302900+ entries in 0.186s
mircea_popescu: phf>
the graph for
that is X->Y->Z, X->Z <<
this is incorrect.
the graph for
that is "X->Y->Z". "X->Z" is not correct because Z requires b 3 and x does not provide b3.
☟︎ mircea_popescu: phf> but you
then run into
the issue is
that
the conflict might be introduced up
the graph chain << fuck, waht !?
a111: Logged on 2016-05-08 03:06 mircea_popescu: phf i don't
think polysemy is actually desirable in
this context. edge should really mean one
thing reliably.
mircea_popescu: phf i don't
think polysemy is actually desirable in
this context. edge should really mean one
thing reliably.
☟︎ mod6: that one sounded HUGE
though
mod6: fuck i hate
those
things.
ben_vulpes: again, if i recall correctly, asciilifeform's v will press all of
the same-leaf-level patches as
the given patch, /up
to
the given patch (inclusive)/ in alphabetical order
ben_vulpes: as in, given
this patchset, and
this particular patch, apply all of
the antecedents in
the correct order.
ben_vulpes: it's only ever made sense
to me in
the context of pressing a specific patch.
phf: but if you want
to
talk about presses you have
to look at
the whole subgraph, hmm
phf: i need
to just write
this stuff up, but at
this point i'm leaning
towards
the idea
that you can't really make a better graph
than "shares a hash", and all
that communicates is
that "this guy uses content from
this other guy"
phf: v's graph says "these
two patches share a common hash", but
that doesn't mean
that one can be pressed on
top of another in isolation
phf: i'm not
talking about gnudiff and presses
phf: asciilifeform:
that's not how v's ~graph~ operates
phf: asciilifeform: no conflict doesn't have much
to do with gnudiff. "conflict" in
this case means
that
the patches contain at least
TWO files one has shared hash, and
the other one hash different hashes
ben_vulpes: what
they have secret combudder/aerospace engineering programs?
phf: and
that's where my
thinking gets fuzzy, because i didn't really finish
this whole exercise
phf: but you
then run into
the issue is
that
the conflict might be introduced up
the graph chain
phf: so you can do a single pass (and
that's what my graph does)
to eliminate all
the edges
that have conflicts
phf: X and Z share a's hash, but
they have conflicting b hashes
phf: but you can't actually press from X->Z, even
though
they are "antecedents", because Z requires Y
to be pressed first
phf: the graph for
that is X->Y->Z, X->Z
phf: patch Z
takes a from 1
to 8, b from 3
to 4
phf: patch Y
takes b from 2
to 3
phf: patch X
takes a from null
to 1, b from null
to 2
phf: say you have
two files, a b. patch X
takes a from null
to hash 0 and b from null
to hash 0
ben_vulpes: > press as a graph
transition << would you elaborate?
phf: but
that doesn't guarantee a clean press as a graph
transition. it only guarantees clean press when you
topo sort
the graph
shinohai:
That's what I'm here for
to break it or fuck it up :D
mod6: thats not a bad
test we wanna
try
to break
this fucker
mod6: shinohai: ok
thats an interesting
test case
ben_vulpes: phf: iirc
that notion descends from stan's original V.
mod6: i just
tried dumping a priv key from an invalid bitcoin address, error says: 'error {"code":-5,"message":"Invalid bitcoin address"}'
shinohai: sweet. I kept getting
that error when I first patched it because I forgot
to decrypt my wallet like an idiot.
mod6: Ok i just
tried
to import an invalid private key
mod6: today is not
that day
mod6: some day im gonna have
to write a boatload more cucumber
tests for
this shit
mod6: i just
tested
trying
to dump a key from an address not in
the wallet and it says:
phf: note
that
the graph edge can actually have multiple different meanings. on mod6 graph, an edge means "patches share a common hash". on my graph edge means "patch can be applied on
top of other patch without conflict"
shinohai: hmmm ... good question on
the invalid key
thing.
That should be
tested sure
mod6: heh,
that's a good
thing (tm)
shinohai: At least it didn't destroy my whole wallet like last
time I used pybitcointools
phf: there were suggestion "just do X" and i would
try
them and discover
that
they wouldn't actually solve
the problem
mod6: does it work everytime as long as
the privkey is valid?
shinohai: I have done it at least 20-30
times with no major issues so far, mostly with vanitygen addys
phf: mircea_popescu:
that's what i was working on before
the log issue became critical, and i might need
to just write it up since i'm having hard
time communicating
the graph/press discrepancy
mod6: oh. hm. we better
test
the shit out of importing priv keys
shinohai: but it exports fine, which is really important
too lol
shinohai: back, yeah sometimes it acts a little weird when you import keys, like mine will hang a really long
time.
mod6: not
the end of
the world.
the whole
thing needs a rewrite anyway.
mod6: i
tried
to reimport
that same key - it gave me an error, which, it probably should, but it's a bit non-descript.
mod6: ok so
this
thing dumped out a privkey just fine.
mod6: anyway shoot me an addy, and i'll
throw you some digits
mod6: it's a bug
though. gotta dig into it. we have a
ticket for it.
mod6: ok,
truth be
told, im not sure what its malfunction is.
shinohai: mine will show previous balances based on
time it is synced on
the blockchain
mod6: gotta
test with real numbers
mod6: shinohai: oh its showing way more
than i actually have in
the wallet. it's re-adding
the change
to
the overall balance.
shinohai wishes we still had
testnet in a way. :/
mod6: u need some hacker
tokens?
shinohai: I haven't got
to
test send/receive yet ... still no place
to buy coins for cash in
this godforsaken country.
mod6: alright, i'll
try out some of
this stuff.
shinohai: lol mine built fine
too on Gentoo, gonna
test on Deb later
mod6: 'tail: /mnt/btc-dev/.bitcoin/debug.log: file
truncated' :D
mod6 builds
trb with key
tools
mod6: i
think i'll
take
this chance
to
truncate
this 22G log file
too
mod6: ok lemme see if i can build
this
thing
mod6: shinohai: since we're
talkin about it... looks like 'funken_prikey_tools.vpatch' had key.h based off genesis, instead of mod6_der_high_low_s.vpatch :
http://dpaste.com/0DWTN8S.txt mircea_popescu: i mean... graph's utility is mostly
that it mirrors
the press, no ?
mod6: nice
to know v does stuff
that works
mod6: if I use
the raw patch from
the deal press pukes on an unmatched hash. (a new feature! lol)