1 2011-05-24 00:00:22 <gmaxwell> sherpacore: no need to merge, I mean, why? if the issue was too much varience keeping customers away then they could soften the varience out of their own pockets to make theirs equal to deepbit.
  2 2011-05-24 00:00:43 <sherpacore> yeah i spose.. could be easier to do that if they pool their profits
  3 2011-05-24 00:01:46 <gmaxwell> I'm pretty sure that in the interest of the health of the network you could probably talk someone with a lot of coin to make you a float loan while you build up your own variance busting warchest.
  4 2011-05-24 00:01:55 <jrmithdobbs> sherpacore: the problem is none of the other pools have their shit together
  5 2011-05-24 00:02:02 <gmaxwell> But I don't really think that getting minimum variance is the issue.
  6 2011-05-24 00:02:13 <jrmithdobbs> sherpacore: they go down for 15+ minutes at a time regularly
  7 2011-05-24 00:02:16 <gmaxwell> jrmithdobbs: I think none of the pools really have their shit togeather, in fact.
  8 2011-05-24 00:02:30 <jrmithdobbs> gmaxwell: yes but deepbit is the lesser of all evils atm unfortunately
  9 2011-05-24 00:02:47 <BlueMatt> http://forum.bitcoin.org/index.php?topic=9591.0
 10 2011-05-24 00:02:51 <BlueMatt> wx2.8 :)
 11 2011-05-24 00:02:59 <jrmithdobbs> gmaxwell: but we've discussed this at length privately so i wont rehash ;P
 12 2011-05-24 00:03:09 <sherpacore> sweet
 13 2011-05-24 00:03:38 <BlueMatt> and they said it couldnt be done :)
 14 2011-05-24 00:03:43 <DontMindMe> hey. is bridging 2 bc clients who dont have port forward activated a good idea? because i basically tried that today. maximum was 600 clients at once :D
 15 2011-05-24 00:03:44 <gmaxwell> jrmithdobbs: hey, using three non-deepbit pools with automatic fast failover gets fantastic uptime. Better than deepbit. :)
 16 2011-05-24 00:04:31 <jrmithdobbs> gmaxwell: so publish a way to do it that's not a modified miner ;P
 17 2011-05-24 00:04:37 <jrmithdobbs> heh
 18 2011-05-24 00:07:45 <luke-jr> jrmithdobbs: Eligius
 19 2011-05-24 00:07:59 <luke-jr> jrmithdobbs: yes, there's been a lot of outages lately, but not 15+ minutes regularly
 20 2011-05-24 00:08:22 <luke-jr> generally, it's back within 5 minutes
 21 2011-05-24 00:08:40 <luke-jr> less than a minute on US now :p
 22 2011-05-24 00:09:03 <Asphodelia> It seems like the 0.01/KB rule hasn't really been revised at all since the value of bitcoins was several orders of magnitude ago. By the end of the year a transaction will cost more than a postage stamp. Maybe it's time to adjust?
 23 2011-05-24 00:09:13 <gmaxwell> Asphodelia: it's revised in git already.
 24 2011-05-24 00:09:27 <Asphodelia> Ah, nice. To what?
 25 2011-05-24 00:09:36 <gribble> {"ticker":{"high":7.45,"low":6.86,"vol":25456,"buy":7.06,"sell":7.0999,"last":7.0603}}
 26 2011-05-24 00:09:36 <JFK911> ;;bc,mtgox
 27 2011-05-24 00:09:37 <gmaxwell> 0.0005 is the MIN_TX now.
 28 2011-05-24 00:09:38 <jrmithdobbs> and there are nodes on the live network already using the new rules
 29 2011-05-24 00:09:47 <Asphodelia> Right on.
 30 2011-05-24 00:10:08 <jrmithdobbs> though not many with much hashing power behind them
 31 2011-05-24 00:10:27 <sherpacore> pixar is setting up a btc mining operation
 32 2011-05-24 00:10:40 <sherpacore> they are using spare cycles on their massive render farm to mine for coin
 33 2011-05-24 00:10:49 <Asphodelia> I find that unlikely; presumably they have more lucrative uses for their GPUs.
 34 2011-05-24 00:12:01 <luke-jr> Asphodelia: it's revised on Eligius since it opened
 35 2011-05-24 00:12:17 <luke-jr> Asphodelia: Eligius will accept 0.00004096 BTC per 512 bytes
 36 2011-05-24 00:12:26 <sherpacore> yeah the timestamp attack of +- 70 is purely theoretical.. if pools/miners use system time instead of network time and if merchants do the same or require a good number of confirms before deliverying the goods, they're fine.
 37 2011-05-24 00:12:35 <ArtForzZy> also, wrong
 38 2011-05-24 00:12:57 <luke-jr> Asphodelia: but that's below the p2p bandwidth costs, so you need to connect to the free-relay net to do it
 39 2011-05-24 00:13:11 <sherpacore> ArtForzZy: ?
 40 2011-05-24 00:13:12 <ArtForzZy> the min output rule and fees per kB werent touched in mainline
 41 2011-05-24 00:13:20 <gmaxwell> sherpacore: good number of confirms is already a widely known best practice.
 42 2011-05-24 00:13:30 <ArtForzZy> main difference is min fee is now 0.005
 43 2011-05-24 00:13:33 <ArtForzZy> err
 44 2011-05-24 00:13:42 <ArtForzZy> 0.0005
 45 2011-05-24 00:14:00 <gmaxwell> wtf is the point of chaning the min fee without changing the per kb /foward rule?
 46 2011-05-24 00:14:07 <luke-jr> gmaxwell: subcent txns
 47 2011-05-24 00:14:07 <yebyen> is anyone playing blackjack?
 48 2011-05-24 00:14:09 <jrmithdobbs> err ya
 49 2011-05-24 00:14:12 <sherpacore> yes.. but lots of nodes use network time. hopefully the miners dont
 50 2011-05-24 00:14:24 <gmaxwell> It's just going to enable people to create TXN that won't get forwarded.
 51 2011-05-24 00:14:42 <luke-jr> gmaxwell: no, it won't create the txns either
 52 2011-05-24 00:14:44 <ArtForzZy> errr... sorry
 53 2011-05-24 00:14:49 <luke-jr> gmaxwell: the change is ONLY for forwarding (and maybe mining)
 54 2011-05-24 00:14:53 <ArtForzZy> fee whas changed to 0.0005/kB
 55 2011-05-24 00:15:00 <gmaxwell> Thats what I though.
 56 2011-05-24 00:15:03 <Asphodelia> If a cheap double-spend attempt happens, will the standard client raise an alert, or silently discard the extra?
 57 2011-05-24 00:15:07 <gmaxwell> luke-jr: the GUI also enforced the minimum
 58 2011-05-24 00:15:08 <ArtForzZy> min output is still 0.01
 59 2011-05-24 00:15:17 <luke-jr> gmaxwell: the GUI still enforces 0.01
 60 2011-05-24 00:15:31 <ArtForzZy> luke-jr: huh?
 61 2011-05-24 00:15:38 <gmaxwell> No, the gui enforces the minimum payment, but the minimum fee has been reduced.
 62 2011-05-24 00:15:41 <ArtForzZy> GUI just calls GetMinFee
 63 2011-05-24 00:16:02 <luke-jr> ArtForzZy: if the GUI/bitcoind interfaces send a fee based on 0.0005 BTC, then things are going to break
 64 2011-05-24 00:16:13 <gmaxwell> in .21 the GUI will not let you send a very low prior tx without a fee of at least 0.01.
 65 2011-05-24 00:16:23 <ArtForzZy> gmaxwell: water is wet.
 66 2011-05-24 00:16:24 <luke-jr> the plan as I understood it, was to delay that change until the version AFTER the relay rules were updated
 67 2011-05-24 00:16:38 <ArtForzZy> yep
 68 2011-05-24 00:16:41 <ArtForzZy> 4:16
 69 2011-05-24 00:16:51 <BlueMatt> and you are up because?
 70 2011-05-24 00:16:55 <sherpacore> das coin
 71 2011-05-24 00:17:04 <luke-jr> BlueMatt tha'ts obvious
 72 2011-05-24 00:17:14 <luke-jr> http://xkcd.com/3
 73 2011-05-24 00:17:17 <luke-jr> http://xkcd.com/386/
 74 2011-05-24 00:17:17 <sherpacore> ArtForzZy: do you have a pool?
 75 2011-05-24 00:17:19 <ArtForzZy> because I can.
 76 2011-05-24 00:17:22 <ArtForzZy> sherpacore: nope
 77 2011-05-24 00:17:32 <BlueMatt> luke-jr: lol
 78 2011-05-24 00:17:49 <sherpacore> ArtForzZy: do you have to go to work.... or does btc pay you enough
 79 2011-05-24 00:17:51 <ArtForzZy> luke-jr: old, but always a nice one
 80 2011-05-24 00:17:55 <gmaxwell> sherpacore: sounds like in addition to the good confirm seasoning advice, high value targets out to be advised to establish good time sources and fix themselves to local time. (and also to peer with each other, though I think this is already happening to some extent)
 81 2011-05-24 00:18:01 <ArtForzZy> I work whenever I want to
 82 2011-05-24 00:18:05 <luke-jr> ArtForzZy: because it's so true
 83 2011-05-24 00:18:10 <sherpacore> ArtForzZy nice
 84 2011-05-24 00:19:11 <BlueMatt> oh well, its 4 am...gnight all, please take a look at wx2.8 and point out how terrible it is :)
 85 2011-05-24 00:19:31 <sherpacore> gmaxwell, yeah makes sense... i mean i dont really see anything wrong with putting everyone on system time (they will just fail to accept blocks for some period of time if they are 2h behind, and blocks they create if theyre solo mining will be out of time bounds)
 86 2011-05-24 00:20:10 <ArtForzZy> also, I currently don't have any major projects for work (well, other than trying to untangle a DOS-based pascal + custom ISA HW mess)
 87 2011-05-24 00:20:14 <sherpacore> >50% hashpower votes on acceptable 2h time range
 88 2011-05-24 00:20:31 <luke-jr> BlueMatt: you did it?
 89 2011-05-24 00:20:41 <BlueMatt> luke-jr: yep works afaict
 90 2011-05-24 00:20:48 <BlueMatt> but havent had much of a chance to test
 91 2011-05-24 00:20:54 <BlueMatt> http://forum.bitcoin.org/index.php?topic=9591.0
 92 2011-05-24 00:21:00 <BlueMatt> implemented very poorly though
 93 2011-05-24 00:21:15 <luke-jr> interesting
 94 2011-05-24 00:21:17 <luke-jr> poorly how?
 95 2011-05-24 00:22:01 <sherpacore> even if everyone picked their own time randomly without a time source.. whatever time was favored by >50% would create a longest chain
 96 2011-05-24 00:22:35 <sherpacore> or a plurality rather
 97 2011-05-24 00:24:21 <sherpacore> the other thing is to encode the # of sunspots at a particular time or level of background radiation somewhere in the universe since thats a physical observation everyone can make
 98 2011-05-24 00:25:46 <Asphodelia> Doesn't that require specialized hardware or something, though? You're probably better off with stock prices or something.
 99 2011-05-24 00:26:00 <Asphodelia> Or something.
100 2011-05-24 00:26:32 <sherpacore> yeah i spose.. you could measure the gravitational effect which would be unique based on the exact wobble of the earth
101 2011-05-24 00:26:50 <sherpacore> but time itself is easily measured with radioactive decay so.. everyone can agree
102 2011-05-24 00:27:30 <sherpacore> in the future youll send entangled particles to your peers, and they read one off every 1 second. if someone reports a wrong spin, you know theyre lying because you can just read the particle from your own stream
103 2011-05-24 00:29:11 <gmaxwell> sherpacore: it would be trivial to talk to the local ntp daemon if there is one, and if it thinks its time is good  use that. Interesting question if NTP is more trustworthy though.
104 2011-05-24 00:29:51 <gmaxwell> sherpacore: since on most systems 0wning up one or two NTP servers is all that would be required.
105 2011-05-24 00:30:32 <gmaxwell> sherpacore: thats worse than a sybil attack that requires drones on a bunch of different /16s.
106 2011-05-24 00:31:16 <gmaxwell> (the advantage of talking to NTP is that it actually knows if its probably right or not)
107 2011-05-24 00:31:50 <sherpacore> afaik it is now taking any ip, regardless of subnet but i see your point
108 2011-05-24 00:32:27 <sherpacore> its not too hard to connect to tor and spam out the world
109 2011-05-24 00:32:38 <newbit> is the price of bitcoins likely to go down in the future
110 2011-05-24 00:33:00 <newbit> isnt bitcoins just a giant pyramid scheme
111 2011-05-24 00:33:10 <Asphodelia> sherpacore: Most tor exit nodes block mail-sending ports... are you talking about a DOS on bitcoin?
112 2011-05-24 00:33:24 <gmaxwell> Asphodelia: no he's talking about a sybil attack on network time.
113 2011-05-24 00:33:36 <sherpacore> err i mean spam as in send a bunch of versions or other requests that update the network time
114 2011-05-24 00:33:57 <kika_> does anyone know why fedora 14 freezes after a few hours mining with diablominer?
115 2011-05-24 00:34:15 <gmaxwell> kika_: overheating your video card/ psu?
116 2011-05-24 00:34:30 <kika_> gmaxwell: that might do it freeze?
117 2011-05-24 00:34:39 <gmaxwell> yes, that might make it freeze.
118 2011-05-24 00:34:41 <kika_> gmaxwell: the whole system got froozen
119 2011-05-24 00:34:56 <kika_> gmaxwell: well the PSU is a silent pro gold 1200 Watts
120 2011-05-24 00:35:06 <kika_> gmaxwell: and its just one ati 6990 card
121 2011-05-24 00:35:26 <sherpacore> if you pwn ntp, then you either are controlling the victim's uplink and routing him elsewhere, or you pwned the time server.. in the first case, if you control uplink you can isolate him from bitcoin and fork him... in the second case... you pwn the world
122 2011-05-24 00:36:03 <sherpacore> everyone should have their own atomic clock on a card
123 2011-05-24 00:36:12 <gmaxwell> yea, if you own the uplink you can already do their own sybil attack.
124 2011-05-24 00:36:15 <gjs278> haven't had the need for an atomic clock yet
125 2011-05-24 00:36:26 <gmaxwell> sherpacore: I linked to cheap/awesome gpsdo's on ebay.
126 2011-05-24 00:36:28 <ArtForzZy> sherpacore: I just have one in a 3U :/
127 2011-05-24 00:36:42 <sherpacore> nice
128 2011-05-24 00:36:55 <gmaxwell> the thunderbolts are great, even with the gps down they keep time fantastically once they are up and going.
129 2011-05-24 00:37:08 <ArtForzZy> gmaxwell: well, kinda
130 2011-05-24 00:37:25 <ArtForzZy> their stock OCXO is rather... meh.
131 2011-05-24 00:37:25 <gmaxwell> (they contain really nice OXCOs)
132 2011-05-24 00:37:33 <gmaxwell> I think it's pretty good.
133 2011-05-24 00:37:34 <jrmithdobbs> gmaxwell: i don't like clients actually talking to a cdma/gps time device
134 2011-05-24 00:37:54 <jrmithdobbs> gmaxwell: set the local ntp server off the device and clients talk to that
135 2011-05-24 00:37:57 <ArtForzZy> well, it's better than pretty much all TCXOs, but for a crystal oven... not so great
136 2011-05-24 00:38:06 <gmaxwell> jrmithdobbs: sure sure.
137 2011-05-24 00:38:13 <jrmithdobbs> gmaxwell: as long as it's not a dell it usually has a better internal clock than the device
138 2011-05-24 00:38:17 <jrmithdobbs> in case of outages
139 2011-05-24 00:38:46 <gmaxwell> jrmithdobbs: no way man, thats why the thunderbolts are nice, their internal clock is pretty good...
140 2011-05-24 00:39:14 <jrmithdobbs> gmaxwell: the cdma devices i've had to work with would drift pretty bad sometimes if they lost signal :(
141 2011-05-24 00:39:19 <gmaxwell> well, okay, depends on if you want it as a frequency standard or a time source. there are different degrees of good. :)
142 2011-05-24 00:40:08 <vsrinivas> hi; when trying to compile bitcoin, i get this error: https://gist.github.com/988052 ; anyone seen anything like it?
143 2011-05-24 00:40:16 <ArtForzZy> yep
144 2011-05-24 00:40:30 <sherpacore> if you fix the difficulty curve ahead of time, the chain doesn't need to know about time, you are just making assumptions about moores law and such... but it probably wouldnt harm anyone if all 21 million appeared tomorrow
145 2011-05-24 00:40:53 <Namegduf> sherpacore: The point of difficulty is to limit block generation rate
146 2011-05-24 00:41:00 <Namegduf> Incredibly rapid block generation is bad for the network
147 2011-05-24 00:41:08 <Namegduf> Too slow means transactions don't go through for ages.
148 2011-05-24 00:41:10 <jrmithdobbs> sherpacore: no it'd be awful
149 2011-05-24 00:41:25 <jrmithdobbs> sherpacore: the p2p code would choke on itself trying to xfer that many blocks in that short a time
150 2011-05-24 00:43:08 <jrmithdobbs> gmaxwell: so what besides transactions can be used to influence time?
151 2011-05-24 00:43:29 <jrmithdobbs> i'm wondering how plausible this sibil attack is
152 2011-05-24 00:44:10 <vsrinivas> w
153 2011-05-24 00:44:12 <jrmithdobbs> transactions, block generation, anything non-trivial (ie, not requiring massive hashing power or coins to throw around)
154 2011-05-24 00:44:20 <sherpacore> downloading 420k blocks (21M btc) isnt a big deal... after that the fees wouldnt justify massive compute power
155 2011-05-24 00:44:22 <jrmithdobbs> that does influence?
156 2011-05-24 00:45:03 <sherpacore> the other way is for each btc to represent 1 vote on the production curve
157 2011-05-24 00:45:17 <gmaxwell> jrmithdobbs: it's really the version message on connect that matters.
158 2011-05-24 00:45:31 <jrmithdobbs> gmaxwell: wait seriously?
159 2011-05-24 00:45:50 <sherpacore> first you register as a voter (so lost bitcoins arent considered registered voters), then you vote on the production rate.. the plurality wins
160 2011-05-24 00:46:23 <sherpacore> if miners vote to generate 1 block every nanosecond they will vote bitcoin into uselesness (but they wouldnt would they)
161 2011-05-24 00:46:27 <gmaxwell> well the block themselves set one clamp. and it's clamped to ???70 min local time.
162 2011-05-24 00:46:43 <gmaxwell> sherpacore: voting on it would reduce the determinisim of the system.
163 2011-05-24 00:47:12 <sherpacore> yes.. but you don't require a time source and the production rate would maximize the utility for the btc shareholders
164 2011-05-24 00:49:08 <sherpacore> in most times the majority wouldn't vote to increase production - that would devalue their supply.. they would do it tho if too much btc is held in reserve for whatever reason.. so they adjust based on their own perception of utility
165 2011-05-24 00:49:27 <sherpacore> a democratic money supply
166 2011-05-24 00:49:45 <AAA_awright> Bitcoin is not democratic
167 2011-05-24 00:50:05 <sherpacore> in one sense it is - the network votes on the current time
168 2011-05-24 00:51:48 <sherpacore> i would prefer if it were one vote per person, but you cant test for personhood, so one vote per btc is the next best thing
169 2011-05-24 00:55:58 <sherpacore> so by proxy >50% of the miners could increase production by agreeing to post and accept 2x timestamps - for every 10 real minutes, the chain appears to go by 20 minutes.. other nodes will accept the first 6 blocks (a lead they cant outhash).. in the next difficulty adjustment, the difficulty goes down
170 2011-05-24 00:56:38 <gmaxwell> no votes determinism is a better thing on many basis.
171 2011-05-24 00:56:51 <gmaxwell> Besides there are plenty of "democractic" currencies now.
172 2011-05-24 00:59:34 <sherpacore> yeah.. it seems a mining majority can speed up production by doubling timestamps..  it would just take 60 real minutes to get 1 confirmation if they did that
173 2011-05-24 01:00:52 <gmaxwell> well they can't because eventually (quickly!) they'll go outside of two hours and their blocks will start getting ignored.
174 2011-05-24 01:06:28 <sherpacore> seems complicated
175 2011-05-24 01:06:31 <phantomcircuit> sherpacore, this has been discussed a lot, the attack works, but only once and only for about 4 blocks
176 2011-05-24 01:06:35 <phantomcircuit> sherpacore, so whatevah
177 2011-05-24 01:06:48 <sherpacore> which attack are we talking about
178 2011-05-24 01:06:56 <phantomcircuit> pushing the timestamps forward
179 2011-05-24 01:08:56 <gmaxwell> phantomcircuit: he suggested another one which is perhaps a bit more potent. open a 100 connections to each miner from different IPs (gogo tor) and drive their network time +70, then connect to mtgox from 100 IPs and drive its time -70.  Now mine, and your blocks are the only ones it heeds, because the real miners are outside of the two hour window. Lets you to temporary blockchain takeovers with less hashpower.
180 2011-05-24 01:09:37 <jrmithdobbs> right
181 2011-05-24 01:09:47 <jrmithdobbs> i think this is actually a plausible attack with real consequences
182 2011-05-24 01:10:21 <phantomcircuit> gmaxwell, sure but you'd need to be EXACTLY +-70
183 2011-05-24 01:10:42 <jrmithdobbs> phantomcircuit: no you can push it up/down slowly with multiple connects
184 2011-05-24 01:10:47 <phantomcircuit> gmaxwell, and since the exchanges all have fixed peering lists...
185 2011-05-24 01:10:50 <phantomcircuit> good luck with that
186 2011-05-24 01:11:06 <sherpacore> dont those 1-level removed peers talk to other untrusted peers
187 2011-05-24 01:11:15 <gmaxwell> yea, I pointed out that both pools and (I guesseD) exchanges are doing selective peering.
188 2011-05-24 01:11:15 <jrmithdobbs> yup.
189 2011-05-24 01:11:17 <sherpacore> eventually they have to accept a version from an unclean person
190 2011-05-24 01:12:05 <phantomcircuit> jrmithdobbs, no in the ~1s it takes a block to propagate the miner woudl have to be +70 and mtgox -70, exactly
191 2011-05-24 01:12:14 <phantomcircuit> jrmithdobbs, fat chance
192 2011-05-24 01:12:25 <jrmithdobbs> phantomcircuit: except that you have the 5-10 minutes between blocks to do it
193 2011-05-24 01:12:41 <jrmithdobbs> phantomcircuit: and just need to appear as >50% to the target
194 2011-05-24 01:13:07 <jrmithdobbs> phantomcircuit: maybe doing it to mtgox specifically is hard but there are other targets
195 2011-05-24 01:13:13 <gmaxwell> phantomcircuit: would there be anything bad about systems with robust time (e.g. pools and exchanges with local ntp stratum 1 clocks) just disabling the network offset stuff?
196 2011-05-24 01:13:57 <gmaxwell> (or adding a bit of code to query the local ntp daemon and use its time if it thinks its sane, network time otherwise)
197 2011-05-24 01:14:49 <jrmithdobbs> why is the network using itself for time at all?
198 2011-05-24 01:17:25 <phantomcircuit> gmaxwell, not afaict
199 2011-05-24 01:18:09 <phantomcircuit> jrmithdobbs, no actually you dont, the time drift has to be within network latency from the miner to the target
200 2011-05-24 01:18:15 <phantomcircuit> jrmithdobbs, so again... fat chance
201 2011-05-24 01:18:42 <gmaxwell> I'm missing why.
202 2011-05-24 01:18:47 <jrmithdobbs> me too
203 2011-05-24 01:18:58 <phantomcircuit> because the block time is checked when it's received
204 2011-05-24 01:19:21 <jrmithdobbs> sure, so between blocks you bump the clock, what am I missing?
205 2011-05-24 01:19:24 <phantomcircuit> maybe im just missing something
206 2011-05-24 01:19:30 <gmaxwell> Right and if theblock time and the target's time disagree by >120 minutes, the block will be ignored.
207 2011-05-24 01:19:44 <gmaxwell> You've driven both clocks out of wack with your sybil attack.
208 2011-05-24 01:20:05 <gmaxwell> the bound isn't tight because you can get them 140 minutes apart.
209 2011-05-24 01:20:12 <jrmithdobbs> phantomcircuit: eg, you would start right after a block was found, bombard your targets with bogus connects that drive up and down the clock and get it to +/-70 on either side before the next block
210 2011-05-24 01:21:01 <jrmithdobbs> phantomcircuit: what's being missed?
211 2011-05-24 01:21:32 <phantomcircuit> one sec
212 2011-05-24 01:22:29 <phantomcircuit> jrmithdobbs, oh i see there is a 10 minute gap
213 2011-05-24 01:22:34 <phantomcircuit> hmm
214 2011-05-24 01:22:42 <gmaxwell> I wish I could figure out why we bother with network time at all.. other than as something to compare the local clock to and fail throw an error when its too far off.
215 2011-05-24 01:22:43 <phantomcircuit> yeah you could do it
216 2011-05-24 01:23:08 <phantomcircuit> the patch is simply to change 70 to 60
217 2011-05-24 01:23:11 <gmaxwell> phantomcircuit: yea if it was ???60 allowed from local time then this would be okay
218 2011-05-24 01:23:12 <phantomcircuit> and magic it's impossible
219 2011-05-24 01:24:06 <gmaxwell> though if you're going to do 60 you're going to screw people with wrong timezones, might as well go tighter and reduce sillyness some more.
220 2011-05-24 01:24:39 <gmaxwell> (I assume timezone error is why it's at 70 now)
221 2011-05-24 01:25:01 <phantomcircuit> well i've learned that just because constants seem arbitrary doesn't mean they are...
222 2011-05-24 01:25:06 <gmaxwell> yea&
223 2011-05-24 01:25:32 <gmaxwell> I still don't see why hosts with working NTP couldn't just check the ntp status and use that if ntp reports good health.
224 2011-05-24 01:25:45 <phantomcircuit> ntp is centralized
225 2011-05-24 01:26:02 <gmaxwell> if the hosts time is from ntp then it's local time is already from ntp.
226 2011-05-24 01:26:12 <phantomcircuit> yeah im aware
227 2011-05-24 01:26:15 <jrmithdobbs> phantomcircuit: no it's not
228 2011-05-24 01:26:31 <jrmithdobbs> phantomcircuit: stratum 1 devices that can check multiple sources are cheap
229 2011-05-24 01:26:31 <phantomcircuit> jrmithdobbs, it totally is, the dns is centrally controlled
230 2011-05-24 01:26:35 <gmaxwell> jrmithdobbs: well it is kinda, it's driven from GPS clocks almost everywhere...
231 2011-05-24 01:26:50 <jrmithdobbs> it's just a matter of picking your sources wisely
232 2011-05-24 01:26:52 <gmaxwell> (or cdma clocks which are driven from GPS clocks)
233 2011-05-24 01:27:20 <jrmithdobbs> and if you want to argue that then time is centralized
234 2011-05-24 01:27:30 <jrmithdobbs> since governments have decided to arbitrarily fuck with DST/etc
235 2011-05-24 01:27:38 <gmaxwell> phantomcircuit: right now the fact that if a node listens you can pretty much set its time freely in the allowed window (and change the allowed window if you own NTP too) seems kinda ugly.
236 2011-05-24 01:28:08 <phantomcircuit> gmaxwell, well you cant actually if the node isn't listening for inbound connections
237 2011-05-24 01:28:20 <jrmithdobbs> phantomcircuit: then you target it's peers.
238 2011-05-24 01:28:41 <phantomcircuit> yeah and now the scale of your attack is hilarious
239 2011-05-24 01:28:48 <jrmithdobbs> not really
240 2011-05-24 01:28:48 <phantomcircuit> like the entire network more or less
241 2011-05-24 01:28:53 <gmaxwell> it's a cheap attack though.
242 2011-05-24 01:28:57 <gmaxwell> you need 100 ips.
243 2011-05-24 01:29:02 <jrmithdobbs> no you just need to appear as half the nodes on the network to the targets
244 2011-05-24 01:29:03 <gmaxwell> and lots of connections from each one.
245 2011-05-24 01:29:19 <jrmithdobbs> it's hard, especially with the stock code, to have more than 50-100 connections at a given time
246 2011-05-24 01:29:19 <phantomcircuit> jrmithdobbs, >half
247 2011-05-24 01:29:35 <gmaxwell> and you need to do nothing but the version string stuff.
248 2011-05-24 01:29:46 <phantomcircuit> to split network time for two areas of the network you'd need to control a significant % of nodes
249 2011-05-24 01:29:51 <phantomcircuit> probably around 98%
250 2011-05-24 01:30:00 <gmaxwell> yea the splitting is hard, you've got me there.
251 2011-05-24 01:30:00 <jrmithdobbs> basically the /16 limitation that was added to increase reliability opens the network up to this
252 2011-05-24 01:30:26 <phantomcircuit> gmaxwell, well only if you're not very clever
253 2011-05-24 01:30:27 <phantomcircuit> hehe
254 2011-05-24 01:30:36 <gmaxwell> but you don't need to control much, as mentioned you only need 100 IPs or so
255 2011-05-24 01:30:46 <jrmithdobbs> aka, tor
256 2011-05-24 01:30:49 <gmaxwell> phantomcircuit: the spliting while also trying to connect to all the peers of the hidden nodes.
257 2011-05-24 01:31:07 <gmaxwell> they can be on the same /24 for all the time code cares
258 2011-05-24 01:31:18 <jrmithdobbs> ya, finding the peers of mtgox and such is the hard part
259 2011-05-24 01:31:27 <jrmithdobbs> and if that's the hardest part i think it's a real problem
260 2011-05-24 01:32:16 <phantomcircuit> gmaxwell, no you'd need to control a large % of the network, since you'd need to completely split the network up
261 2011-05-24 01:32:30 <jrmithdobbs> phantomcircuit: no you don't
262 2011-05-24 01:32:37 <gmaxwell> phantomcircuit: why? just overwhelm the median operaiton.
263 2011-05-24 01:32:37 <jrmithdobbs> phantomcircuit: you just need to split your targets up
264 2011-05-24 01:32:49 <gmaxwell> *operation.
265 2011-05-24 01:32:57 <jrmithdobbs> phantomcircuit: that's enough to cause havoc
266 2011-05-24 01:33:24 <phantomcircuit> gmaxwell, each ip gets 1 vote
267 2011-05-24 01:33:35 <phantomcircuit> gmaxwell, 128 connections max
268 2011-05-24 01:33:38 <phantomcircuit> per node
269 2011-05-24 01:33:48 <phantomcircuit> assumign most peers are ~ right time
270 2011-05-24 01:34:23 <jrmithdobbs> phantomcircuit: eg, you'd shoot for pushing deepbit+slush forward 70min and then whoever your real target is back by 70, ify ou can do it in <10 minutes the majority of blocks genned will be rejected by your target
271 2011-05-24 01:34:34 <phantomcircuit> (120+60)/128
272 2011-05-24 01:34:42 <jrmithdobbs> phantomcircuit: now you have 1-2 hours to make use of it
273 2011-05-24 01:34:55 <jrmithdobbs> phantomcircuit: do it a few times a day for a couple of days and you'll get lucky
274 2011-05-24 01:35:08 <jrmithdobbs> it's sibil+race condition (the race condition being based on your hash rate)
275 2011-05-24 01:35:52 <sherpacore> it seems as if you have more then 66% of the hash power (double everyone else) you can do the push time forward thing, since they will take 30 minutes to create a block (by which time your next +2h block is already accepted, and it only took you 15 minutes).. they will keep popping blocks off your chain ad infinitum (and youve already paid the future bounties to yourself)
276 2011-05-24 01:36:30 <jrmithdobbs> phantomcircuit: the hard part is doing it to targets where you get monetary gain
277 2011-05-24 01:36:33 <phantomcircuit> sherpacore, if you have 66% of the hash power you can do whatever the hell you want
278 2011-05-24 01:37:07 <phantomcircuit> jrmithdobbs, if that's your goal then sophisticated timing attacks are a total waste of time
279 2011-05-24 01:37:13 <phantomcircuit> you could trivially DoS the network
280 2011-05-24 01:37:23 <phantomcircuit> seriously you could do it with a single 100mbps server
281 2011-05-24 01:37:25 <phantomcircuit> no problem
282 2011-05-24 01:37:26 <sherpacore> (but by that logic a 66% splinter off the original 66% group could usurp 1/3 of the conspiracy)
283 2011-05-24 01:37:32 <jrmithdobbs> phantomcircuit: yes but this is a builtin limitation of the protocol, not an issue with the current implementation
284 2011-05-24 01:37:54 <phantomcircuit> jrmithdobbs, the 70 minutes point can be narrowed
285 2011-05-24 01:40:19 <jrmithdobbs> phantomcircuit: shouldn't be trusting time from *connecting* nodes in the first place though
286 2011-05-24 01:41:02 <gmaxwell> ah, incoming connection limits would twart doing this attack quickly.. without actually really participating in the network.
287 2011-05-24 01:41:26 <phantomcircuit> gmaxwell, yes that was my point
288 2011-05-24 01:42:17 <sherpacore> isnt it true if you accept any time offset you can be hacked?  push deepbit/megamine down 5 minutes, everyone else +5 minutes, create a block +125 minutes (megamine rejects and continues mining a stale block... the rest of the network posts a new block and in 10 minutes megamine now must accept 2 blocks (or mined 2x times everyone else)
289 2011-05-24 01:43:01 <sherpacore> so megamine never gets to mine anything
290 2011-05-24 01:43:10 <phantomcircuit> lol yes
291 2011-05-24 01:43:15 <phantomcircuit> that's actually completely true
292 2011-05-24 01:43:21 <phantomcircuit> wow
293 2011-05-24 01:43:22 <phantomcircuit> epic fail
294 2011-05-24 01:43:24 <jrmithdobbs> phantomcircuit: see, this has other implications than trying to cashout in mtgox
295 2011-05-24 01:43:36 <jrmithdobbs> that's what i was trying to say and failing to illustrate, haha
296 2011-05-24 01:43:58 <jrmithdobbs> at the very least it's an effective attack for monkeying with difficulty given good targets
297 2011-05-24 01:45:20 <jrmithdobbs> unless BlueMattBot/sipa/jgarzik/gavin can point out something we're missing
298 2011-05-24 01:50:08 <sherpacore> what if the 2hours is based on localtime+2h instead of nettime.. then the max lag could only be a few seconds assuming high value nodes have good time sources.. and continue using nettime for all the other calculations
299 2011-05-24 01:53:31 <gmaxwell> sherpacore: it would mean nodes with boinked clocks would be forwarding the wrong blockchain frequently all on their own with no attack
300 2011-05-24 01:53:43 <gmaxwell> (e.g. no intentional attack at least)
301 2011-05-24 01:54:52 <sherpacore> i think its highly theoretical without any practical app... given the pools use system time and only talk to trusted hosts.. but a client without sufficient guards could be subtly messed with - the cost of doing it wouldnt be worth it tho heh
302 2011-05-24 01:56:25 <jrmithdobbs> sherpacore: what cost? boredom and a tor client?
303 2011-05-24 01:56:49 <gmaxwell> sherpacore: well they're defended by normal good practices: wait until you get many confirmed before doing anything you might regret.
304 2011-05-24 01:57:33 <jrmithdobbs> gmaxwell: this kind of redefines  how many confirmations are safe to rely though
305 2011-05-24 01:57:42 <jrmithdobbs> gmaxwell: taking it from to 6 to at least 13
306 2011-05-24 01:58:07 <gmaxwell> does it? the attacker has to mine all the confirm blocks too, while keeping the attack up.
307 2011-05-24 01:58:10 <jrmithdobbs> gmaxwell: and that's assuming hash power hasn't increased significantly since last diff change. pretty big assumption
308 2011-05-24 01:58:42 <jrmithdobbs> gmaxwell: that's just a race condition, do it enough times and you get lucky
309 2011-05-24 01:58:51 <jrmithdobbs> gmaxwell: this isn't exactly something that's going to be immediately noticed
310 2011-05-24 01:58:57 <jrmithdobbs> not until it actually causes impact
311 2011-05-24 01:59:50 <gmaxwell> well thats another obvious improvement, network health checks inthe client, and a paranoid mode that leave tx as unconfirmed if it thinks the network is unhealthy.
312 2011-05-24 02:00:05 <gmaxwell> a bunch of valid current diff blocks being rejected based on time is unhealthy.
313 2011-05-24 02:01:19 <jrmithdobbs> that would help, but it doesn't address the fact that you could disturb txn processing during those times
314 2011-05-24 02:01:25 <gmaxwell> every current difficulty invalid time is a really expensive warning that something is wrong wrong wrong.
315 2011-05-24 02:01:30 <jrmithdobbs> at least to some specific segment of the network
316 2011-05-24 02:01:47 <gmaxwell> yea but you could flood the same node right off the network.
317 2011-05-24 02:01:56 <gmaxwell> bitcoind has a bunch of fairly obvious DOS vectors too.
318 2011-05-24 02:02:32 <jrmithdobbs> just throw a decent ammount of legit miners without the async-io rpc patches, for instance
319 2011-05-24 02:02:35 <jrmithdobbs> lol
320 2011-05-24 02:03:17 <sherpacore> seems like with +-70 you get a 140 minute window to feed the victim phony blocks.. if you have 50% hashpower, you can send him 7 confirms.. so deepbit could probably empty out mtgox in a couple hours heh
321 2011-05-24 02:03:21 <gmaxwell> e.g. if bad network health is mostly detected based on valid difficulty but invalid time blocks thats far too expensive to fake as a DOS attack.
322 2011-05-24 02:03:56 <sherpacore> pump up everyone +70, create a block thats 70+120 = 190  which is+260 from the victims view (so he needs 140 minutes till he accepts it)
323 2011-05-24 02:04:30 <gmaxwell> well, as mentioned paranoid peering practice probably make it impossible for that particular target.
324 2011-05-24 02:04:41 <jrmithdobbs> ya mtgox is a bad example
325 2011-05-24 02:04:52 <jrmithdobbs> but joe the plumber accepting bitcoin for his services is not
326 2011-05-24 02:05:05 <sherpacore> during that 140 minutes, you are deepbit sized so you can send 7 confirms.. and then he has a +7 chain vs a +1 chain.. so add another 60 minutes for the victim to resync to the main chain.. thats like 10 confirms in total
327 2011-05-24 02:05:09 <sherpacore> yeah
328 2011-05-24 02:05:44 <sherpacore> an unprotected client could theoretically get 10 confirms if hes up against deepbut, and he accepts time offset, and he has open peers
329 2011-05-24 02:05:56 <gmaxwell> part of the joe plumber solution should be commercial reversal insurance services in any case.
330 2011-05-24 02:06:17 <gmaxwell> Since he'll want faster than one block confirms for some business, and an insurer could provide for that.
331 2011-05-24 02:06:55 <gmaxwell> as a side effect he'll also guard against a lot of these attacks.
332 2011-05-24 02:08:12 <sherpacore> for the full on attack, you have to up the rest of the network's time.. so you need to be able to upset the other main miners.. they have acl's and system time, so not really possible unless theyre in on it
333 2011-05-24 02:09:51 <sherpacore> otherwise you have just a 70 minute window to feed phony confirms... so you can do 3-4 maybe if youre deepbit
334 2011-05-24 02:12:30 <sherpacore> oh no not 10 confirms.. you only have 140 minutes, so like 7 confirms.. then hes already back on the main chain
335 2011-05-24 02:15:04 <sherpacore> but if you have many trials... say 1,000, your chances are better of doubling your phony chain... if you are attacking an exchange and they have a few billion, the expected returns are high heh
336 2011-05-24 02:15:34 <gmaxwell> the fact that time is only taken on connect complicates this I think.
337 2011-05-24 02:15:56 <jrmithdobbs> complicates but doesn't nullify imho
338 2011-05-24 02:16:04 <gmaxwell> even if you flood target's guard with connects you can't move his time until he reconnects to it in any cas.e
339 2011-05-24 02:16:16 <gmaxwell> agreed.
340 2011-05-24 02:16:29 <sherpacore> // Each connection can only send one version message
341 2011-05-24 02:16:30 <jrmithdobbs> this + a few well targeted short term ddoses to cause reconnects ...
342 2011-05-24 02:16:32 <jrmithdobbs> you see where I'm going
343 2011-05-24 02:17:09 <gmaxwell> hm. I think you can use this to overcome the connection limit.
344 2011-05-24 02:17:29 <jrmithdobbs> especially considering the (lack of) timeout handling in net.cpp
345 2011-05-24 02:17:30 <gmaxwell> say he only has one slot open, you cycle through 200 IPs in that one slot, I don't think they go away from the set on disconnect.
346 2011-05-24 02:18:00 <jrmithdobbs> wait really?
347 2011-05-24 02:18:32 <gmaxwell> I don't see anything take an item out of vTimeOffsets.
348 2011-05-24 02:18:54 <gmaxwell> // Ignore duplicates
349 2011-05-24 02:18:55 <gmaxwell> if (!setKnown.insert(ip).second)
350 2011-05-24 02:18:56 <gmaxwell> ...
351 2011-05-24 02:18:58 <gmaxwell> static vector<int64> vTimeOffsets;
352 2011-05-24 02:19:00 <gmaxwell> vTimeOffsets.push_back(nOffsetSample);
353 2011-05-24 02:20:05 <gmaxwell> so, pluses: takes more connects to overwhelm, minus: takes only one concurrent connect, .. can also run a node out of memory if you get enough IPs I guess. :)
354 2011-05-24 02:20:26 <gmaxwell> (does this support IPv6? :) )
355 2011-05-24 02:22:43 <jrmithdobbs> it kind of pretends to
356 2011-05-24 02:22:59 <jrmithdobbs> i've not bothered actually trying to connect to bitcoind via ip6 though, let me see
357 2011-05-24 02:24:46 <jrmithdobbs> wow allowing ip6 rpc would be tedious as fuck
358 2011-05-24 02:25:30 <gmaxwell> yea, but it looks like you can push any node you can connect to over and over again from different IPs to a hard ??? 70 from which it won't ever recover from.
359 2011-05-24 02:26:22 <gmaxwell> just take a /20 and connect from each IP one at a time and feed it -69.9 or +69.9 and it'll be so stuffed with bad data that it won't know what to do.
360 2011-05-24 02:26:43 <sherpacore> when does the client actually send a version? is that an optional command
361 2011-05-24 02:26:45 <jrmithdobbs> how do i not have nc6 installed
362 2011-05-24 02:27:50 <gmaxwell> it's mandatory.
363 2011-05-24 02:28:00 <gmaxwell> // Must have a version message before anything else
364 2011-05-24 02:28:02 <jrmithdobbs> gmaxwell: looks like right now it only binds to 0.0.0.0:8333 not ::8333
365 2011-05-24 02:28:03 <sherpacore> hmm
366 2011-05-24 02:28:25 <jrmithdobbs> err you know what i mean
367 2011-05-24 02:28:36 <sherpacore> is that in rpc.cpp where it constructs the version msg
368 2011-05-24 02:28:49 <jrmithdobbs> no it's in net.cpp
369 2011-05-24 02:29:59 <jrmithdobbs> gmaxwell: that looks to be more a bug/oversight though, it really should be binding to 0.0.0.0:8333 and :::8333
370 2011-05-24 02:31:11 <gmaxwell> well with v6 anyone with a v4 address can run your node out of ram eventually. ;)
371 2011-05-24 02:31:50 <sherpacore> if it takes the adjustedtime when sending version, then the victim is going to slightly adjust his peers when he versions them, but probably not by a large amt
372 2011-05-24 02:31:51 <gmaxwell> just by connecting once from every IP in a 6to4 /64 and getting an entry added to the seen IP set.
373 2011-05-24 02:31:54 <sherpacore> the bitcoin code is awesome
374 2011-05-24 02:32:45 <sherpacore> seems like its easy to add stuff. the obejct model is really well designed
375 2011-05-24 02:32:55 <jrmithdobbs> starting to wonder if satoshi is hiding in shame ;P
376 2011-05-24 02:33:10 <gmaxwell> pshaw. well, I think it's pretty readable.
377 2011-05-24 02:33:23 <gmaxwell> At least we can answer these questions without much work, and mostly reach right conclusions.
378 2011-05-24 02:33:33 <sherpacore> it uses iterators so no worrying about buffer overflows and the like which is nice
379 2011-05-24 02:34:07 <gmaxwell> I skimmed through all of it before getting on IRC and blabbing about bitcoin and didn't have any difficulty doing it. (though I didn't understand a fair bit, and didn't read it that critically)
380 2011-05-24 02:34:36 <sherpacore> bitcon is perfect!
381 2011-05-24 02:34:50 <sherpacore> looking up the source gives me some confidence to buy more.. its rock solid
382 2011-05-24 02:35:34 <jrmithdobbs> freudian slip or intentional commentary? ha
383 2011-05-24 02:35:41 <jrmithdobbs> 23:34 < sherpacore> bitcon is perfect!
384 2011-05-24 02:35:46 <sherpacore> err bitcoin
385 2011-05-24 02:37:00 <sherpacore> anyway i dont think timing can be abused, except in a hyper theoretical sense..  seems like the 2h window for accepting/rejecting a block should just be based on localtime, then the only consequence is youll lag behind and take longer to get confirmations if your system clock is way behind
386 2011-05-24 02:37:09 <sherpacore> or make it +2h from the median time of the past 11
387 2011-05-24 02:37:28 <sherpacore> min(real+2h, median+2h)
388 2011-05-24 02:39:23 <sherpacore> that way everyone excepts the block because the only criteria it cant be more than >2h the median  (even if it really took longer, the chain acts like a drifting clock that adjust slowly to prevent abuse)
389 2011-05-24 02:41:19 <sherpacore> that is, the criteria for new block timestamp is 0-2h from the median11..
390 2011-05-24 02:41:55 <sherpacore> so even if your nettime is pushed down 70 minutes, you will still accept the same blocks everyone else does based on the median
391 2011-05-24 02:42:26 <sherpacore> and if youre pushed up, you cant be fooled into accepting a block no one else would (since everyone is pegged to the median)
392 2011-05-24 02:42:52 <sherpacore> so it seems like just making it symmetrical would fix any discrimination based on differing times
393 2011-05-24 02:43:56 <sherpacore> if mining power drops.. then the time has to decay 2 hours for every 11 blocks...
394 2011-05-24 02:44:25 <sherpacore> so the difficulty is adjusted 5 times slower than it would be or something
395 2011-05-24 02:45:03 <sherpacore> you could have it back off like Tcp.. median+2h for 11... then median+4h....
396 2011-05-24 02:46:06 <sherpacore> anyway not an issues except philosophically.. bitcoin is totally secure
397 2011-05-24 02:53:45 <sherpacore> if the philosophical goal is to adjust production over time, then the +2h threshold should proably be based on real time (not what sybils claim the time is)..  over time means real world time, so the network must do a secret ballot (each determines his own time and hashpower votes on what the time realy is)  if the philosophical goal is just to slow the rate of production... then +2h should
398 2011-05-24 02:53:47 <sherpacore> be based on median of past blocks...
399 2011-05-24 02:56:31 <sherpacore> in the second an attacker generating medianlen/2 blocks can now make it 4h, so the medianlength/2 should be # of conseuctive blocks that would be beyond an evil person to do.. 20 consecutive blocks would be hard for an evil person with ~50%..  99.9999046%  it wont happen
400 2011-05-24 03:28:46 <bushing> does anyone have a little bit of time to help me troubleshoot why I can no longer make any transactions?  I can receive money, but no outgoing transactions seem to make it onto the network
401 2011-05-24 03:29:18 <ne0futur> bushing: upgrade to the last version ( .21 )
402 2011-05-24 03:29:25 <bushing> maybe it thinks my coins are double-spent?  this all started when I was trying to fix a ghost transaction
403 2011-05-24 03:29:25 <doublec> bushing: have you looked on bitcoincharts.com/bitcoin for them?
404 2011-05-24 03:29:35 <ne0futur> if you re still using .20 confirmations for sent money can take 12 hours or more
405 2011-05-24 03:30:13 <bushing> yes, I have & I was running .7 for a long time, accidentally, which actually seemed to work fine & now I'm running .21, and I've waited >24 hours, and no, they don't make bitcoincharts or blockexplorer & moment
406 2011-05-24 03:30:24 <ne0futur> bushing: and probably you ll have more answers on #bitcoin-dev
407 2011-05-24 03:30:52 <ne0futur> bushing: and you dont have a firewall somewhere ? or an ISP blacklisting the btc port ?
408 2011-05-24 03:31:19 <jrmithdobbs> ne0futur: this is #bitcoin-dev
409 2011-05-24 03:31:23 <bushing> this is bitcoin-dev.  I have like 140 connections
410 2011-05-24 03:31:32 <bushing> here's the transaction info, and the wallet tx: http://pastie.org/private/teimy5o14wxixfknwxr7la
411 2011-05-24 03:31:36 <ne0futur> "this all starurps, again i m on the bad channel , sorry ;(
412 2011-05-24 03:31:51 <bushing> the first transaction in the tx log is visible on the network but the last one isn't
413 2011-05-24 03:32:13 <bushing> and before this, I was seeing almost instant confirmations of some sort
414 2011-05-24 03:32:39 <bushing> so 2d3029f11& never hit the network
415 2011-05-24 03:33:17 <bushing> I've tried deleting the transactions from wallet.dat and rebuilding them using bitcoin -rescan; I'm not sure if that hurt, but it hasn't helped
416 2011-05-24 03:33:33 <bushing> (deleted the transactions using bc_tx)
417 2011-05-24 03:34:18 <bushing> to my uneducated eye, it would almost seem like my transactions are getting rejected
418 2011-05-24 03:34:27 <jrmithdobbs> http://blockexplorer.com/tx/9eb16c3af37b0f341dc4bd04b740dc5fa56c3ba3db2ba452f388b46518a36706
419 2011-05-24 03:34:33 <jrmithdobbs> it went through?
420 2011-05-24 03:34:41 <bushing> [22:31:51] <bushing> the first transaction in the tx log is visible on the network but the last one isn't
421 2011-05-24 03:34:50 <bushing> those are the last 2 txns in wallet.dat
422 2011-05-24 03:34:58 <jrmithdobbs> well fuck dude just paste the right txn id
423 2011-05-24 03:35:11 <jrmithdobbs> instead of making me scroll through the shittiest pastebin site ever
424 2011-05-24 03:35:13 <bushing> [22:32:38] <bushing> so 2d3029f11& never hit the network
425 2011-05-24 03:35:51 <jrmithdobbs> good luck, 'night
426 2011-05-24 03:35:55 <bushing> thanks
427 2011-05-24 03:37:07 <phantomcircuit> bushing, the resend timer is set to every 30 minutes
428 2011-05-24 03:37:08 <phantomcircuit> so
429 2011-05-24 03:37:11 <phantomcircuit> just wait
430 2011-05-24 03:37:19 <phantomcircuit> make sure you're connected to the network and wait
431 2011-05-24 03:37:26 <phantomcircuit> dont restart the client
432 2011-05-24 03:37:33 <phantomcircuit> since it's not smart enough to save the timer
433 2011-05-24 03:38:12 <bushing> i let it sit 8 hours, it's 22:37:18 local time here
434 2011-05-24 03:38:15 <bushing> (connected, etc)
435 2011-05-24 03:38:44 <bushing> so that's why I wonder if it's simply getting rejected
436 2011-05-24 03:39:52 <ne0futur> I d go to sleep and see tomorrow if I were you
437 2011-05-24 03:40:15 <ne0futur> and go back to a backup of my wallet if its not better tomorrow
438 2011-05-24 03:40:52 <bushing> I've done that; that was just my most recent attempt :( I've now gone through the process of sending a test transaction, waiting a couple of days, then trying to repair my wallet and trying again 3 times now
439 2011-05-24 03:41:22 <bushing> should the client get any feedback if it sends an invalid txn?
440 2011-05-24 03:42:42 <BurtyBB> hi bushing, didn't expect to see you here :)
441 2011-05-24 03:43:22 <bushing> meh, I'll look and see what I have in the way of older wallets, sigh
442 2011-05-24 03:45:12 <jrmithdobbs> bushing: try moving addr.dat out of the way, reconnecting, and going to bed
443 2011-05-24 03:45:29 <jrmithdobbs> bushing: you might just be broadcasting to/preferring some broken nodes
444 2011-05-24 03:45:41 <bushing> hm, okay, will do, thanks
445 2011-05-24 03:50:26 <cosurgi> stupid question - does mainline client finally support encryptrd wallet?
446 2011-05-24 03:50:51 <cosurgi> I remember there was a branch for that
447 2011-05-24 04:06:56 <_ape> ah even famous people use bitcoin apparently, haha :P
448 2011-05-24 04:08:12 <dissipate> _ape, what famous people?
449 2011-05-24 04:08:21 <dissipate> is charlie sheen using bitcoin?
450 2011-05-24 04:10:28 <_ape> well hes famous enough for sony to go after anyway :P
451 2011-05-24 04:11:20 <noagendamarket> lol
452 2011-05-24 04:11:56 <bushing> :P
453 2011-05-24 05:04:47 <kevin__> Hi,I am interested in bitcoin and i want to know the  mechanism of bitcoin.
454 2011-05-24 05:05:29 <kevin__> Could anyone tell me what is blocks?
455 2011-05-24 05:15:10 <stuhood> https://en.bitcoin.it/wiki/Block
456 2011-05-24 05:34:32 <diki> wake uuuuppp people
457 2011-05-24 05:36:08 <Clarence> everybody needs sleep
458 2011-05-24 05:36:13 <DontMindMe> diki: NO!!
459 2011-05-24 05:37:15 <diki> 5 hours are needed for sleep
460 2011-05-24 05:37:18 <diki> no more
461 2011-05-24 05:37:38 <diki> i usually go to bed 2-3-4AM and wake up at ~7:30
462 2011-05-24 05:37:39 <Clarence> well, most are getting it right now
463 2011-05-24 05:37:52 <Clarence> aren't you special
464 2011-05-24 05:37:54 <diki> there isn't time dude
465 2011-05-24 05:38:26 <Clarence> there isn't time not to
466 2011-05-24 05:42:35 <uppe> most people would be awake right now actually
467 2011-05-24 05:42:54 <uppe> given that europe + asia is in broad daylight right now
468 2011-05-24 05:43:11 <Clarence> what category of people?
469 2011-05-24 05:43:28 <uppe> Clarence: there's a world outside your pretty little island
470 2011-05-24 05:43:38 <Clarence> I stand corrected
471 2011-05-24 05:44:28 <diki> yeah it's 10 am here....
472 2011-05-24 05:44:32 <DontMindMe> try "The Uberman". http://a81.img-up.net/sleep3zgrm.jpg
473 2011-05-24 05:44:36 <Clarence> but I must interject... I could say that most people life on another planet where they sleep at this time.. there's a world outside of yours too
474 2011-05-24 05:45:36 <uppe> Clarence: but you can't prove that :-)
475 2011-05-24 05:45:42 <uppe> i can prove my statement
476 2011-05-24 05:45:47 <Clarence> oh well, it's the internet :)
477 2011-05-24 05:49:35 <diki> The internet commands you: wake up :P
478 2011-05-24 06:09:46 <Nesetalis> -throws a boot at diki-
479 2011-05-24 06:39:39 <gribble> The operation succeeded.
480 2011-05-24 06:39:39 <sipa> ;;later tell BlueMatt nice! 03:51:17 <@BlueMatt> AHHHH, guess who just got bitcoin running on wx2.8
481 2011-05-24 06:54:23 <gjs278> gribble, fetch me the mtgox prices
482 2011-05-24 06:54:26 <gjs278> ;;bc,mtgox
483 2011-05-24 06:54:26 <gribble> {"ticker":{"high":7.23,"low":6.86,"vol":16502,"buy":6.9999,"sell":7.0767,"last":7}}
484 2011-05-24 06:54:37 <gjs278> 6.99 is unacceptable robot
485 2011-05-24 07:15:03 <Egolibuster> hi totally new to bitcoin and not a geek myself
486 2011-05-24 07:15:21 <da2ce7> Egolibuster, welcome.
487 2011-05-24 07:15:23 <sipa> welcome!
488 2011-05-24 07:15:49 <Egolibuster> I would love to be able to accept bitcoin on my magento shop. Too stupid to understand how to adapt all that php stuff.
489 2011-05-24 07:15:57 <da2ce7> ;view 2883
490 2011-05-24 07:16:03 <da2ce7> ;;view 2883
491 2011-05-24 07:16:04 <gribble> #2883 Mon May 23 22:31:49 2011 da2ce7 SELL 1.0 noagendamarket date @ 1 btc (None)
492 2011-05-24 07:16:08 <Egolibuster> anyone here working on such a module?
493 2011-05-24 07:16:17 <sipa> Egolibuster: i've seen questions about magento on the forum
494 2011-05-24 07:16:31 <Egolibuster> cool, will have a look
495 2011-05-24 07:16:32 <da2ce7> ;;google magento
496 2011-05-24 07:16:32 <gribble> Magento: <http://www.magentocommerce.com/>; Home page - Magento Commerce Demo Store: <http://demo.magentocommerce.com/>; Magento: <http://www.magento.com/>
497 2011-05-24 07:16:38 <AAA_awright> Isn't there a channel where we don't talk about elliptic curve cryptography and key-exchanges and SHA algorithms written in OpenCL?
498 2011-05-24 07:16:53 <Diablo-D3> AAA_awright: no.
499 2011-05-24 07:17:12 <da2ce7> AAA_awright, are you dreaming?
500 2011-05-24 07:17:25 <AAA_awright> :)
501 2011-05-24 07:17:45 <Egolibuster> sipa, you meant the bitcoin forum, not the magento forum, right?
502 2011-05-24 07:17:45 <sipa> AAA_awright: #bitcoin :)
503 2011-05-24 07:17:49 <sipa> Egolibuster: yes
504 2011-05-24 07:18:06 <AAA_awright> Do we own that channel now?
505 2011-05-24 07:18:11 <sipa> yes
506 2011-05-24 07:18:17 <AAA_awright> Nice
507 2011-05-24 07:18:20 <da2ce7> er. isn't it spamy still from old clients?
508 2011-05-24 07:19:02 <da2ce7> hmm... werid.  I've never been in #bitcoin/ is it used much?
509 2011-05-24 07:19:11 <sipa> there are 10 left
510 2011-05-24 07:19:21 <sipa> it's the general discussion channel
511 2011-05-24 07:19:27 <sipa> what used to be #bitcoin-discussion
512 2011-05-24 07:19:42 <AAA_awright> I thought Freenode G-lined them
513 2011-05-24 07:20:51 <da2ce7> ah so we don't have old client spam anymore.
514 2011-05-24 07:21:49 <da2ce7> sipa, how is your wallet cypto going?
515 2011-05-24 07:21:58 <sipa> BlueMatt is doing that
516 2011-05-24 07:23:16 <da2ce7> ah using the ECDH insted of RSA?
517 2011-05-24 07:23:30 <sipa> BlueMatt's implementation uses AES256
518 2011-05-24 07:23:36 <sipa> no public key cryptography
519 2011-05-24 07:23:46 <da2ce7> gah.
520 2011-05-24 07:23:58 <da2ce7> what you were researching was far seperior.
521 2011-05-24 07:24:14 <da2ce7> *superior
522 2011-05-24 07:25:13 <sipa> it had one disadvantage: someone with write access to your wallet could add his own keys in your key pool
523 2011-05-24 07:31:02 <da2ce7> so you have your all your ECDH-DSA keys signed by and encripted by a master ECDH-DSA key... you have to enter your passowrd too generate new addresses also.
524 2011-05-24 07:31:21 <diki> sipa, are you the person maintaining bitcoin?
525 2011-05-24 07:31:39 <da2ce7> it gets kinda complicated...
526 2011-05-24 07:31:58 <da2ce7> but no secure system is single layed.
527 2011-05-24 07:34:17 <eps1> ;;bc,stats
528 2011-05-24 07:34:20 <gribble> Current Blocks: 126269 | Current Difficulty: 244139.48158254 | Next Difficulty At Block: 127007 | Next Difficulty In: 738 blocks | Next Difficulty In About: 2 days, 22 hours, 55 minutes, and 48 seconds | Next Difficulty Estimate: 396182.38043487
529 2011-05-24 07:34:21 <CIA-103> DiabloMiner: Patrick McFarland master * r8a2419a / (2 files in 2 dirs): Updated license headers by request - http://bit.ly/kUoGFj
530 2011-05-24 07:34:25 <eps1> ;;bc,mtgox
531 2011-05-24 07:34:25 <gribble> {"ticker":{"high":7.2,"low":6.86,"vol":16541,"buy":7.0787,"sell":7.0999,"last":7.0787}}
532 2011-05-24 07:45:26 <diki> why are people saying that if you invest in mining rig you wont even get 20% of the investment back?
533 2011-05-24 07:45:41 <Diablo-D3> because people are retarded
534 2011-05-24 07:46:06 <mtrlt> and retarded people can't do maths
535 2011-05-24 07:46:18 <diki> to be honest with bitcoins i was able to buy a cooler for my cpu, buy a domain and a bit more to buy another 5850
536 2011-05-24 07:46:43 <mtrlt> with bitcoins i have been able to buy 4 6990s
537 2011-05-24 07:47:15 <diki> and i like got 3-4 times more monay than the actual power used
538 2011-05-24 07:47:20 <diki> *money
539 2011-05-24 07:47:35 <mtrlt> yep, 1 BTC costs me $0.56 in power
540 2011-05-24 07:47:45 <mtrlt> even at the current diff.
541 2011-05-24 07:47:52 <Diablo-D3> ;;bc,mtgox
542 2011-05-24 07:47:52 <gribble> {"ticker":{"high":7.2,"low":6.86,"vol":15436,"buy":7.0787,"sell":7.1,"last":7.0999}}
543 2011-05-24 07:48:16 <diki> i calculate my power by my wattage used 24/7 for 30 days i.e a month
544 2011-05-24 07:50:10 <diki> My CPU is 125w tdp, and gpu is 151 tdp at stock which it is = ~275 watts + some other peripherals = ~300
545 2011-05-24 07:50:27 <diki> but since my cpu is idle, even a bit underclocked
546 2011-05-24 07:50:46 <mtrlt> i use a power meter.
547 2011-05-24 07:50:53 <diki> i wish i had a kill-a-watt
548 2011-05-24 07:50:55 <mtrlt> around 850 watts for 2x6990 at 910mhz
549 2011-05-24 07:51:14 <diki> another 5850 would add me like 151 watts more
550 2011-05-24 07:56:04 <Diablo-D3> I dont recommend more than 2x6990 on a single board unless you're power modding
551 2011-05-24 07:56:33 <diki> why not?
552 2011-05-24 07:56:46 <diki> afaik, there no multi-board mobos
553 2011-05-24 07:56:51 <diki> dual-socket, yeah, but not dual mobo
554 2011-05-24 07:57:02 <Diablo-D3> why not? because ATX shafted you.
555 2011-05-24 07:57:07 <Diablo-D3> or actually, AMD technically did
556 2011-05-24 07:57:28 <diki> if you are talking about me damaging my card\n640105
557 2011-05-24 07:57:53 <diki> I think at least...
558 2011-05-24 07:58:06 <diki> however it is stable if i mine at 98% gpu usage
559 2011-05-24 07:58:09 <diki> hasn't crashed once
560 2011-05-24 07:58:20 <Diablo-D3> diki: no
561 2011-05-24 07:58:22 <Diablo-D3> damaging the mobo
562 2011-05-24 07:58:24 <diki> But if i load it with OCCT using linpack...then oh man..
563 2011-05-24 07:58:44 <Diablo-D3> 6990s pull more than 75w off the slot rail at stock
564 2011-05-24 07:58:46 <diki> the only way a mobo can be damaged if it's a 4+1 phase and you stick a thuban there and OC it to 4Ghz
565 2011-05-24 07:58:53 <Diablo-D3> see above.
566 2011-05-24 07:59:00 <diki> and the slot only provides 75 watts
567 2011-05-24 07:59:05 <diki> you cant pull mroe than that
568 2011-05-24 07:59:08 <Diablo-D3> not quiiite.
569 2011-05-24 07:59:08 <diki> *more
570 2011-05-24 07:59:12 <diki> Yes quiite
571 2011-05-24 07:59:27 <diki> everything else is pulled from the power connectors
572 2011-05-24 07:59:44 <Diablo-D3> okay, if the ATX 12v pins supply x, and you pull more than y*4, where y is a slot, and y is 75 watts....
573 2011-05-24 07:59:51 <diki> which in a 6990 is 1x 6-pin and 1x 8 pin or 2x8pin
574 2011-05-24 07:59:52 <Diablo-D3> and y > x
575 2011-05-24 07:59:56 <Diablo-D3> then you're fucked.
576 2011-05-24 08:00:17 <Diablo-D3> diki: 6990 is more than 375 stock. it has 2 8 pins. each is 150.
577 2011-05-24 08:00:42 <diki> so you burned a 6990 or what?
578 2011-05-24 08:00:57 <Diablo-D3> no, the mobo will be damaged, pay attention
579 2011-05-24 08:01:04 <diki> has it been proved?
580 2011-05-24 08:01:07 <diki> Is this just a theory?
581 2011-05-24 08:01:10 <Diablo-D3> an ATX24 plug can only supply about 150w.
582 2011-05-24 08:01:25 <Diablo-D3> a 6990 pulls 75 or more off one rail.
583 2011-05-24 08:01:30 <Diablo-D3> what is 75 and 75.
584 2011-05-24 08:01:52 <diki> go on
585 2011-05-24 08:02:03 <Diablo-D3> this is basic math, diki
586 2011-05-24 08:02:14 <Diablo-D3> 2 slots at max load is already hitting the maximum
587 2011-05-24 08:02:32 <diki> <Diablo-D3> what is 75 and 75.<- There is no quesrion mark, thus not a question
588 2011-05-24 08:02:39 <diki> It could be a rhetorical qustion
589 2011-05-24 08:02:42 <Diablo-D3> nope, it is a statement.
590 2011-05-24 08:02:54 <Diablo-D3> 6990 + 6990 = max atx24 load on 12v.
591 2011-05-24 08:03:09 <Diablo-D3> dont add any more cards.
592 2011-05-24 08:03:25 <diki> so you are saying i shouldnt buy another 5850 for my mobo?
593 2011-05-24 08:03:31 <Diablo-D3> bingo.
594 2011-05-24 08:03:41 <diki> I am still gonna buy tho
595 2011-05-24 08:03:51 <Diablo-D3> yes, and you will be like two other people in here
596 2011-05-24 08:04:02 <mtrlt> diki has 2x6990?
597 2011-05-24 08:04:15 <diki> i *has 1x5850
598 2011-05-24 08:04:18 <diki> soon to be 2x5850
599 2011-05-24 08:04:22 <mtrlt> mmyea
600 2011-05-24 08:04:31 <Diablo-D3> everyone doing more than 150w off pci-e total are using power modded risers
601 2011-05-24 08:04:32 <mtrlt> Diablo-D3: see it was me talking about having 2x6990
602 2011-05-24 08:04:37 <mtrlt> :P
603 2011-05-24 08:04:53 <diki> my mobo has 2x pci-e x16 slots
604 2011-05-24 08:04:59 <diki> one is running at x4 tho
605 2011-05-24 08:05:02 <Diablo-D3> mtrlt: no, he asked
606 2011-05-24 08:05:05 <Diablo-D3> [05:56:04] <Diablo-D3> I dont recommend more than 2x6990 on a single board unless you're power modding
607 2011-05-24 08:05:06 <Diablo-D3> [05:56:33] <diki> why not?
608 2011-05-24 08:05:21 <diki> a 5850 is 151w tdp
609 2011-05-24 08:05:22 <mtrlt> yea but why did you say that line :P
610 2011-05-24 08:05:33 <diki> that one crappy watt can't possibly destroy a mobo
611 2011-05-24 08:05:37 <Diablo-D3> [05:50:55] <mtrlt> around 850 watts for 2x6990 at 910mhz
612 2011-05-24 08:05:50 <mtrlt> and no-one talked about having more than 2x6990 :(
613 2011-05-24 08:05:57 <Diablo-D3> diki: no, 5850 has a 6 and an 8 usually, and its pulling almost all the power off the plugs
614 2011-05-24 08:06:12 <diki> a 5850 has 2x6pin connectors
615 2011-05-24 08:06:18 <Diablo-D3> mine has 2 and 8.
616 2011-05-24 08:06:24 <diki> then you are screwed
617 2011-05-24 08:06:35 <Diablo-D3> on the contrary, its a much better card
618 2011-05-24 08:06:35 <diki> Although mine is reference
619 2011-05-24 08:06:51 <Diablo-D3> 2x6 gives you 150 total.
620 2011-05-24 08:07:06 <diki> and the max tdp of a 5850 is?
621 2011-05-24 08:07:20 <Diablo-D3> 151 @ 725
622 2011-05-24 08:07:31 <Diablo-D3> any additional power comes off the mobo
623 2011-05-24 08:07:47 <diki> if 2x6pin give 150 watts total
624 2011-05-24 08:08:03 <diki> then, those power connectors are already powering the card
625 2011-05-24 08:08:11 <Diablo-D3> yes, but anything OVER that comes from the mobo
626 2011-05-24 08:08:12 <diki> if not, then that means there is a small headroom for OC
627 2011-05-24 08:08:23 <Diablo-D3> mines 765 factory stock
628 2011-05-24 08:08:35 <Diablo-D3> and 6+8 gives you 225
629 2011-05-24 08:08:38 <diki> the other card i am going to buy is 730mhz stock
630 2011-05-24 08:08:57 <Diablo-D3> and 8+8 is 300, like I said earlier
631 2011-05-24 08:09:50 <diki> well, VC MSI R5850 TWIN FROZR II is 725 mhz but it's a bit over my budget
632 2011-05-24 08:10:03 <diki> so i might get the asus one which is 765mhz, but
633 2011-05-24 08:10:29 <diki> However, in crossfire if i choose to do, the faster card downclocks itself to the settings of the slower card
634 2011-05-24 08:10:52 <diki> even if i don't i may downclock on my own
635 2011-05-24 08:12:21 <Diablo-D3> diki: not quite.
636 2011-05-24 08:12:33 <Diablo-D3> driver just stalls rendering
637 2011-05-24 08:13:02 <diki> I need 300 bux for that card
638 2011-05-24 08:13:12 <diki> this includes shipping and any fees from paypal
639 2011-05-24 08:13:19 <diki> i currently have $130
640 2011-05-24 08:13:26 <diki> 135 actually
641 2011-05-24 08:13:55 <Guest67877> ?
642 2011-05-24 08:14:04 <diki> ??
643 2011-05-24 08:14:36 <diki> Anyway, diablo, the issue i will face is cooling..
644 2011-05-24 08:14:53 <diki> i don't want a rocket inside my hous so i must do something...
645 2011-05-24 08:14:58 <diki> *house
646 2011-05-24 08:19:03 <diki> so diablo, what hashing to use for storing my passwords? i chose md5 but not just md5(pass) it's a custom 3-4way hashing
647 2011-05-24 08:19:16 <diki> with salt
648 2011-05-24 08:29:30 <Diablo-D3> 100000 round PBKDF2 with a 128 bit salt using SHA512 as the hash
649 2011-05-24 08:29:47 <diki> is that even feasable with php?
650 2011-05-24 08:30:27 <Diablo-D3> lolphp
651 2011-05-24 08:30:44 <diki> Deepbit is written in php
652 2011-05-24 08:30:47 <diki> slush also
653 2011-05-24 08:31:03 <diki> php wins anytime
654 2011-05-24 08:31:05 <slush> no, python
655 2011-05-24 08:31:09 <diki> it's used in Facebook
656 2011-05-24 08:31:10 <Diablo-D3> php is fucking fail
657 2011-05-24 08:31:29 <Diablo-D3> diki: er, dude, facebook is a java shop.
658 2011-05-24 08:31:35 <diki> orly?
659 2011-05-24 08:31:43 <diki> maybe you hadnt noticed .php on most pages
660 2011-05-24 08:31:54 <Diablo-D3> dude, I dont have a facebook account
661 2011-05-24 08:31:57 <Diablo-D3> only losers have those
662 2011-05-24 08:32:18 <sipa> but they have their own php-to-c compiler, i believe
663 2011-05-24 08:32:25 <sipa> ... weird idea
664 2011-05-24 08:32:43 <Diablo-D3> sipa: then they can stop using java
665 2011-05-24 08:34:17 <sipa> don't know if they use java, but it's certainly possible
666 2011-05-24 08:34:31 <diki> google is multi-platform but they largely use python
667 2011-05-24 08:35:04 <da2ce7> sipa, n' co: http://hostinga.imagecross.com/image-hosting-01/7853bitcoincypto.png
668 2011-05-24 08:35:11 <da2ce7> this is what I had in my head.
669 2011-05-24 08:36:34 <sipa> why two keys per address?
670 2011-05-24 08:36:47 <Diablo-D3> because its over 9000
671 2011-05-24 08:37:14 <da2ce7> because you shouldn't encrypt and sign with the same key.
672 2011-05-24 08:37:23 <sipa> you don't
673 2011-05-24 08:38:00 <sipa> per key you'd have: a ECDSA pubkey, an ECDH pubkey, an ECDH-encrypted ECDSA privkey
674 2011-05-24 08:38:40 <sipa> bitcoin doesn't use encryption itself
675 2011-05-24 08:38:48 <da2ce7> you also need a ecdh, private key...
676 2011-05-24 08:39:11 <BlueMatt> sipa: btw, the only actual issue with wx was string conversion between wxString, std::string, wxChar, char
677 2011-05-24 08:39:11 <sipa> read http://en.wikipedia.org/wiki/Integrated_Encryption_Scheme
678 2011-05-24 08:39:15 <BlueMatt> (and string literal crap)
679 2011-05-24 08:39:21 <diki> tbh i don't understand all thse pub,private,intimate keys at all
680 2011-05-24 08:39:57 <sipa> da2ce7: read http://forum.bitcoin.org/index.php?topic=7830.0?
681 2011-05-24 08:40:43 <diki> so basically you are all working on encrypting the wallet and making sure no one can fetch it via memory either?
682 2011-05-24 08:41:12 <sipa> da2ce7: there are 3 types of keys in the system: the master key, the per-key ephemeral key, the real ecdsa keys
683 2011-05-24 08:41:49 <sipa> BlueMatt: seen my comment in the forum?
684 2011-05-24 08:41:51 <BlueMatt> sipa: you cant override functions based on return in c++ can you?
685 2011-05-24 08:42:02 <sipa> ah, that's the problem
686 2011-05-24 08:42:07 <BlueMatt> (as that is whats required)
687 2011-05-24 08:42:31 <BlueMatt> yea wx2.9 converts gracefully automatically, you have to do it yourself in 2.8
688 2011-05-24 08:42:49 <BlueMatt> and _ needs to return std::string, wxString, and char* depending on the situation
689 2011-05-24 08:42:51 <sipa> where does _ come from now?
690 2011-05-24 08:42:55 <BlueMatt> wx
691 2011-05-24 08:43:02 <BlueMatt> (or util.h ifndef GUI)
692 2011-05-24 08:43:08 <sipa> but it does string->string?
693 2011-05-24 08:43:17 <BlueMatt> no wxString -> wxString
694 2011-05-24 08:43:20 <sipa> ah
695 2011-05-24 08:43:42 <BlueMatt> or...I think its string-> wxString
696 2011-05-24 08:43:51 <BlueMatt> no, sorry its char* -> wxString
697 2011-05-24 08:43:59 <sipa> and GetTranslationChar?
698 2011-05-24 08:44:14 <BlueMatt> for when you need to return a char*
699 2011-05-24 08:44:29 <Sage> hey guys
700 2011-05-24 08:44:52 <BlueMatt> (and the crazy chars showing up, I think, is just crap because I stupidly didnt read and did a bunch of .mb_str() on stuff that isnt temporary)
701 2011-05-24 08:44:56 <BlueMatt> hi S
702 2011-05-24 08:44:59 <BlueMatt> Sage:
703 2011-05-24 08:45:10 <da2ce7> so ECIES is justa more effecent way of doing the same thing as I did with two sets of keys.
704 2011-05-24 08:46:03 <Sage> has it ever happened to you that your screen got distorter after setting memory to 400mhz on 6990?
705 2011-05-24 08:46:08 <sipa> ECIES = ECDH on master key + ephemeral key + symmetric encryption using the hash of the resulting agreed pubkey as key
706 2011-05-24 08:46:17 <sipa> ECIES = ECDH on (master key + ephemeral key) + symmetric encryption using the hash of the resulting agreed pubkey as key
707 2011-05-24 08:46:28 <diki> serious trolling here: http://horriblesubs.org/
708 2011-05-24 08:46:35 <BlueMatt> are we still on that encryption scheme?
709 2011-05-24 08:46:38 <diki> they should change it to a comedian site
710 2011-05-24 08:47:09 <sipa> BlueMatt: don't bother :)
711 2011-05-24 08:49:11 <sipa> BlueMatt: hmm, maybe you could make some utility functions str, wxstr, cstr that all have 3 versions, taking all combinations?
712 2011-05-24 08:49:25 <sipa> meh, nvm
713 2011-05-24 08:49:33 <BlueMatt> I pretty much do...just different names
714 2011-05-24 08:49:43 <BlueMatt> but not wxstr, I just use _ for that
715 2011-05-24 08:49:47 <sipa> ok
716 2011-05-24 08:50:03 <sipa> but not all strings that need this conversion, need to be translated, right?
717 2011-05-24 08:50:18 <BlueMatt> yep, thats what _T() is for
718 2011-05-24 08:50:30 <BlueMatt> string literals still need converted via macros to wx crap
719 2011-05-24 08:50:48 <BlueMatt> and the other stuff needs something like wxString(strThing.c_str(), wxConvUTF8)
720 2011-05-24 08:50:55 <BlueMatt> or string(wxString.mb_str())
721 2011-05-24 08:51:09 <BlueMatt> (the second one returns a temp pointer, which I think is where a fucked up a ton
722 2011-05-24 08:51:25 <diki> BlueMatt: and i thought you said you didn't know C/C++ :)
723 2011-05-24 08:51:32 <diki> that's knowing a lot
724 2011-05-24 08:52:00 <BlueMatt> diki: thats almost all from here: http://wiki.wxwidgets.org/Converting_everything_to_and_from_wxString
725 2011-05-24 08:52:01 <BlueMatt> :)
726 2011-05-24 08:52:38 <diki> i still have no idea what macros is
727 2011-05-24 08:52:44 <diki> apart from a macro picture
728 2011-05-24 08:53:01 <BlueMatt> oh, thats just when you see #define...
729 2011-05-24 08:53:21 <BlueMatt> like #define FUNCTION(thing) otherFunctionThatActuallyGetsCalled()
730 2011-05-24 08:53:36 <diki> isn't define used to make constants?
731 2011-05-24 08:53:38 <BlueMatt> which I should probably have done largely instead of the wxString conv crap
732 2011-05-24 08:54:23 <BlueMatt> define can be used for macros and constants which can be tested with #ifdef/#if
733 2011-05-24 08:54:26 <BlueMatt> but not with if()
734 2011-05-24 08:54:36 <BlueMatt> those constants are static const int ...
735 2011-05-24 08:54:57 <diki> isn't #ifdef used on compile time to determine some stuff
736 2011-05-24 08:55:03 <sipa> yes
737 2011-05-24 08:55:04 <BlueMatt> yep
738 2011-05-24 08:55:13 <BlueMatt> (that is all out my ass and may all be wrong)
739 2011-05-24 08:55:19 <diki> then it doesnt actually do checks when the program has been compiled and you run it
740 2011-05-24 08:55:32 <BlueMatt> yea
741 2011-05-24 08:55:34 <sipa> BlueMatt: #define'd things don't exist at all at compile time, they have been resolved in the preprocessor stage
742 2011-05-24 08:55:44 <BlueMatt> if() if for run, #if is for compile
743 2011-05-24 08:55:46 <sipa> they don't really have a time
744 2011-05-24 08:56:01 <BlueMatt> meh, I call preprocessor compile
745 2011-05-24 08:56:02 <sipa> *typw
746 2011-05-24 08:56:14 <BlueMatt> if its running in make, its compile as far as Im concerned
747 2011-05-24 08:56:19 <sipa> ok sure, but still, they are very different from a const static int variable
748 2011-05-24 08:56:31 <BlueMatt> yes, veryt
749 2011-05-24 09:01:22 <diki> i dont understand how inline assembly works
750 2011-05-24 09:01:33 <diki> so you have a function and inside it some assembly
751 2011-05-24 09:01:43 <diki> how the hell do they intereact together...
752 2011-05-24 09:01:47 <diki> *interactr
753 2011-05-24 09:02:07 <sipa> simple: the compiler copies the assembly to the output
754 2011-05-24 09:02:13 <sipa> instead of compiling it
755 2011-05-24 09:19:35 <beppu> diki: it's been a while since I've done assembly, but the variables from the C/C++ side should just be addresses in memory that the inline asm should be able to resolve.
756 2011-05-24 09:22:08 <beppu> diki: Also, the contents of the stack are predictable so there is another set of data the asm side can easily access.
757 2011-05-24 09:23:40 <beppu> diki: There are rules for how function parameters are supposed to be pushed to the stack.
758 2011-05-24 09:24:57 <lfm> if you look at the compiler's assembly output intermediat stage you might get an idea how its done
759 2011-05-24 09:25:52 <diki> i've seen assembly output
760 2011-05-24 09:26:00 <diki> quite a lot when i tried to unpack Crysis2.exe
761 2011-05-24 09:26:01 <BlueMatt> sipa: slight fixup pushed, care to test and see any strings you saw as broken are fixed?
762 2011-05-24 09:26:03 <beppu> Was it gcc -S that outputs asm?
763 2011-05-24 09:26:37 <lfm> beppu: ya
764 2011-05-24 09:26:56 <beppu> thx.  it has been aa while
765 2011-05-24 09:28:10 <diki> 64bit assembly addressed the registers with RAX rather than EAX, right?
766 2011-05-24 09:28:21 <diki> rax,rbs etc
767 2011-05-24 09:28:45 <beppu> my experience ended in the 32bit era
768 2011-05-24 09:28:46 <lfm> diki dissassembled binary is actually quite different from compiler asm output
769 2011-05-24 09:28:55 <diki> how so?
770 2011-05-24 09:29:20 <lfm> well the variable names and even line numbers of the C source are in the asm file
771 2011-05-24 09:29:32 <beppu> more readable
772 2011-05-24 09:29:34 <BlueMatt> can I get people to test: https://github.com/TheBlueMatt/bitcoin/tree/wx2.8
773 2011-05-24 09:29:42 <diki> I see.
774 2011-05-24 09:29:55 <diki> Ollydbg is a nice disasm
775 2011-05-24 09:30:33 <diki> I prefer v1 cause it has plugins allowing me to hide olly from Isdebuggerpresent
776 2011-05-24 09:30:50 <sipa> BlueMatt: compiling
777 2011-05-24 09:31:24 <lfm> diki well open source has no need for that
778 2011-05-24 09:32:55 <beppu> thank god ;)
779 2011-05-24 09:33:45 <beppu> in open source, you can see aglimmer of what a civil society might be like.
780 2011-05-24 09:34:53 <lfm> ya, it can be kinda idealistic
781 2011-05-24 09:36:24 <sipa> BlueMatt: works better already, but the status column still sometimes still has empty strings, and the bottom bar has superfluous characters sometimes
782 2011-05-24 09:36:43 <BlueMatt> ah, ok I see it...one sec
783 2011-05-24 09:37:53 <sipa> oh, and scrolling is very slow, but that may be related
784 2011-05-24 09:37:58 <rlifchitz> ;;bc,stats
785 2011-05-24 09:38:00 <gribble> Current Blocks: 126292 | Current Difficulty: 244139.48158254 | Next Difficulty At Block: 127007 | Next Difficulty In: 715 blocks | Next Difficulty In About: 2 days, 20 hours, 31 minutes, and 15 seconds | Next Difficulty Estimate: 396359.62100705
786 2011-05-24 09:38:21 <BlueMatt> sipa: hm, odd...scrolling is slow?
787 2011-05-24 09:38:42 <BlueMatt> arg I dont have a scrollable wallet on this machine, you have a testnet one I can borrow?
788 2011-05-24 09:38:43 <sipa> BlueMatt: yes, but that's maybe because it 'detects' a lot of strings changed, and redraws them
789 2011-05-24 09:38:53 <BlueMatt> it very well could be
790 2011-05-24 09:39:14 <sipa> so, first try to fix the overwritten strings-issue :)
791 2011-05-24 09:39:21 <BlueMatt> fair enough
792 2011-05-24 09:40:11 <sipa> and i'm testing with a real wallet actually (don't worry, no actual available funds in it you can steal :p)
793 2011-05-24 09:40:47 <sipa> oh no, wait, now you're going to implement a features that extracts all private keys and sends them to you...
794 2011-05-24 09:40:52 <BlueMatt> awww, my backdoor to steal sipa's coins didnt work...
795 2011-05-24 09:41:26 <BlueMatt> I keep my real wallet on a server and dont move it off unless it is encrypted
796 2011-05-24 09:41:46 <BlueMatt> this machine just has a ton of test wallets I use for testing, but none have enough txes to scroll
797 2011-05-24 09:43:05 <sipa> i don't get any connections though
798 2011-05-24 09:43:08 <sipa> even when using -connect
799 2011-05-24 09:43:16 <sipa> ;;bc,blocks
800 2011-05-24 09:43:17 <gribble> 126294
801 2011-05-24 09:43:27 <BlueMatt> I can always connect to my 24/7 fallback node
802 2011-05-24 09:54:14 <BlueMatt> ok sipa easily fixed, stupid mistake, yet again
803 2011-05-24 09:54:22 <BlueMatt> ( I think)
804 2011-05-24 10:00:11 <sipa> BlueMatt: same
805 2011-05-24 10:00:17 <BlueMatt> hm?
806 2011-05-24 10:00:47 <BlueMatt> in that case, I cant reproduce, what is the error?
807 2011-05-24 10:00:50 <sipa> i still see empty status fields
808 2011-05-24 10:00:59 <sipa> when i scroll, it changes
809 2011-05-24 10:01:05 <BlueMatt> which one is the status field?
810 2011-05-24 10:01:11 <sipa> the first column
811 2011-05-24 10:01:21 <BlueMatt> oh, I have no txes...on minute
812 2011-05-24 10:02:39 <BlueMatt> ok, now I see crazy errors like htat
813 2011-05-24 10:02:41 <BlueMatt> that*
814 2011-05-24 10:04:54 <BlueMatt> sipa: try now
815 2011-05-24 10:09:19 <sipa> BlueMatt: a lot better, but still slow scrolling :(
816 2011-05-24 10:09:35 <BlueMatt> I thought as much, care to send a wallet over?
817 2011-05-24 10:09:40 <BlueMatt> (testnet obviously)
818 2011-05-24 10:09:56 <lfm> bluematt or I could send you some testnet coins
819 2011-05-24 10:10:08 <sipa> don't have any tnBTC here, i think
820 2011-05-24 10:10:21 <BlueMatt> well Im tracking down a bug with testnet on the wx2.8 branch atm so...not really
821 2011-05-24 10:10:25 <BlueMatt> but really I need txes not coins
822 2011-05-24 10:10:57 <lfm> I can send you 100 to make what you want of them
823 2011-05-24 10:11:15 <BlueMatt> alright, well let me work out this bug 1st
824 2011-05-24 10:12:15 <sipa> BlueMatt: valgrind reports an error here
825 2011-05-24 10:12:38 <sipa> BlueMatt: http://pastebin.com/c0mkjFei
826 2011-05-24 10:13:05 <BlueMatt> ah, yep ok the conversion fails...arg
827 2011-05-24 10:23:05 <diki> on average...how much ram do you think a pool server needs?
828 2011-05-24 10:23:14 <diki> It's not much but i've only given my VM 360...
829 2011-05-24 10:23:27 <diki> cause 1GB is already used by Windows...
830 2011-05-24 10:23:40 <sipa> why would you run a pool server on windows?
831 2011-05-24 10:23:51 <diki> Cause i don't have any other machines?
832 2011-05-24 10:23:52 <BlueMatt> why would you run any server on windows?
833 2011-05-24 10:23:56 <BlueMatt> vm
834 2011-05-24 10:24:29 <lfm> or is mswin in another vm?
835 2011-05-24 10:24:42 <diki> Windows7->VM->ubuntu
836 2011-05-24 10:24:56 <BlueMatt> should be ubuntu -> VM -> win7
837 2011-05-24 10:24:56 <lfm> oh you poor boy
838 2011-05-24 10:25:25 <sipa> oh ok, so the pool server runs on ubuntu
839 2011-05-24 10:25:26 <sipa> nvm
840 2011-05-24 10:25:45 <diki> the problem is that the machine has only 360mb ram left for it
841 2011-05-24 10:25:52 <diki> and Bitcoind happily takes 70
842 2011-05-24 10:26:07 <lfm> use swap
843 2011-05-24 10:26:18 <diki> Swap=load on hdd
844 2011-05-24 10:26:25 <lfm> ya
845 2011-05-24 10:26:38 <diki> And windows7 will start writing on hdd if ram usage >=75%
846 2011-05-24 10:26:40 <BlueMatt> well if you are going to try to run it on linux on a vm on win7 you are going to have to
847 2011-05-24 10:26:44 <beppu> buy more ram?
848 2011-05-24 10:26:44 <diki> so both OSes writing in swap=bad idea
849 2011-05-24 10:26:48 <BlueMatt> why do you have to run win on there?
850 2011-05-24 10:26:59 <diki> Games, notepad++ everything?