper nick and register your pgp with deedbot.
fromphuctor: in particular i have the SUSPECT that router from juniper and cisco might build random keys that are not random at all
phf: well, perhaps he should load it into datomic
mircea_popescu: fromphuctor i dunno if you've seen the logs, but the idea to add ssh keyscan was accepted upon discussion.
a111: Logged on 2016-05-02 19:14 mircea_popescu: asciilifeform but he has a point ; once we get conversion going running a ssh-keyscan werker is good idea.
mircea_popescu: you're also very likely right re nat router keys. junk of the junk.
deedbot: fromphuctor_ voiced for 30 minutes.
deedbot: fromphuctor__ voiced for 30 minutes.
fromphuctor: if you want visibility you should try openssl, openssh, libressl, openbsd ssh keys
fromphuctor: at the moment , i tried, they are not very interested in your reserach
mircea_popescu: everyone gotta eat, and the general majority of these people eats at usg.*
fromphuctor: because they think their implementation is fine
phf: mircea_popescu: just irked, there's a rare "human record" kind of data that ben_vulpes deals with that exceeds 8gb. why don't you taunt us with your hadron collider data dumps some more?
mircea_popescu: so no, i very much expect they "wouldn't be interested".
mircea_popescu: hey, this is what the internet does & is for my dear phf : taking a very reasonable statement way the fuck out of context.
fromphuctor: another thing to test when you collect ssh keys is for duplicate
mircea_popescu: fromphuctor you know, you could in principle write all this stuff yourself. if it's any good odds are it'll get accepted/used. if not, odds are you'll get a very scathing review from alf making you doubt your sexuality and other life choices.
mircea_popescu: also, did i mention lisp isn't used in production anywhere today ? wouldn't want to skip a day :D
deedbot: fromphuctor voiced for 30 minutes.
phf: no wonder all the talk of "i'll pay out of pocket!!"
mircea_popescu: (anyway no, it's by no means expensive for the specs, we got a nice one-offsie on account of other business etc)
mircea_popescu: note that trilema blows the openssl attempt at legitimization way the fuck out of the water.
Apocalyptic: 12:18 <+mircea_popescu> Apocalyptic ^ i propose as best candidate for m-r testing. // I'm already working on a more approachable one, also I don't know what's with you and m-r but I don't think m-r is a good factorization algo, I was and am running p-1 , p+1 and ECM
Apocalyptic: "14:16 <+mircea_popescu> i recall him trying one" // and succeeded, see ascii's rating, afaik he included the log line where I posted all the factors
Apocalyptic: th modulus residue I'm working on has ~ 260 digits and it seems the remaining primes are pretty big, +35/40 digits from what I see
mircea_popescu: Apocalyptic you know, i vaguely recall there was a challenge for the "largest factor found in factorization" or somesuch ?
mircea_popescu: dunno what the record is by now, but at some point it was like 60 digits
diana_coman: hm, trying to but waiting on wotpaste.cascadianhacker.com
mircea_popescu: Apocalyptic nice mechanism for checking factoring algo claims he came up with too.
mircea_popescu: incidentally - perhaps a good idea would be a tmsr factorization program. something like seti@home, that people can just run on spare machines if they feel like.
Apocalyptic: indeed nice checking algo, sorta switching the usual role of d and e
Apocalyptic: mircea_popescu: that would be a great idea indeed, in the meantime if anyone feels like helping me factor my modulus residue let me know
Apocalyptic: it doesn't have to be coordinated in any way since ECM runs are totally independant, just willing to dedicate some cores
Apocalyptic: mircea_popescu: download ecm from inria, compile it, run ECM with the residue I supply
mircea_popescu: incidentally, can you be induced to look into that cado thing ? it has some sort of multimachine-over-ssh thing
Apocalyptic: i'm not really interested in cado or setting up such a thing
Apocalyptic: a pitty they don't sign source archives though...
Apocalyptic: if trying ECM don't bother picking B1 under 11e6
ben_vulpes: diana_coman: timeout? works fine from here
diana_coman: ben_vulpes, yes, it timed out in the end, no idea why
phf: maybe needs a 250gb machine
deedbot: ascii_butugychag voiced for 30 minutes.
mircea_popescu: in other news, phf's computer so fat, its foo's out the bar!
mircea_popescu: but the cool part is : derp is essentially bringing a wot-argument-by-another-name.
Apocalyptic: "If you look closely, all of the keys shown on the website are only self signed." // clearly he has looked closely
mircea_popescu: in other lulz : undefined reference to `__sync_add_and_fetch_8'.
mircea_popescu: ascii_butugychag a coupla hours ago i made a cursory attempt to install a factorizer lib (linked in logs). i threw up minutes later because obviously it doens't work. put out a general call, slavegirl picked up the task, installed python 4.4. wrangled for a while with broken cmake settings (pro tip : cmake is known to not correctly identifgy your python, workable hack in this case is to disable the test and manually fix the f
mircea_popescu: ;;google undefined reference to `__sync_add_and_fetch_8'
mircea_popescu: EVERYTHING. from rtorrent to juce to what have you. let's break all the things!
phf: i will venture "wrong gcc version/flags" on account of that _8 there
mircea_popescu: isn't it the case that 99% of compile failures get called "wrong gcc version/flags" ? as if that does anything ?
ascii_butugychag: mircea_popescu: the 'consensus shapers' are all working the angle of 'these terrorists planted the keyz'
mircea_popescu: ascii_butugychag which is proof positive nsa did in fact plant them.
mircea_popescu: the notion that usg's word vs tmsr's word has a chance in hell is lulzy.
mircea_popescu: you clearly spend too much time on reddit/tardspedia/etc.
Apocalyptic: "14:18 <+ascii_deadfiber> i do suspect that the enemy knows some factoid re prime distribution that makes factoring randints easier
Apocalyptic: " // I would be curious as to why you suspect such a thing
ascii_butugychag: i was hired, for instance, partly on the basis of my ancient shithub acct
ascii_butugychag: Apocalyptic: because of the 'deniable' sabotage of replacing moduli with random ints
mircea_popescu: ascii_butugychag in that case, let's put it in storage i guess.
deedbot: dotblank voiced for 30 minutes.
ascii_butugychag: mircea_popescu: my point was that in nazi land i get ranked as simply a very lazy nazi who spends time on ???? rather than properly drilling with the other ss men
ascii_butugychag: it is, perhaps, why i'm still alive - if 'accidented', then, e.g., trb, suddenly becomes more interesting to a buncha folk
ascii_butugychag: 'iz zis nazi lande sooo goood ?? we would leeeeeve it, if we could!' -- 'donald duck in naziland'
deedbot: fromphuctor voiced for 30 minutes.
deedbot: ascii_butugychag voiced for 30 minutes.
deedbot: meuh voiced for 30 minutes.
ascii_butugychag: i suspect i've already been killed and now in a hell of sorts.
ascii_butugychag: speaking of, i'm told that latest recycled shitoshi is now being paraded on american tv
trinque: read an article yesterday about "hooray, $shitoshi can help resolve the 'block size debate'" etc
ascii_butugychag: and the man has to be a lunatic, who would sign up for this? it is like signing up as court alchemist in 17th c
ascii_butugychag: i find it strange that they wouldn't go for broke and break the idiot's dsa key for the occasion
trinque: ascii_butugychag: the guy's going to drop fucking NSAcoin
meuh: It doesn't feel right
deedbot: ascii_butugychag voiced for 30 minutes.
mircea_popescu: this is good for tracking. it's also normal in case you sign other people's keys.
meuh: mircea_popescu: think i got it, the site reports "Modulus Seen Elsewhere! Please make sure these are yours:" for signatures made on other keys, and I have to check I've made those signatures with one of my key
mircea_popescu: ascii_butugychag it's some anon russian guy. if you think rando usg endowment is a better shot...
ascii_butugychag: meuh, mircea_popescu : this happens IF AND ONLY IF the modulus recurrs in its entirety inside another key.
deedbot: fromphuctor__ voiced for 30 minutes.
ascii_butugychag: mircea_popescu: the fella who unplugged Dulap-1 every weekened was also rando ru guy.
ascii_butugychag: negative zero integers) was reported to OpenSSL by Huzaifa Sidhpurwala
meuh: ascii_butugychag, mircea_popescu: you make me feel rather uncomfortable, it puzzle me how common modulus can appears in those keys:
deedbot: meuh voiced for 30 minutes.
mircea_popescu: ascii_butugychag the particular link noob dropped IS ALREADY IN THE LOG.
meuh: ascii_butugychag, mircea_popescu: you make me feel rather uncomfortable, it puzzle me how common modulus can appears in those keys:
meuh: coz' they have the same modulus ?
ascii_butugychag: say you, like a n00b might, sign the message 'yes, meet me at midnight on the crossroad'
ascii_butugychag: then someone else takes your modulus, and welds own username string to it
ascii_butugychag: there is no way to watch for mischief where modulus is lifted WHOLESALE, without also listing a boatload of this rubbish, unfortunately.
meuh: OK, I need to investigate more (going to open again that OpenPGP RFCs) and don't waste your time. I originally tought there was a bug in the interface that was incorrectly reporting my keys as dupes because I've made a single armored export of the public key and feeds this as a single blob to the submission page
meuh: ascii_butugychag: thanks
mircea_popescu: so before it was "error handling request" ; then it was cycling indefinitely in "queue". now it's "network error".
ascii_butugychag: mircea_popescu: and at any rate 'they planted'em in 2013! who you gonna believe, obamitler or some ru rando'
mircea_popescu: ascii_butugychag seeing how sks itself reports them from 2002 to whatever, 2012, i dunno. lol.
mircea_popescu: because, much like in the case of the miners getting caugvht with pants down, there is THIS interval right here.
mircea_popescu: and no matter what "science" derps might come up with, here it is and here it lives.
mircea_popescu: stupid ho screaming "nonono" with hands stuffed in ears is still getting raped, just like any other.
mircea_popescu kinda sees this future, derpy eggs on wheels going around smouldering ruins that used to be us towns, "STOP! Or I'll call you a bad name!"
ascii_butugychag: 'ASN1 Strings that are over 1024 bytes can cause an overread in applications
deedbot: Echoplex voiced for 30 minutes.
trinque: ah that didn't work; I guess that links to the "do you want to archive" page
deedbot: ascii_butugychag voiced for 30 minutes.
mircea_popescu: trinque it hasta be clicked, because otherwise you realise, botted into the ground.
a111: Logged on 2016-05-03 00:42 asciilifeform: ben_vulpes: ever play with symbolics 'joshua' ?
trinque: ah life at the end of the world
deedbot: ascii_butugychag voiced for 30 minutes.
mircea_popescu: if anyone wants to get the archive, base64 and deed it...
a111: Logged on 2016-05-03 22:08 ascii_butugychag: circa 1994.
mircea_popescu: "this doesn't solve the problem of the sun looking at me funny" "this is not a problem" "am i thick or what ?!" "no, you're just the sort of fellow that's decided afore the fact that the sun can look at you funny and this matters."
mircea_popescu: such problems can only be solved by the problematizer and for himself.
mircea_popescu: there's a difference between "why should i suck your cock" "because it's here" and "why should i suck your cock, because he says i'm a slut ???".
mod6: i had the same reaction mp
shinohai: btcbase is fast becoming my go-to site >.<
mircea_popescu: we were kinda in the middle of things with that when interrupted.
shinohai: $v FFE78FDD065EC28BAE4EE3B39F5E3C8A52BEBF5DDE280133786FB5E4082EF291
deedbot: shinohai updated rating of phf from 1 to 2 << btcbase.org
mod6: the nice part is that i can just click and see the vpatch too
mod6: how does one load the vpatch in there in the first place? just submit to phf?