161300+ entries in 0.052s
mircea_popescu: asciilifeform incidentally do you recall that very ingenious thing that worked essentially as a vinyl record player needle but on a random chunk of clay pot ?
☟︎ mircea_popescu: "if thge best you can come up with is otp - give me back the cpu and take this mechanical gearbox instead"
☟︎ mircea_popescu: the punishment for otp is "you may not use computer anymore"
mircea_popescu: if i tell the woman to "spend an hour cooking, this 30 seconds cookings are bs" this doesn't mean i'm inviting her to pick her nose for 59.5 minutes.
mircea_popescu: i specifically don't fucking want sub second cryptography.
mircea_popescu: fuck that stupid shit. if the computer massaging my 10kb message takes a minute i couldn't care the fuck less - it took me 10 minutes to get all the guns arrayed to make a ml submission.
mircea_popescu: asciilifeform eh, you know what i mean. "use xor it's cheap!"
mircea_popescu: this "make the block small" and "make the cipher computable" restraints are EXACTLY wrong.
mircea_popescu: the idea that "better box" rather than "larger blocks" prevents from that is not unlike the fundamental idea at the basis of whitening.
mircea_popescu: but give it a GOOD read aforehand, it's crappy as found in .8
mircea_popescu: actually polarbeard if you're looking for more backportage stuff to do ... rawtx thing.
☟︎☟︎ mircea_popescu: stream cypher = particularly braindamaged blockcypher, ie one with a tiny block/
mircea_popescu: basically, do everything exactly opposite to the "wisdom" of a very inept if deeply self-pompositous field
mircea_popescu: there's plenty of interesting constructions in alt-algebras, number and set theory etc.
☟︎ mircea_popescu: incidentally, thinking vaguely along the lines of making a b-a call for papers for symm cypher, what would we actually want ? i'm thinking a) block sizes of 1, 4, 16, 64 kbytes. none of that bit-denominated bs, wtf is this, 64 bits. fuck that. b) key size of 64kb fixed. c) bonus points for proved hardness, as-hard-as-x etc d) bonus points for not using just basic arithmetics. fuck this shift-and-xor hack-and-slash bs,
☟︎☟︎ mircea_popescu: asciilifeform i think salsa was djb's own version of shiva, ie, "playing with ; don't use in anger"
mircea_popescu: punkman i've been thinking about it too, seems in point of fact we're mangling two distinct things. 1) patches-as-code ; 2) patches-as-human-readable.
mircea_popescu: <polarbeard> that's it, I'm patching every .c on my disk together with F 1000 and compiling it << lmao. STONE SOUP!
mircea_popescu: ascii_butugychag> my intent was that the user would include ALL of the leaves he is interested in by using a custom patch dir, and then press to the last. << this is not a bad approach (for dev work). i wouldn't want it in the press-for-deploy, but different story.
mircea_popescu: <jurov> and my other question about maintaining A and A' in "mempool" ? << everything is in the mempool al ltime anyway.
mircea_popescu: really, a properly behaving o shouldn't be a big deal to write neh ?
mircea_popescu: jesus i wonder how much "mission critical code" was really written by patch itself.
mircea_popescu: mod6 but does it press ? from what i see it correctly says no antecedents found etc
mircea_popescu: creates thriving communities and supports the implicit "fragmentation=strength" properties of the wot something fierce.
mircea_popescu: but as polarbeard among others can testify, this makes writing code fucking difficult, thanks god.
mircea_popescu: because if i sign random shit that doesn't actually connect to the tree even if it claims to
mircea_popescu: well yes because he never actually published them, ie, by sending to ml or w/e the process is.
mircea_popescu: mod6 as he says, turns out he wrote it on top of patches that didn't make in