log☇︎
236000+ entries in 0.152s
asciilifeform: if my sig is of 'A35A727799941F46F4500F25389F21F4E995F64AC65341080052EC014A8BACD76D992D0C7A5B0250502D', that thing will appear ~to naked eye~ in the pcode as 'zA35A727799941F46F4500F25389F21F4E995F64AC65341080052EC014A8BACD76D992D0C7A5B0250502D' .
asciilifeform: e.g., want to sign hash? get it whereever you like, and drop it IN THE PCODE
asciilifeform: it is POSSIBLE to write, e.g., sha512, or keccak, in p, but not essentially necessary, because one of the philosophical turtles of p is that you SEE what you are signing
asciilifeform: there are no special-purpose ops to produce hashes
asciilifeform: i was not going to expand on the 'p' thread until the proggy is done, but this is probably a good time to say 1 more ☟︎☟︎
asciilifeform: when they are poor; but 100x easier when poor+stupid also.
asciilifeform: it is not so hard to buy people, without even saying a word;
Framedragger: ah, well, that does make one feel.. more uneasy..
asciilifeform: they approve of the type of work that gpg turd is.
Framedragger: /me received one from google even, once. (for doing stuff related to... tor!!11eleven)
Framedragger: ohno he received a cheque from microshit??? that hitler!! cmon now.
a111: Logged on 2016-12-11 21:50 mircea_popescu: and then koch wants to go around pretending like he's one of us.
asciilifeform: http://btcbase.org/log/2016-12-11#1581840 << i dunno that he even bothers to pretend, he is a confessed recipient of microshit cheques ☝︎
a111: Logged on 2016-12-11 21:50 trinque: and the process is not reversible, couldn't even put the information in like "this had lf" somewhere
asciilifeform: http://btcbase.org/log/2016-12-11#1581842 << this is entirely so, and iirc i smashed my head into this very wall when jurov made ( per my scheme ) 'turdatron' ml system ☝︎
Framedragger: http://btcbase.org/log/2016-12-11#1581116 << very nice to have, much thx! ☝︎
ben_vulpes: either i remember a decrypted clearsign's hash not matching or am inventing the memory
ben_vulpes: yeah that sounds familiar, actually.
trinque: --decrypt gives you a freshly resquashed turd your machine never saw before.
trinque: and the process is not reversible, couldn't even put the information in like "this had lf" somewhere ☟︎
mircea_popescu: how, pray tell ? and check out how easy it is to unearth the heathens.
mircea_popescu: and then koch wants to go around pretending like he's one of us. ☟︎
trinque: ben_vulpes: was it you and me that ran into this problem at some point? I cannot remember regarding what
trinque: clearsig "canonicalization" includes line ending conversion to crlf before sig, and lf-only endings probably represent *all* submissions to the ML
mircea_popescu: but in general speaking, usg.koch blessed us with an idiocy which has to be resolved somehow.
mircea_popescu: then eg. the notes field could include link to ml.
mircea_popescu: trinque at the very least a result here could be where you machine-verify that X item deedbotted is equal to X' v-item through the process of a) X valid clearsigned b) X' valid detach0igned c) X = preprocessor(X')
mircea_popescu: trinque the clearsign iirc is a stricter operation ; but plenty of signedtxts emerge untouched from both processes.
trinque: so then if I've got a sig of an un-transformed item in ML, it may contain things the transformer would've stripped/modified if it were clearsigned
trinque: mircea_popescu: it seems to me the two paths (clearsign and detached-sign) sign completely different piles of bits, if the clearsign first goes through some transformer
mircea_popescu: oic. well it's the first here.
asciilifeform: mircea_popescu: this is neither 1st nor 10th 'faux computer' i make.
asciilifeform: you ~compose~ p operations by physically concatting the p-progs.
asciilifeform: (N is expected on top of stack)
mircea_popescu: asciilifeform this sounds pretty delishious tbh.
asciilifeform: but this is mere convention.
trinque: and whatever else it does to the poor clearsigs
trinque: so then the original item would have to have been detach-signed with line endings already mangled to get a clearsigned version that'd verify
asciilifeform: to verify a sig, you feed ptron this VALID PROGRAM ^ -- which creates the pubkey in memory -- and then on top of that, the sigogram
asciilifeform: what this does is store the modulus in variable #13 ('m' simply pushes integer 13 to stack, '!' stores top of stack in var # second-of-stack)
trinque: I'd have to spelunk further to confirm, but it looks like clearsigning is canonicalize -> sign
mircea_popescu: trinque mind that there'\s a loose cr/lf spuriously at the end of plaintext payload
mircea_popescu: ("find the longest ascii file so that a) lzw compression is under 1% and b) the encryption scheme therein discussed is trivially breakable)
asciilifeform: anyway there are other important aspects (e.g., a pubkey looks like:
asciilifeform: if overflow, ever -- terminates with sad.
asciilifeform: for instance, p code is required to first thing put on the stack 1) how many p-instrs are permitted 2) how many bits wide the fixints are ☟︎
asciilifeform: there are things that are not in p yet but have to be
asciilifeform: and yes this is a prog language where all ops are 1 character of an american kbd
trinque: mircea_popescu: reading up on the mangling process, maybe
phf: wait so X multiplies da9 and 100, = compares the result with 2ff, 9ae???
asciilifeform: phf: http://wotpaste.cascadianhacker.com/pastes/ApmIh/?raw=true << same example but before i stripped out the comments
asciilifeform: phf: with which the = ends up comparing the output of X
mircea_popescu: ah yes, there is that. so can you try an actual messagelist item ?
asciilifeform: phf: it is the known result
trinque: mircea_popescu: did, the signed item has to have already been run through gpg's idiot preprocessor
mircea_popescu: trinque theoretically at least it should be just transform the binary sig into a plaintext packet and glue it in.
phf: asciilifeform: what's the purpose of an extra number in the example?
asciilifeform: also this is bad example, in real life no use of p will look like this
asciilifeform: to make the parser almost nonexistently simple
mircea_popescu: trinque care to look into if you can bash the signature out of detachsig and attach it as clearsigned ?
asciilifeform: = takes 2 numbers off stack and puts 0 if they are equal, etc
asciilifeform: there are other commands, e.g., X takes 2 numbers off stack and modexponentiates
asciilifeform: 'z' simply puts a 0 on the stack, and if ptron sees 0-9 or A-F it shoves the 0 leftward 4 bits, and stuffs in the nibble in place
asciilifeform: moves the complexity, such that remains, into the payloads.
asciilifeform: trinque: it is a very simple thing, think 'rpn calculator' and you almost have it. ☟︎☟︎
trinque: new features could come on the P side.
trinque: alright, but I am willing to entertain running gpg and P in parallel for some time, ftr.
asciilifeform: it won't be fired in battlefield tonight, that is for sure.
asciilifeform: trinque: i cannot answer this, because it is an experimental weapon
asciilifeform: but i don't see a good argument for fucking it but not to deaoth
mircea_popescu: fuck standard gpg. in principle - this must work.
asciilifeform: afaik: not with standard gpg, or such that standard gpg will eat the result
asciilifeform: thread began with my asking people to think about this one.
asciilifeform: i dunno if there exists upon this cursed earth a rope that doesn't break when i go and pull on it...
mircea_popescu: it just eats the patch straight neh ?
asciilifeform: is there any means for doing this ?
asciilifeform: which brings back to #1 horror -- how do i deedbot a vpatch and its sig..?
mircea_popescu: "cold equations" is bad enough when you ~fuck up~, but let me tell you about when you didn't even fuck up in the first place.
mircea_popescu: good thing signature mechanisms are still human-administered or else we'd be sol.
asciilifeform: exists but i ain't signing that horror show.
asciilifeform: the schem will be nonclearsigned, the sig -- posted. the boards -- physically wrapped in copies of the schem. is all i can offer.
mircea_popescu: it'd be too fucking easy a hole for adlais to slither into.
asciilifeform: a 'everyone scrolls past when reading' turd does NOT belong in v.
mircea_popescu: hey, i don't want unreadable turds in v trees either.
asciilifeform: and tmsr-fab, and many other things.
mircea_popescu: yeah. but do you agree the actual problem is the lack of a tmsr-cad ? as far as "right things" go ?
asciilifeform: simply that i am stuck in this corner, where there is a thing that needs doing, that i do not have with what to do.
asciilifeform: it was not my favoured hypothesis that they must be fixed.
mircea_popescu: but if that's what it is - neither v nor deedbot need fixing for it.
mircea_popescu: so then tmsr-cad ?
asciilifeform: the actual semantic content of that schem is miniscule.
mircea_popescu: give me something other than a 5mb dxf to work with, as per this "Short when i wrote it" ?
asciilifeform: and is to STAY short.
mircea_popescu: at this point i'm not at all clear what problem you're trying to solve here, but it seems to have become "i wish to pack an arbitrarily long poem into an arbitrarily short notation in such a way it will never be forgotten."
asciilifeform: and tomorrow everyone ends up on boxes with gtk44444 and it fetches ALL urls.
jurov: as opposed to <!DOCTYPE shit blabla "http://webserver/blabla.dtd"> - if you feed this to xml decoder, it gets tempted to fetch it
jurov: it's just an attribute. when i include some TEXT chunk in png with an url, nothing happens either.
mircea_popescu: holy shit you're going to argue from "theoretical logic" ?
jurov: asciilifeform: the mircea's file does not have dtd declaration anyway...
asciilifeform: mircea_popescu: png is a bitmap format! where does it ask for www turd? can haz snippet ?