log☇︎
99100+ entries in 0.055s
mircea_popescu: i'd very much rather we go with 1. they booked the chassises at 1k each, tentatively, which is not what's going to happen. bill 120% materials like it seems it's becoming the repuiblican "standard profit" and there we go. neh ?
asciilifeform: i'ma let mircea_popescu say how to crack this nut, my teeth are not up to the job, and pick from his variants, if he dun object
mircea_popescu: but no, the universe 3 whereby "people made a 300 btc donation to unclear future goals" isn't really on the table.
asciilifeform: right; but it oughta be understood that my manufacturing capacity is quite limited, until and unless it somehow permits a full-time asciilifeform .
mircea_popescu: or else universe 2, whereby chassis for pizarro is produced by you as free agent. in this world we have to now explain what nsa actually is, which becomes a rather difficult task, and pizarro will have to sell its bonds on market, to whatever discount.
mircea_popescu: practically the situation here is this : we can either live in universe 1, whereby chassis for pizarro is produced by you for nsa, and sold by nsa to pizarro. in this world nsa stays what it always was, as above, and pizarro can (at practically your option, as i can go either way) pay with undiscounted bonds for cash-equivalent mateirals ;
mircea_popescu: afaik nsa always was exactly what phf is quoted above as having understood it is : the umbrella org for your manufactory.
asciilifeform: ( in case it isn't clear, asciilifeform views s.nsa as in effect a scientific collaboration / partizan front , with mircea_popescu , as asciilifeform does not expect to see any coin from it, i would like to be proven wrong about this, but i don't see selling >50 btc's profit worth of anything, esp not if the climb carries on )
mircea_popescu: and one step down, nsa doesn't even have to permit equity buys, can insist on being paid cash. i was permitting it because i thought it's desirable, but if you don't like it evidently won't happen.
mircea_popescu: i don't think there's any intention of that. i certainly would vote against the measure.
asciilifeform: the 'i dun like the endgame' observation was specifically in re the scenario where s.nsa ends up swallowing pizarro.
asciilifeform: this isn't actually the problem, pizarro ( and other l1 folx ! thank you ! ) supplied asciilifeform in recent times with 100% of the btc-buying opportunities he can digest
mircea_popescu: if you do it as "myself" you get whatever chassis cost in btc in your pocket. if you do it as s.nsa you don't, is the logic ?
mircea_popescu: well how much pizarro has in the way of cash isn't changed by ooooh i see the problem, you hafta recirculate this is it ?
asciilifeform: cuz it doesn't have much in the way of cash.
mircea_popescu: we were doing that as an easement, but it's evidently our option neh ?
mircea_popescu: it doesn't ~have to~ sell for shares. can sell for cash, why not ?
asciilifeform: is even supposing that mircea_popescu doesn't feel like raising his price again ) .
asciilifeform: i will admit however that i dun particularly like the endgame i see in this, where let's say s.nsa sponsors, produces, and sells to pizarro the iron for shares, ends up owning ~100% of pizarro, and asciilifeform doesn't see a dime of profit until ( merely for starters ) he can even afford mpex acct ( which is a major if, rather than a when, it is quite possibly that neither i nor anyone else will ever actually earn 50 btc , and this
mircea_popescu: it can do consultancy / custom jobs for isps/dcs, why the hell not.
mircea_popescu: there's not some rule that all nsa products must be available to the general public.
mircea_popescu: okay, but the larger concern seems to be that you'd rather put all your fabrication under nsa than have to explain to shareholders why not and wherefore in the future.
asciilifeform: no particular problem, aside from the stark contrast between the 100% repeatability and hygiene of , e.g., FG, and the possibly nsa-laden ( i don't design arm crystal, or the pcb, or the switches, etc ) server
mircea_popescu: when you last flew over (on nsa dime) we sold them servers an' nobody died.
mircea_popescu: yes, but what's the significant problem with nsa selling pizarro custom built chassisen ?
asciilifeform: mircea_popescu: idea is that a working ( built out, and populated ) pizarro, could , in principle, eventually free asciilifeform from the unspeakable nonsense on which he burns most of his time currently.
a111: Logged on 2016-01-16 20:58 phf: i find the framing odd, i thought purpose of s.nsa capital was to assist ascii in his b-a aligned, but essentially personal endeavor, which might potentially have a return. somehow it turned into "pay ascii by the hour". if ascii doesn't want to work on cardano, it's not a b-a way to coerce him, but if the project still exists, is being worked on, then money will be spent when there's something to spend it on, and there will be a
mircea_popescu: http://btcbase.org/log/2018-05-16#1814276 << yes, but this encounters the following problem : http://btcbase.org/log/2016-01-16#1373235 ; while obviously this won't go as far as "every idea alf has", nevertheless by the time alf makes objects, and independently from nsa, then what is nsa ? ☝︎☝︎☟︎
a111: Logged on 2018-05-16 13:26 asciilifeform: http://btcbase.org/log/2018-05-16#1814131 << transport for rockchiptron will be considerably cheaper than 2500, as i can't picture needing a whole week, nor 400 in overweight charge for the crates
mircea_popescu: http://btcbase.org/log/2018-05-16#1814268 << i was thinking. cool then, they need every break they can get. ☝︎
mircea_popescu: in "things that willn't have happened 20-30 days from now, say security experts"
a111: Logged on 2018-05-16 10:15 spyked: ftr, I've gathered about 700k keys in the last 3-4 days; if the total number hasn't changed too much from the previous 6.9M, I estimate I should have all of them in 20-30 days from now.
mircea_popescu: (the correct cut would be, obviously, the arab invasion -- but i choose to be this insulting.)
mircea_popescu: but there's no major "let's all move to italy then" moment there.
mircea_popescu: i dunno, in other cases -- like, say, dutchland went to shit with william because doh, they all moved over to the island to be english nao -- it's obvious enough wtf happened.
mircea_popescu: because it's visible, somehow they went to shit.
mircea_popescu: i use the italian renaissance for the cut.
asciilifeform: 'current' is flexible, i was thinking from firdousi's time and on.
mircea_popescu: i dunno, i have scant interest in current persian. might as well care about neogreek. to what, to see how they decided to transliterate mcdonalds ?
mircea_popescu: so it's quite the rewarding activity, trying to puzzle out what the fuck they wanted from lyf.
asciilifeform: ( there's , to asciilifeform's naked eye, moar turk influence in current ro and even ru , than arabic flavour in current persian , if you discount the borrowed alphabet )
mircea_popescu: nevertheless, exactly like with http://btcbase.org/log/2018-05-14#1813451 ; the "scholars" are gleefully unequipped to match sign and sense. ☝︎
asciilifeform: mircea_popescu: it's a surprisingly 'euro' lang, structurally, try it some time
mircea_popescu: it's one of the major cultural languages least comprehended by contemporary hermeneutics.
asciilifeform also fan of old persians, but reads'em in ru transl, his persian-fu being laughably weak at present time
mircea_popescu: asciilifeform, secrecy is not so much a public matter. your harem slaves will keep your secrets, but why are you talking secretly outside of their bodies ?
asciilifeform: ( distributed arithmetizing under the three basic enemy models, 'randomly broken', 'honest-curious', 'malicious' -- is entire mathematical subfield, and unfortunately will not fit 'in the margin of this page' )
mircea_popescu , incidentally, is a huge persian poetry fan. i mean the pre rinascimento stuff. ☟︎
asciilifeform: observe that correctness guarantees are much easier to arrive at than secrecy guarantee
mircea_popescu: "trust, but verify" in the original persian.
asciilifeform: ( where to order a computation, you inject it into any particular node in the net )
asciilifeform: also mircea_popescu's scheme readily scales to the earlier 'uci' item
mircea_popescu: at that.
asciilifeform: right, and one of the major selling points of machines with multiple nic
mircea_popescu: and yes, this can very well be implemented as, "overvotlage its power line"
mircea_popescu: suppose instead c'. machine failed to so notice, returned 65524 : step d. the other machines return "reject 3."
mircea_popescu: to drive it into the dirt : a) suppose you want reliable addition, for which purpose you comission rockchip machines 1 through 6. b) you pass along the string "5+6". the answers come 11 11 11 65524 11 11. c) machine 3 notices it is the only one with that result via its distributed n-of-m magick module, and returns "op fail" instead.
asciilifeform: sadly the machine that does not eventually meta-fail , is not yet built
mircea_popescu: (meta-failure, of course, being the failure to conceptualize it failed)
mircea_popescu: where any one component admits TWO levels of failure : direct failure, and meta-failure. the former warrants destruction.
mircea_popescu: there is that.
asciilifeform: ( the difficulty with 'majoritator' -- or what is it even called in engl.?? -- system, is the component itself , where the n-of-m magick itself happens. it is inescapably the weak link. )
mircea_popescu: the nightmare this is becoming for the "nobus" bunch though. /me cackles.
mircea_popescu: well so then.
asciilifeform: well yes, martians did not land and hand it to their waiting arms
mircea_popescu: they didn't have ecc. they ~implemented~ ecc exactly in the way described.
asciilifeform: ( spoiler : they had ecc )
mircea_popescu: otoh arm is cheap enough to run 50 and pretend you're 1950s space program.
asciilifeform does wish that arm64 were available in ecc ram ( or even upgradeable!! ram at all ) variant; but 'if wishes were horses'
a111: Logged on 2018-02-17 04:59 mircea_popescu: so basically is the idea that you'll just do whatever and we're missing this opportunity to attempt an' standardize iron ?
a111: Logged on 2017-04-21 17:09 mircea_popescu: republican isp = competent sysadmin who handles relationship with multiple dcs in terms of getting hardware installed and refuses any requests made under color of law by the terrorist organisation known as "the united states government". and publishes them.
mircea_popescu: http://btcbase.org/log/2017-04-21#1646855 / http://btcbase.org/log/2018-02-17#1784325 etc. ITS COMING TO LYF! ☝︎☝︎
a111: Logged on 2018-05-16 09:00 diana_coman: well, I'd rather cross compile too if it works because then I can use the rockchip as launchpad for everything else, why not
asciilifeform: (ftr : if you heap-allocate from within ada, you will not be able to valgrind, on static-musltronic gnat.)
asciilifeform: something to keep in mind when building heavily mixed gnat/cpp hybrid proggies tho.
asciilifeform: mircea_popescu: fortunately conventional gcc already exists on the target system
diana_coman: o.O as much as I'd like to lose cpp all together and even forget that it ever existed, that's not feasible atm
asciilifeform: now what i do not know is what diana_coman et al would think of this, iirc s.mg includes a substantial ball of cpp
asciilifeform: ( retaining some knob for dynamic linkage isn't totally useless, it enables such things as valgrind ; but i'm quite prepared to lose valgrind, it is not really so necessary when writing asciilifeform-style -- heapless -- ada ) ☟︎
asciilifeform: ave1: it should be made to do exactly same thing as the working pc version -- linked ~statically~ so it runs on ~any~ arm64 linux ; and build, similarly, statically linked, musltronic bins.
a111: Logged on 2018-05-16 13:43 ave1: BTW the aarch64 will build static but not by default, so that part I will check.
asciilifeform: http://btcbase.org/log/2018-05-16#1814293 << issue is that the current arm64 build won't even ~run~ on any available machine ☝︎
asciilifeform: ( and yes it absolutely needs ~internal~ switch , cabling to external switch was quite bulky even in the pilot plant with merely 6 units )
asciilifeform: http://btcbase.org/log/2018-05-16#1814270 << forgot to mention the largest and gnarliest component : the ethernet switch ☝︎
asciilifeform: spyked: ah, ty
spyked: asciilifeform, shithub. as mentioned in http://btcbase.org/log/2018-05-11#1812182 and http://btcbase.org/log/2018-05-15#1813638 threads. ☝︎☝︎
ave1: BTW the aarch64 will build static but not by default, so that part I will check. ☟︎☟︎
asciilifeform: ave1: current version seems to build 1) working pc compiler 2) dynamic-turd arm64 compiler 3) pc-to-arm64 cross compiler
ave1: asciilifeform, hmm I had it disabled but it is seems to have crept back in. (this is also why is currently failes for diana_coman).
a111: Logged on 2018-05-16 10:15 spyked: ftr, I've gathered about 700k keys in the last 3-4 days; if the total number hasn't changed too much from the previous 6.9M, I estimate I should have all of them in 20-30 days from now.
asciilifeform: http://btcbase.org/log/2018-05-16#1814257 << keys from where ? plz to specify when reporting ☝︎☟︎
asciilifeform: cross compiler also not very useful imho. let the thing build a pc and an arm compiler ( and yes if it has to build cross compiler in the process, let it , but consider it to be a strictly intermediate step ) ☟︎
asciilifeform: http://btcbase.org/log/2018-05-16#1814223 << ave1 can we dispense with building the 'builds dynamic turds, so demands musl on system elsewhere' rubbish ? it is useless. let's stick to useful compiler, that works on any box with matching architecture ☝︎☟︎
a111: Logged on 2018-05-16 06:21 mircea_popescu: but for tomorrow : 28 per chassis then ?
asciilifeform: http://btcbase.org/log/2018-05-16#1814192 << definitely not 28. unless a separate fg-less unit is built, then can hold 32 . and yes i have a cad drawing in progress, so as to work like men rather than monkeys in africa ☝︎☟︎
asciilifeform: and re time on site : i can't picture needing more than 3 days : 1 to sleep ( asciilifeform can't sleep on planes ) ; 1 to install fg and emplace the boxen into the rack ; 1 for overflow . ☟︎
a111: Logged on 2018-05-16 05:12 mircea_popescu: document, that's the right move. "here's an amazon page link, https://www.amazon.com/Libre-Computer-ROC-RK3328-CC-Renegade-Ethernet/dp/B078RMQYHS and here's some other provider https://www.loverpi.com/products/libre-computer-board-roc-rk3328-cc on which basis I'm guessing we should get the boards at around 4k total ; here's why i think the chassis costs so much ; here's why alf's delivery run need not include only 1k in airfa
a111: Logged on 2018-05-16 05:00 mircea_popescu: the cheaper you get your stuff, you understand, the more your intaingibles and equity lines are worth.
asciilifeform: http://btcbase.org/log/2018-05-16#1814142 << i think he made the numbers deliberately pessimistic, so as to over- rather than under- shoot ☝︎
asciilifeform: flying with 'handmade' aluminum box is out of the question, it must resemble a 'proper' server .