asciilifeform: i.e. if someone were to submit an antipatch which brings a file hash back to a previously-seen value.

asciilifeform: re: earlier thread: a few hours of thinking led me to another potential boojum in 'v' : cyclic graphs ☟︎

asciilifeform: gotta nail this down, once and for all.

asciilifeform: (which i am trying to strip of all unnecessary 'intelligence')

asciilifeform: this is not acceptable in 'v'

asciilifeform: we resolved it before by doing shell gymnastics with the dir

asciilifeform: there is inconsistency

asciilifeform: but also in all earlier patches

asciilifeform: in the original 'orchestra'

asciilifeform: and is retarded

asciilifeform: presently some of them are, e.g., '+++ b/src/foo' and this slipped past my eyes

asciilifeform: to be specific, all diff hunks should take the form seen in 'genesis', where '--- a/bitcoin/src/foo', '+++ b/bitcoin/src/foo'

asciilifeform: can either do this yourself, or wait several days until i get around to it.

asciilifeform: without descending into 'bitcoin'

asciilifeform: mod6: they also need to be rebased (paths adjusted) so that they ~all~ apply from the root dir (where the patches live.)

asciilifeform: mod6: also might want to try applying the patches, sequentially.

asciilifeform: (new set should have no timestamps.)

asciilifeform: mod6: if so, then safe to sign.

asciilifeform: mod6: diff with the old set, and see that nothing but timestamps is at variance

asciilifeform: if not, BROKEN

asciilifeform: fluffypony: is the process ~exactly~ the same as on a gentoo box?

asciilifeform: fluffypony: now install a py package, say

asciilifeform: fluffypony: the 'fun' on mac begins when you diverge from traditional bsd utils and try using items like perl, python, or - satan forbid - attempt to work with init scripts or other customary denizens of /etc/

asciilifeform: less so today, in digital circuits, than in the old days of analogue amps in everything

asciilifeform: http://log.bitcoin-assets.com/?date=22-08-2015#1248408 << my understanding is that individually-measured and selected components are still sop in Machines Which Must Work ☝︎

asciilifeform: in unrelated nyooz, tr0l0l0l, http://www.loper-os.org/?p=1482&cpage=1#comment-17278 ☟︎

asciilifeform: is winblows next ? ☟︎

asciilifeform: ferfuckssake, is it really necessary to build on toy os ?

asciilifeform: a 'release' can simply be a deedbotted manifest.

asciilifeform: ben_vulpes: i'm still not entirely convinced that particular knob is necessary

asciilifeform: won't eat patches, no, but will eat sigs.

asciilifeform: now the ideal thing would be to hook deedbot straight up to this.

asciilifeform: deedbot- http://dpaste.com/2R8J8C5.txt

asciilifeform: congrats trinque !

asciilifeform: sig verifies.

asciilifeform: importantly, these need to be detached (own file) signatures. specifically ~not~ like this: http://therealbitcoin.org/ml/btc-dev/2015-August/000150.html ☟︎

asciilifeform: ;;later tell mircea_popescu ^

asciilifeform: ^ me.

asciilifeform: http://therealbitcoin.org/ml/btc-dev/2015-August/000153.html

asciilifeform: and, speaking of which,

asciilifeform: or 'v' will not eat descendant patches from said person.

asciilifeform: it is important that everyone who works on therealbitcoin sign 'genesis' and distribute said signature. ☟︎

asciilifeform: no vpatch, with the exception of 'genesis', should ever be longer than a few pages of text.

asciilifeform: the closest thing we have to a 'commit' here is the act of creating, signing, and distributing a vpatch.

asciilifeform: if any aspect of the basic theory here is the least bit unclear to any of you, i want to hear about it asap.

asciilifeform: because if not, it is not necessary.

asciilifeform: to further work the example, the 'antipatch' is only necessary if ben_vulpes's chain builds on any of asciilifeform's patches which have the unwanted patch as antecedent.

asciilifeform: aficionados of 'functional programming' will recognize the 'v' data structure as so-called 'immutable'.

asciilifeform: i am emphatically ~not~ recreating 'git' or the like.

asciilifeform: it avoids polluting the mechanism with special cases.

asciilifeform: this is my view.

asciilifeform: aha.

asciilifeform: creating the antipatch is likely to require some manual manipulation. ☟︎

asciilifeform: the thing he ends up having to do is to create an antipatch for that one, and place it on the head of his longest chain.

asciilifeform: one potentially warty aspect of 'v' is that let's say ben_vulpes agrees with all of asciilifeform's longest-chain patches but one, somewhere in the middle.

asciilifeform: but theoretically it is now possible to submit sigs for the same thing, and they will match.

asciilifeform: jurov's current script, unfortunately makes it necessary to post the payload (genesis) with the sig, in order for the message to come through.

asciilifeform: then sign and post your sig.

asciilifeform: trinque: see that your genesis.vpatch is bitwise-identical to mod6's.

asciilifeform: 'v' will recursively walk the hashes in the release and apply the necessary vpatches to produce the tree. ☟︎

asciilifeform: this is a 'release'.

asciilifeform: i am also considering a knob whereby one can sign a manifest

asciilifeform: say, one can ask for 'asciilifeform's longest chain.'

asciilifeform: this makes it possible to mechanically trace a 'patchchain' ☟︎

asciilifeform: is the idea.

asciilifeform: everything that anyone is to be signing is a vpatch.

asciilifeform: trinque: i did the genesis.vpatch thing solely to simplify 'v' build system, which i was working on; such that there is no longer any shenanigans with tarballs, copying of crud, etc

asciilifeform: with caveat that gnupatch will no longer keep you from applying the same patch twice.

asciilifeform: which left the timestamps. then experimented with abolishing timestamps in diffs entirely. this works

asciilifeform: earlier there were problems with different folks getting different (bitwise) genesis.vpatch - which i traced down to 'locale' retardation in gnudiff

asciilifeform: it is.

asciilifeform: they're the same files!

asciilifeform: ah in that sense, no shit, matches

asciilifeform: but if you were to switch the hash in vdiff to 256, then will match.

asciilifeform: so won't match in the bitwise sense

asciilifeform: the original manifest used sha256

asciilifeform: diff -uNr $1 $2 | awk 'm = /^(---|\+\+\+)/{s="sha512sum \"" $2 "\" 2>/dev/null " | getline x; if (s) { split(x, a, " "); o = a[1]; } else {o = "false";} print $1 " " $2 " " o} !m { print $0 }'

asciilifeform: mod6 used it to produce the latest genesis.

asciilifeform: i posted a new vdiff.sh

asciilifeform: damnit

asciilifeform: nope

asciilifeform: (ask how they destroyed fine art? not by 'hiding it in vaults', no. by elevating steaming piles of shit like jackson pollack's splashes, on account of them being supposedly 'authenticable' in his bucket swirling patterns, to 'fine art')

asciilifeform: http://log.bitcoin-assets.com/?date=22-08-2015#1248271 << compare this crock of shit to the algo for authenticating a bitcoin ! ☝︎

asciilifeform: propping up the 'world order' or however one calls it.

asciilifeform: no usg confiscates them, to keep the price from rocketing to alpha centauri. the 'art market' not only destroyed fine art as a going concern, but is one of the giant turtles on which the elephants stand

asciilifeform: there is no buterin's waterfall for rembrandts and vermeers

asciilifeform: http://log.bitcoin-assets.com/?date=22-08-2015#1248273 << in my eyes, a ~skilled~ art forger is a hero second only to a hypothetical fella who puts an actual bullet through a bilderberger. why, exactly, should i have any sympathy for the folks hitching a ride in the only 100% effective inflation shelter known, at the expense of literally everybody else ? ☝︎☟︎

asciilifeform: (mircea_popescu's bag of tricks is ultra-handy for items like this. specifically, in re: the ancient art of not wasting time with spurious crud)

asciilifeform: if the folks allegedly interviewed want to me to read this shit, they can sign it.

asciilifeform: http://log.bitcoin-assets.com/?date=22-08-2015#1248244 << ahahaha this is ~very~ easy. does the thing contain any material signed with the original leak's pgp key? no? then straight to /dev/null !! ☝︎

asciilifeform: see also the 'rating folks not in wot?' and 'branding a snake' thread

asciilifeform: (also serial #s tend to work poorly unless reliably tattooed on.)

asciilifeform: let the rousseaus bite one another's ears off, fighting over who is rousseau.

asciilifeform: what next, serialize the ants in the anthill ?

asciilifeform: mircea_popescu: what i don't grasp is why you're feeling so generous - giving orcs unique names.

asciilifeform will soon learn what, if anything, ru academiderpers did re: crypto in '90s, among other things.

asciilifeform learned, perhaps to his grief, on account of phf (shame on you!) that 'ozon' now takes shitpal. now bought a container of fresh b00kz, aha.

asciilifeform: orcs only need to be distinguishable from one another quickly & reliably.

asciilifeform: orcs can grunt to one another whatever names they like, but only men get unique name

asciilifeform: (were issued serial #s, tatooed on)