396900+ entries in 1.907s
mircea_popescu: but on
the subject :
the only
thing
that matters is
the value
transferred. any other approach is a waste of
time.
☟︎ assbot: Logged on 24-09-2015 21:29:38; phf: ascii_field, mircea_popescu, et al: i've been experimenting with idea of
trusted nodes in bitcoind. so far: address has levels of
trust, ultimate,
trusted and untrusted. adding an address with -connect gives it ultimate
trust, with -addnode makes it
trusted, otherwise (reported by other nodes, random connections, etc.) are untrusted. nodes have
the same level of
trust as
their addresses.
trusted nodes (i.e. bot
mircea_popescu: check it out asciilifeform, "experts at Morgan Stanley" fixed
the world!
mircea_popescu: ed by
tools
that should soon be available as open source.”
mircea_popescu: Stroustrup said: “You can write C++ programs
that are statically
type safe and have no resource leaks. You can do
that without loss of performance and without limiting C++’s expressive power.
This supports
the general
thesis
that garbage collection is neither necessary nor sufficient for quality software. Our core C++ guidelines makes such code simpler
to write
than older styles of C++ and
the safety can be validat
☟︎ BingoBoingo: Nah, "Trend Piece" aka not news
that sneaks into
the news
mircea_popescu: "Adopting Russian Orthodox Christian
terminology for an important house of worship,
the building is called
the Moscow Cathedral Mosque. Its main golden dome and
tall minaret reflected
the style of many Orthodox churches, except for
the Islamic crescents atop
them."
BingoBoingo: "Don't get anesthesia in Peru,
They mess with
the functional parts"
BingoBoingo: It's nice
that
they put
that warning up
there
mircea_popescu: hands off
the functional parts, yo! stick
to
tits and hips and whatnot
mircea_popescu: last fucking place you want
the dudes
to get involved,
the cunt and
the other set of labia.
mircea_popescu: in other news,
the pin of
the association of plastic surgeons from lima, peru depicts as a logo a stylisized cunt.
mircea_popescu: it can just be published as an item, you don't even need a "connection" between
the
two.
mircea_popescu: this way B can send any payments its customers want, and A will idem send whatever payments its customers want, and at
the end of
the day
the whole balance is BTC settled at
the agreed upon price and
that's
that.
mircea_popescu: also
the correct way
to handle
this is
through some sort of repo agreement. "A agrees
to receive any sums from B
throughout
today, Sept 25th, and remit no later
than by Sept 26th, at 8:00 gmt at address so and so a sum of btc equal
to :
the sum of btc received ; plus
the sum of X currency received divided by Vx ; plus
the sum of Y currency received divided by Vy ; plus [etc]."
mircea_popescu: if you are going
to put something, put pgp. not pki, and in no case
http for a stateful machine.
assbot: Logged on 24-09-2015 14:20:54; asciilifeform: (a panel which fits on a passenger car roof - let's even assume one on
the hood, as well - would be lucky
to pick up 200W at high noon on a cloudless day.)
mircea_popescu: and yes it will have
to be redone if nothing else
then for
that reason.
assbot: Logged on 24-09-2015 14:13:39; asciilifeform: if
the protocol had been designed by sane people, ALL SIGNATURES WOULD HAVE SAME FORMAT regardless of for what
the signature is - for a file, or for
the key itself, whatever
mircea_popescu: you don't pay
the credit card, you move
the balance on a bigger credit card. duh.
assbot: Logged on 24-09-2015 13:50:30; hdbuck: raspi + 3D printed case!
they'll sell at least 100 of
them, so
that's around $1.2 million a piece
to pay back
the VCs. i am waiting for
the « toasters »
tho.
BingoBoingo: Yeah, Looks like it is just you and
the smaller MPIF share betting on
trump so far with everyone else offering odds
assbot: BitBet - Donald
Trump gets Republican Nomination :: 10.24 B (57%) on Yes, 7.84 B (43%) on No | closing in 6 months 2 weeks | weight: 99`567 (100`000
to 1) ... (
http://bit.ly/1Pzg83A )
brg444: about bitpay, can't say
they don't doesn't deserve what's happening
to
them :/
they're part of
the reasons why we have a
trove of redditard
today who
think everyone and
their grandmother should be using BTC for retail purchases...
phf: shinohai: what prompted
the question?
phf: shinohai: what's
the context?
gernika: It might not have been empty.
That said, I didn't see
the noise in
the logs I would expect
to see if it were connected
to another node.
phf: i don't have intuition for complete eatblock
time, but 3 weeks does sound like surprisingly long
time, even with all
the checks enabled
phf: gernika: you know if
the addr db empty? if you did a connect with
that node at any point, it'll have a populated address database and start connecting immediately on startup
gernika: It was started like
this bitcoind -myip=x.x.x.x -caneat -datadir=/home/gernika/.bitcoin-8-14/
ascii_field: the other important question is whether
the node being synced was also accessing
the net at
the
time
gernika: Yes it did. I did not dump all.dat files at once. I would dump one, eat
through it (in about 24 hours),
then dump
the next.
ascii_field: what i am asking is, did it at any point spend
time waiting for an edible block
to appear on disk.
ascii_field: but was
the entire blockchain already present on disk
to be eaten?
gernika: The node was not on
the network, it was just eating from disk. It did not appear wedged, just processing one block after another.
ascii_field: let me guess, of
the 3 weeks, most of
the
time was spent wedged ?
gernika: wedged - due
to not having
the patch yes
ascii_field: as in, didn't have
the bdb locks fix patch applied ?
gernika: Right. So I made it
to 368xxx and got stuck on
the large block syncing from one of your nodes over
the network. I
then (probably) had a bad shutdown and corrupted
the db. I
then used eatblock
to sync from what I had on disk up
to
that point.
ascii_field: this does NOT
take 3 weeks on my hardware.
ascii_field: the correct way
to use eatblock is with dumpblock coming from an already synced LOCAL node.
ascii_field: i am asking ~specifically~ how
the
thing was put
together
ascii_field: as in, where did
the blocks fed
to 'eatblock' come from ?
gernika: ascii_field
Thought it was real - 32 GB ecc ram, opteron 4226.
ascii_field: mircea_popescu once alluded
to a dirty ad-hoc implementation of
this by chinese operators, where nodes had vpn links
to one another
ascii_field: and for 'trusted' peers, in
the above sense,
to be ones where all comms in both directions are authenticated.
phf: well,
the patch gives special status
to ips
that were explicitly provided. if you're being mitm'd, your only recourse is operator intervention,
the goal of
the patch was
to ensure
that your recourse does not automatically become "use random shmoe"
☟︎ ascii_field: because
the whole
thing was written by
the profoundly retarded
ascii_field: as no useful activity can
take place while
this hangs
phf: where's packet dropping and delay results in a no-consequences disconnect, similar
to
that "no activity in first 60 seconds"
assbot: Logged on 24-09-2015 19:30:42; gernika: Friend is claiming he can do a full sync in 6-8 hours using
the phoundation client (non-ssd), while it
took my OpenBSD box 3 weeks using eatblock. Not sure what I'm doing wrong.
phf: misbehave is only
triggered when counterparty sends malformed, but still recognisable packets, so
that would be a much more noticeable attack
ascii_field: strategic, incidentally, in
that it did not
trigger 'misbehave' listing on my end
ascii_field: (all of
the mitm against my nodes has
taken
the form of vaguely plausibly-deniable packet dropping and delay)
ascii_field: phf: all
this will do is
that you will waste cpu cycles on peers who are getting packetdropped by usg
phf: punkman:
that specific loop can be closed much easier by checking for -connect during misbehave report and bailing
ascii_field: will say
this again, 'trusting' ip p.q.r.s is IDIOCY
assbot: Logged on 24-09-2015 21:29:38; phf: ascii_field, mircea_popescu, et al: i've been experimenting with idea of
trusted nodes in bitcoind. so far: address has levels of
trust, ultimate,
trusted and untrusted. adding an address with -connect gives it ultimate
trust, with -addnode makes it
trusted, otherwise (reported by other nodes, random connections, etc.) are untrusted. nodes have
the same level of
trust as
their addresses.
trusted nodes (i.e. bot
phf: -connect based nodes in large avoid
this problems because
there's a mainloop
that keeps adding same -connect supplied addresses over and over again, so even if elsewhere it's decided
to drop
the node, it'll be added and reconnected again on
the next iteration. never
the less a connect node can still be banned for misbehaving, which is something
that his patch prevents from happening
phf: one way it might help with ascii's problem specifically
though is
that your client will patiently wait for ascii's node, even
though
there might be periods of
time where it would otherwise be completely unresponsive.
assbot: Logged on 30-08-2015 15:45:26; asciilifeform: ;;later
tell mircea_popescu (corrected) dulap still behind; i had a
thought
that
the very possibility of
this problem is an atrocious idiocy - why should an infrastructural node (so0opernoude!!!) rely on
the whims of wild animals
to learn latest blocks? we need
the 'nobles'-never-disconnect
table.
jurov: phf if you recall alf had some kind of middleman problem.. and
this won't help
phf: appreciate any comments on
the subject.
phf:
to misbehave, idle however long and send data as large as
they want. what's not implemented: prioritizing
trusted nodes over others during node selection: you might still lose connection by natural means, in which case -addnode nodes will be dropped, and a standard node selection mechanism is used.
the patch so far is here
http://paste.lisp.org/display/155710. i'm
thinking
that ultimate vs.
trusted distinction might be unnecessary. i would
phf: ascii_field, mircea_popescu, et al: i've been experimenting with idea of
trusted nodes in bitcoind. so far: address has levels of
trust, ultimate,
trusted and untrusted. adding an address with -connect gives it ultimate
trust, with -addnode makes it
trusted, otherwise (reported by other nodes, random connections, etc.) are untrusted. nodes have
the same level of
trust as
their addresses.
trusted nodes (i.e. both
t. and ultimate) are allowed
☟︎☟︎ Pierre_Rochard: from
the public info I read,
they’ll continue with a skeleton crew, could make a
turnaround if
there’s a bubble soon :/
BingoBoingo: BitPay is/was
the non-retarded retail alternative
to CoinBase
BingoBoingo: Because like CoinDesk is going
to have a friend in Atlanta walk by
the BitPay offices
to
take in
the rumoured slaughte
BingoBoingo: Wanted
to commit
the report
to
the
textual record as soon as possible
assbot: Almost 10,000 Syrians have registered
to live in a country
that might not exist -
The Washington Post ... (
http://bit.ly/1KxKnH3 )
mike_c: thx
to kako for rapid server change.
mike_c: would rather not do it server-side
to reduce dependencies
mike_c: I was going
to add an ajax search of
the user page
to
the 'not found' btcalpha wot nick search page