Show Idle (>14 d.) Chans


← 2021-09-20 | 2021-09-22 →
punkman: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-20#1058979 << you don't need to write your own CBC, just use iv=0
dulapbot: Logged on 2021-09-20 21:37:10 thimbronion: Ok - I have enough to go on at least.
asciilifeform: punkman: he asked how the nonce is used, asciilifeform answr'd algorithmically, not knowing what kinda libs thimbronion intends to use, nor whether they're correctly implemented, or contain cbcism, etc
dulapbot: Logged on 2021-09-20 21:30:32 thimbronion: asciilifeform: I am not clear on how to use the nonce in the red packet. I had been thinking it would be passed as the iv argument to the encryption function - then I realized - it's also needed for the decryption function, but it's not clear to me in the current scheme how the decryptor gets the iv.
asciilifeform: ( e.g. the ancient ada serpent asciilifeform dug out in era2 gave only three functions -- key eater, encrypt(16bytes), decrypt(16bytes) )
dulapbot: (trilema) 2017-10-18 asciilifeform: fwiw there is a quite short ada serpent existing, passes the test set ( and branch-free ).
thimbronion: asciilifeform: punkman is correct, I think. I'm not quite clear on why the result of the xor is being anded with 0xffffffff, however: http://paste.deedbot.org/?id=N5gD
asciilifeform: thimbronion: me neither !
asciilifeform: thimbronion: this is why asciilifeform doesn't like 'just use library!' approach
punkman: can also think of it as new_iv = encrypt(red_packet[:16], iv=0) then encrypt(red_packet[16:], new_iv)
punkman: I think X and 0xFF gives you last 8 bits if X is larger than 8bits
punkman: I'm writing Go prototype, just now compiled without errors! doesn't do anything yet though
punkman: good exercise for learning Go
asciilifeform: lol punkman , didja pick up oddball lang just for the occasion ? or habitually used ?
asciilifeform: incidentally, anyone ever used 'd' ?
dulapbot: (trilema) 2017-07-05 asciilifeform: interestingly 'golang' is a pretty straight ripoff of 'D' (by digital mars co.)
asciilifeform recently did megatonne of commercial work in 'd'. aint bad, vaguely like an ad-hoc orcish ada w/ better string handling
punkman: asciilifeform: just learning it now, never used. most of career has been shitscript langs
asciilifeform is no one to discourage folx from use of 'puzzle langs' but cannot guarantee that he will try $proggy if it's in cobol for vms or golang or similar
thimbronion is currently doing a project in BrightScript - yes Roku has its own lang!
asciilifeform has genuinely nfi what ^ is
asciilifeform: thimbronion: seems to resemble lua ?
thimbronion: asciilifeform: I'm not familiar with lua at all, so perhaps. Roku is a set top box thing.
asciilifeform: thimbronion: would ask 'why does tv box need own proprietary interpreter lang' but can guess.
punkman: thimbronion: is it android box?
shinohai: lol fluffypony released from the Fed's AirBnB https://archive.is/mTX0l
asciilifeform: eh, i was hoping he'd get the chair
shinohai: Must have sold watch, or gave seekrit Monero backdoor to NSA finally.
asciilifeform: shinohai: moar likely they simply wanted some btc (perhaps to refill waterfall, hence yest.'s -xx% dump)
shinohai: teh timing ....
thimbronion: punkman: I suspect not - otherwise I'd be using Java, probably. Actually don't know what the OS is.
shinohai: heh Roku's run some sort of proprietary linux, I had one briefly but not a tv person ....sp
jonsykkel: http://www.loper-os.org/?p=2175#selection-1323.44-1339.1 << is not doubly redundant since Stack(SP) alredy guaranteed to be 0000000 after Push; + FZ_Random() overwrites Stack(SP)?
dulapbot: Logged on 2021-09-21 17:26:15 jonsykkel: http://www.loper-os.org/?p=2175#selection-1323.44-1339.1 << is not doubly redundant since Stack(SP) alredy guaranteed to be 0000000 after Push; + FZ_Random() overwrites Stack(SP)?
asciilifeform: jonsykkel: the FZ_Clear(Stack(SP)) is only ~singularly~ redundant! and i'ma explain why included it
asciilifeform: jonsykkel: let's suppose that, for whatever reason, FZ_Random(RNG, Stack(SP)) does ~not~ overwrite Stack(SP) with rngolade. (why not? dunno, dr.evil put a nop there via telepathy. or 'smart' os. or whatever.) wouldntcha rather have easily spottable, with naked eye, result of 0 for the '?' call ? vs 'random-looking' but equally-predictable stack soup.
asciilifeform: jonsykkel: it is naturally impossible to write 'cosmic ray proof' software. but imho is in fact possible to write (and inexpensively) cosmic ray ~resistant~ softs.
asciilifeform: where, if you cannot necessarily prevent a machine from misfiring, at least can make the consequences obvious to naked eye.
asciilifeform: ( most importantly -- to STOP the process and let the operator know that something went off the rails, and anything may've happened, incl. say privkeys leaking into plaintext or whatever calamity -- before consequences are irreversible )
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059019 << epic cocktail of facts, fantasies, outright schizo night meth rantage, 'goring the cape instead of matador' ('they took our trump', 'islam') , a++
jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059020 << are you sure? was looking at a diffrent version of the source but unless im having a stroke the only place SP is modified is in zap push and drop - zap clears all the registers and drop clears the previous top item
dulapbot: Logged on 2021-09-21 21:20:40 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059018 << there is no such guarantee, jonsykkel ! 'Push' doesn't touch the contents of the stack. why didja think it did ?
jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059023 << ah, makes sense. agree one should always be highly suspicious of calling into "other peoples code"
dulapbot: Logged on 2021-09-21 21:23:16 asciilifeform: jonsykkel: let's suppose that, for whatever reason, FZ_Random(RNG, Stack(SP)) does ~not~ overwrite Stack(SP) with rngolade. (why not? dunno, dr.evil put a nop there via telepathy. or 'smart' os. or whatever.) wouldntcha rather have easily spottable, with naked eye, result of 0 for the '?' call ? vs 'random-looking' but equally-predictable stack soup.
jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059024 << should just start doing ones computing buisness on those triple redundancy systems
dulapbot: Logged on 2021-09-21 21:25:01 asciilifeform: jonsykkel: it is naturally impossible to write 'cosmic ray proof' software. but imho is in fact possible to write (and inexpensively) cosmic ray ~resistant~ softs.
jonsykkel: real evil/smart os would fill overwrite ur zero'd stack with random looking predictable soup
jonsykkel: gates shud get on it asap
asciilifeform: jonsykkel: the principal 'enemy' is actually moar mundane -- 'optimizing' compilers (which remove 'redundant') zeroizations. theoretically 'volatile' mark inhibits this behaviour, and it appears in the recent chapters where possible.
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059029 << this is correct. see, however, above. i'd much rather leave a redundant zeroization or 2 in a non-speed-critical routine, than 'oops i thought this gets 0'd somewhere but apparently nope'
dulapbot: Logged on 2021-09-21 22:16:41 jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059020 << are you sure? was looking at a diffrent version of the source but unless im having a stroke the only place SP is modified is in zap push and drop - zap clears all the registers and drop clears the previous top item
asciilifeform fucking hates 'cleverness', 'idiomatic c', and concretely the very psychotype of the people who perpetrate them.
dulapbot: (trilema) 2016-01-24 asciilifeform: riddle time:
asciilifeform: safety-critical program (which is what ffa is) must be obvious in its function like fucking hammer.
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059033 << where to get this, without somehow getting into the iron biz again ?
dulapbot: Logged on 2021-09-21 22:32:02 jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059024 << should just start doing ones computing buisness on those triple redundancy systems
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059035 << a bit harder to hide this 'in the haystack' than simple nop.
dulapbot: Logged on 2021-09-21 22:49:56 jonsykkel: real evil/smart os would fill overwrite ur zero'd stack with random looking predictable soup
jonsykkel: talking about 'optimizing' compilers - wonder if -O2 can possibly interfere with constant-timeness
asciilifeform: jonsykkel: try it! o3 too
asciilifeform: i expect somewhere there's gotta be a broken enuff compiler
jonsykkel: ill do that, gonna test it on my inferior c ffatron as well
jonsykkel: gonna have a bigger compiler selection at least
asciilifeform: ( there's pre-baked tapes there, with different hamming weights )
jonsykkel: excellent
asciilifeform: jonsykkel: i know for a fact there are cpu archs where MUL aint constanttime. i haven't got one to test with, but did write an egyptian word-mul for use on these.
asciilifeform: jonsykkel: in april of '20 i also found a theoretical nonconstanttimeism if the peh state machine straddles a cache line -- because of how hex digits processed. fixed this with rewrite of digit eater (and placing it in the fallthrough of the state mach.) but not released that ch yet
jonsykkel: interesting
jonsykkel: cant do aligned alloc?
asciilifeform: jonsykkel: not as a defined part of language nope
asciilifeform: nor does compiler necessarily know the cache behaviour of where the thing'll run, lol
asciilifeform: maybe tomorrow amd sells box where cache line is 1777 bytes strictly. somehow.
jonsykkel: its very possible
jonsykkel: Bit; Bit; Bit; Bit; Bit; Bit; Bit; Bit; << this is nice also
asciilifeform: gotta stick to the basic two rules of constanttimeism. the hexdigit thing is obvious violation in retrospect.
asciilifeform: jonsykkel: gnat doesn't unroll loops 'magically' ( picture if tried )
asciilifeform: so yes, bit bit bit...
jonsykkel: hehe
dulapbot: Logged on 2021-09-21 23:11:48 asciilifeform: ( activated/inactivated by HaveBarrelShifter knob in iron.ads )
asciilifeform: currently mul_word defaults to iron mul, if you want egyptian gotta change manually.
asciilifeform oughta change this to cleaner knob a la the shifts, via 'renames' device
jonsykkel: gotta make sure those knobs are clean
asciilifeform: anyways the arch where this supposedly matters is iirc ppc32.
asciilifeform: i don't have one anywhere, so have not tried.
asciilifeform: if anyone does ( signpost ? ) plox to write in. (supposing there's even a working gnat for it..)
asciilifeform: ( there may be others, this is the only 1 i happen to recall )
jonsykkel: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059042 << agre. all programs to be used/read by human person shud be hammerlike both in function and interface
dulapbot: Logged on 2021-09-21 23:04:21 asciilifeform: safety-critical program (which is what ffa is) must be obvious in its function like fucking hammer.
asciilifeform: jonsykkel: i suspect you'll get the intended meaning if think about it
dulapbot: Logged on 2021-09-21 23:05:15 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2021-09-21#1059033 << where to get this, without somehow getting into the iron biz again ?
← 2021-09-20 | 2021-09-22 →