1 2012-10-10 01:08:43 <jgarzik> heh
2 2012-10-10 01:08:46 <jgarzik> -reindex is working
3 2012-10-10 01:09:19 <jgarzik> but does not make the process appreciably faster than -loadblock or IBD from a fast network peer ;p
4 2012-10-10 01:09:35 <jgarzik> ACTION kicks BDB in the rear
5 2012-10-10 04:34:45 <sipa> jgarzik: should be roughly the same as loadblock, as blk000?.dat flushing is very infrequent during ibd
6 2012-10-10 05:53:29 <abrkn> hey, i've been thinking about making my first btc app. i was thinking of making a blackjack game where the player has a small edge if he plays perfectly. has this been done before?
7 2012-10-10 05:59:42 <abrkn> all zzz?
8 2012-10-10 06:02:45 <OneEyed> I've restarted a block download (it is quite fast here, SSD on a very good connection) using the git version. Why do I get a *lot* of those? ERROR: ProcessBlock() : already have block 112887 000000000000141cd081
9 2012-10-10 06:03:45 <OneEyed> Does that mean that duplicate blocks are downloaded then discarded? I have so many of them that it is not just a race condition that requested a block that I was currently receiving already, I have more than 100 blocks above this height already
10 2012-10-10 06:05:22 <OneEyed> (and nothing is downloading while my client seems to be receving those duplicate blocks)
11 2012-10-10 06:09:14 <sipa> OneEyed: it happens when a new block is being announcement while you're already doing IBD
12 2012-10-10 06:09:50 <sipa> it gets requested twice, and sonetimes a block gets downloaded twice
13 2012-10-10 06:10:02 <sipa> it shouldn't happen frewuently though
14 2012-10-10 06:10:17 <abrkn> what sort of api do people use to develop btc services now? things like batch transfers to reduce fee, etc would be nice to get pre built
15 2012-10-10 06:13:35 <sipa> abrkn: sendmany exists
16 2012-10-10 06:13:49 <abrkn> ok, so people generally go directly on the bitcoind api?
17 2012-10-10 06:53:48 <OneEyed> sipa: it happens *very* frequently
18 2012-10-10 06:54:13 <OneEyed> And it is so slow compared to the fast download of yet-unknown blocks that it dominates the time spent downloading the blockchain
19 2012-10-10 07:37:32 <UukGoblin> hi, I'm trying to send a transaction with sendmany, but getting error: {"code":-6,"message":"Account has insufficient funds"}
20 2012-10-10 07:37:54 <UukGoblin> when the account should have the exact sufficient funds available
21 2012-10-10 07:38:18 <aurigae> transaction fee?
22 2012-10-10 07:38:28 <UukGoblin> I even tried reducing the sendmany amount by 0.04 BTC to cover any possible transaction fees, but still getting "Account has insufficient funds"
23 2012-10-10 07:38:49 <aurigae> 0.05?!?
24 2012-10-10 07:38:52 <UukGoblin> also, getinfo is showing paytxfee as 0.00000000
25 2012-10-10 07:40:00 <UukGoblin> if it is about a transaction fee, can I get bitcoind to tell me how much it thinks it should be?
26 2012-10-10 07:40:13 <UukGoblin> I'd really like to avoid any fees whatsoever, as I'm mining myself
27 2012-10-10 07:40:28 <UukGoblin> I'm happy to wait until I find the next block
28 2012-10-10 07:40:34 <aurigae> As i know its about a treshhold the coinz have to be in your wallet
29 2012-10-10 07:40:55 <aurigae> otherwise you pay the 0.05 fee
30 2012-10-10 07:41:03 <UukGoblin> it's 0.05 now? :-O
31 2012-10-10 07:41:13 <aurigae> as i know, maybe im lazy
32 2012-10-10 07:41:23 <UukGoblin> it used to be 0.0005 last time I checked
33 2012-10-10 07:41:56 <aurigae> just got up, as sayd, maybe im lazy xD, its def something with 05 , not 04
34 2012-10-10 07:42:06 <kjj_> are any of your outputs very small?
35 2012-10-10 07:42:51 <aurigae> As of 10 June 2012, minimum transaction fees on the original Bitcoin client are: Accept a transaction for inclusion in a block: 0.0005 BTC Relay a transaction to other Bitcoin hosts: 0.0001 BTC A transaction can be sent without fees if both of these conditions are met: It is smaller than 10 (SI) kilobytes (10.000 bytes). All outputs are 0.01 BTC or larger.
36 2012-10-10 07:42:55 <aurigae> https://en.bitcoin.it/wiki/Transaction_fees
37 2012-10-10 07:44:25 <UukGoblin> kjj_, no
38 2012-10-10 07:44:48 <UukGoblin> kjj_, I'm trying a sendmany of 2x 50.47256191
39 2012-10-10 07:44:57 <UukGoblin> kjj_, my account balance is shown as 100.94512382
40 2012-10-10 07:45:09 <UukGoblin> and I never had such problems before the upgrade
41 2012-10-10 07:47:32 <UukGoblin> also tried sending 2x 50.45, which also gave me insufficient funds error
42 2012-10-10 07:47:46 <kjj_> you are solo mining, you said?
43 2012-10-10 07:47:52 <UukGoblin> p2pool mining
44 2012-10-10 07:48:10 <kjj_> oh. heh, you probably have hundreds or thousands of small inputs then
45 2012-10-10 07:49:38 <kjj_> you can use listunspent to get an idea of what is in your wallet
46 2012-10-10 07:59:14 <Joric> Is it possible to speed up AES using GPU? People asking about recovering their wallet.dat passwords.
47 2012-10-10 07:59:40 <_dr> i'd say using a cpu is faster than a gpu
48 2012-10-10 08:00:08 <_dr> if it has the aes-ni instruction set, that is
49 2012-10-10 08:00:52 <Joric> who even proposed to use 25k+ aes rounds? very unpractical
50 2012-10-10 08:01:42 <Joric> jk i know it was written by matt
51 2012-10-10 08:01:43 <sipa> Joric: if anything, try bruteforcing the password, not the 256-bit AES key
52 2012-10-10 08:01:56 <kjj_> _dr: even if the CPU has support for AES, a GPU that is good at mining has several hundred cores
53 2012-10-10 08:02:31 <sipa> even with 100000s iterations, that's most likely orders of magnitudes faster than trying to spend the 2^255 attempts to attack AES with a random key
54 2012-10-10 08:02:43 <_dr> kjj_: we'd have to do the math to be absoluteley sure
55 2012-10-10 08:03:07 <_dr> kjj_: but using an asic (which new cpus have) for the stuff might still be faster
56 2012-10-10 08:05:11 <kjj_> Looks like the hardware support gives about 8x speedup. If the OpenCL implementation is within an order of magnitude of non-hardware supported speed, the GPU will still win
57 2012-10-10 08:05:16 <UukGoblin> kjj_, hundreds, possibly
58 2012-10-10 08:05:28 <Joric> so you're telling AES is not GPU friendly? i honestly didn't look
59 2012-10-10 08:06:13 <UukGoblin> kjj_, 73, in particular
60 2012-10-10 08:06:27 <UukGoblin> can I get an idea of what the expected tx fee should be?
61 2012-10-10 08:06:44 <UukGoblin> why would I get insufficient funds if I have paytxfee set to 0?
62 2012-10-10 08:08:01 <pjorrit> you have insufficient funds available for spending?
63 2012-10-10 08:08:08 <kjj_> UukGoblin: you are probably over the 10k cap
64 2012-10-10 08:08:46 <UukGoblin> pjorrit, I only spend what I have, I'm not american ;-)
65 2012-10-10 08:09:07 <sipa> imho, for parallel attacks, a GPU will be faster than specialized CPU hardware instructions
66 2012-10-10 08:09:15 <sipa> but you shouldn't be attacking the AES part
67 2012-10-10 08:09:28 <UukGoblin> kjj_, so can I (a) get an idea of what sort of a transaction fee would be needed and (2) can I set the transaction to 0 fee and make sure that it's included in a block that I mine?
68 2012-10-10 08:09:40 <Joric> lol sorry it uses sha512, not aes lol sorry
69 2012-10-10 08:10:02 <Joric> sha512 can be gpu accelerated thats for sure
70 2012-10-10 08:10:27 <kjj_> UukGoblin: 1) not really, and 2) not really. you can do it with the raw transaction API
71 2012-10-10 08:10:32 <sipa> it does require 64-bit integer arithmetic, which GPU's aren't great at
72 2012-10-10 08:10:40 <UukGoblin> kjj_, and (3) can I send this as multiple smaller transactions to avoid the 10k limit?
73 2012-10-10 08:10:57 <sipa> UukGoblin: with the raw transaction api you can do anything
74 2012-10-10 08:11:01 <UukGoblin> raw transaction API! now you're talking
75 2012-10-10 08:11:03 <kjj_> UukGoblin: 3) yes. try doing two different transactions
76 2012-10-10 08:11:54 <kjj_> UukGoblin: please, please, please figure out how to use the API on testnet. I'd hate for you to have a 100 BTC "learning experience" on the real network
77 2012-10-10 08:12:04 <UukGoblin> hurr ;-P
78 2012-10-10 08:12:32 <UukGoblin> I guess I'll just keep lowering the amounts
79 2012-10-10 08:13:29 <UukGoblin> I went down to 50 and still getting insufficient funds
80 2012-10-10 08:13:47 <UukGoblin> does it mean the transaction fee would be bigger than 0.94 BTC?
81 2012-10-10 08:14:27 <kjj_> are you using accounts?
82 2012-10-10 08:15:13 <UukGoblin> accounts? I don't think I am, as I don't know what you mean
83 2012-10-10 08:16:25 <kjj_> which account are you specifying for the sendmany to come from?
84 2012-10-10 08:16:36 <UukGoblin> kjj_, ''
85 2012-10-10 08:16:47 <kjj_> doublec: bitcoind listaccounts
86 2012-10-10 08:16:59 <kjj_> make sure that all of your funds are in ""
87 2012-10-10 08:17:31 <kjj_> bah. stupid username expansion.
88 2012-10-10 08:17:40 <UukGoblin> ah. It seems about half of them is in '' and the other half (not precisely half) is in 'p2pool'
89 2012-10-10 08:18:15 <kjj_> try the move command to get them all into the same account, and then try again
90 2012-10-10 08:19:08 <UukGoblin> will that broadcast a transaction, or just issue an in-wallet move?
91 2012-10-10 08:19:28 <kjj_> in-wallet only. accounts don't really exist, they are just numbers stored locally
92 2012-10-10 08:20:21 <kjj_> move just decreases one number and increases the other
93 2012-10-10 08:20:44 <UukGoblin> okay...
94 2012-10-10 08:20:55 <UukGoblin> "" : 105.77509126,
95 2012-10-10 08:20:59 <UukGoblin> "p2pool" : 0.00000000,
96 2012-10-10 08:21:02 <UukGoblin> "test1" : -0.00000001
97 2012-10-10 08:21:12 <UukGoblin> there's a few other accounts that show 0.00000000
98 2012-10-10 08:21:22 <UukGoblin> however when running getinfo, I get:
99 2012-10-10 08:21:23 <UukGoblin> "balance" : 100.94512382,
100 2012-10-10 08:21:36 <kjj_> I still think your first sendmany will fail because you are over 10k
101 2012-10-10 08:22:01 <UukGoblin> wallet backed up... hope it's not too late
102 2012-10-10 08:22:27 <kjj_> you either have some negative accounts, or you have unconfirmed p2pool rewards
103 2012-10-10 08:22:33 <UukGoblin> kjj_, correct, a smaller sendmany now worked
104 2012-10-10 08:22:42 <UukGoblin> ah, uncofirmed pool rewards would make sense
105 2012-10-10 08:22:51 <UukGoblin> this 'test1' was some test I did a while ago...
106 2012-10-10 08:22:58 <UukGoblin> and it's still showing a negative amount
107 2012-10-10 08:23:11 <kjj_> if you do listtransactions, you should see a few immatures
108 2012-10-10 08:24:34 <sipa> UukGoblin: getinfo's result is probably the most accurate value that reflects how much you can spend now
109 2012-10-10 08:24:52 <sipa> UukGoblin: the sum of all accounts should be equal, but for recently generated coins, the semantics may differ a bit
110 2012-10-10 08:25:06 <UukGoblin> kjj_, correct, 4 immatures totalling 4.82996743, it all adds up
111 2012-10-10 08:25:21 <UukGoblin> still weird that I have -1e-8 on one account, though
112 2012-10-10 08:25:37 <UukGoblin> (and my txn ended up having 0.007 BTC as fee :-( )
113 2012-10-10 08:25:46 <kjj_> just move a satoshi into "" if it bothers you
114 2012-10-10 08:26:03 <UukGoblin> right
115 2012-10-10 08:26:27 <kjj_> er, from "" rather
116 2012-10-10 08:26:42 <UukGoblin> (what the hell, let the next miner be happy)
117 2012-10-10 08:26:57 <UukGoblin> right, maaany thanks for your help on this matter :-)
118 2012-10-10 08:27:28 <UukGoblin> I'll have to check out this Raw API later - does it allow me to include arbitrary (not-enough-fee) transactions into my memory pool?
119 2012-10-10 08:27:57 <kjj_> yes
120 2012-10-10 08:28:07 <UukGoblin> cool
121 2012-10-10 08:28:56 <kjj_> it also allows you to shoot yourself in the foot. be careful
122 2012-10-10 08:29:17 <UukGoblin> I can imagine
123 2012-10-10 08:29:21 <UukGoblin> raw powah
124 2012-10-10 08:31:16 <sipa> in particular, if you include inputs worth 100 BTC, and outputs only worth 1 BTC, you will create a transaction with 99 BTC fee :)
125 2012-10-10 08:31:35 <UukGoblin> hahah :-)
126 2012-10-10 08:31:59 <UukGoblin> I'm sure pool operators would love that
127 2012-10-10 08:32:04 <kjj_> heh. you laugh, but we think that a recent block with 100 BTC in fees might have been caused by API accidents
128 2012-10-10 08:32:13 <UukGoblin> OH IT HAPPENED
129 2012-10-10 08:32:16 <sipa> yes
130 2012-10-10 08:32:16 <UukGoblin> lulz
131 2012-10-10 08:32:29 <sipa> ozcoin miners were very happy about a triple-payout block
132 2012-10-10 08:32:52 <UukGoblin> at least it went to the miners - some pools reserve the right to take the txn fees
133 2012-10-10 08:35:20 <UukGoblin> yaay, can now spend multisigs with that raw api
134 2012-10-10 08:35:38 <UukGoblin> I've been away for a while, happy to finally see awesome changes making their way into mainstream
135 2012-10-10 08:37:18 <UukGoblin> still can't find a call to accept a txn to my memory pool
136 2012-10-10 08:38:18 <sipa> UukGoblin: sendrawtransaction
137 2012-10-10 08:39:03 <kjj_> it will get broadcast, but if you make it obnoxious enough, none of your peers will accept or relay it
138 2012-10-10 08:39:13 <UukGoblin> right
139 2012-10-10 08:39:21 <UukGoblin> but my node will remember it and include it
140 2012-10-10 08:39:29 <UukGoblin> can I also do it for someone else's transaction?
141 2012-10-10 08:40:15 <UukGoblin> I guess there's no reason it shouldn't work
142 2012-10-10 08:40:17 <UukGoblin> fun stuff
143 2012-10-10 08:40:20 <kjj_> hmm. I need to look at the mining code. I assume that it ignores all other selection criteria when looking at transactions that include your own keys
144 2012-10-10 08:40:43 <UukGoblin> I'll have to review the options for writing a peer-to-peer poker client now :-)
145 2012-10-10 10:39:19 <Evilmax> hi: please give me some btc for phoning?
146 2012-10-10 11:19:50 <drizztbsd> Evilmax: "phoning"?
147 2012-10-10 11:50:46 <MC1984> http://www.zdnet.com/the-do-not-track-standard-has-crossed-into-crazy-territory-7000005502/
148 2012-10-10 11:50:50 <MC1984> horrific
149 2012-10-10 12:12:50 <JyZyXEL> how do you create a genesis block?
150 2012-10-10 12:32:43 <asa1024> you become the Creator of the block chain
151 2012-10-10 12:35:42 <edcba> JyZyXEL: you fork bitcoin
152 2012-10-10 12:36:18 <JyZyXEL> already done
153 2012-10-10 12:37:16 <JyZyXEL> im trying to figure out the procedure for making the block now
154 2012-10-10 12:37:35 <edcba> ok
155 2012-10-10 12:37:41 <edcba> it's easy
156 2012-10-10 12:37:48 <gmaxwell> ACTION cues Luke-Jr
157 2012-10-10 12:37:59 <edcba> you just create a new block ?
158 2012-10-10 12:38:21 <edcba> with a very low difficulty
159 2012-10-10 12:38:42 <JyZyXEL> whats the easiest way to do that?
160 2012-10-10 12:39:10 <JyZyXEL> im sure the bitcoind can do that with correct parameters in the config file?
161 2012-10-10 12:39:49 <Luke-Jr> JyZyXEL: making a new scamcoin?
162 2012-10-10 12:40:12 <JyZyXEL> no, just a chain for testing
163 2012-10-10 12:40:22 <kjj_> you need to make it by hand
164 2012-10-10 12:40:25 <edcba> btw how the timekoin was working ?
165 2012-10-10 12:40:57 <JyZyXEL> kjj_: there is no script or anything to help me with it?
166 2012-10-10 12:40:58 <edcba> never understood how it could work since you can supposedly create as much identities as you want
167 2012-10-10 12:41:18 <JyZyXEL> i think bitcoind has some code to help with it
168 2012-10-10 12:41:24 <edcba> JyZyXEL you can use bitcoin client to output what you wants :)
169 2012-10-10 12:41:24 <kjj_> JyZyXEL: it was only ever done once. Satoshi might have a script to do it, but I don't think he's published it
170 2012-10-10 12:41:36 <Luke-Jr> JyZyXEL: that's what -testnet is for
171 2012-10-10 12:41:52 <JyZyXEL> Luke-Jr: thats what im making :p
172 2012-10-10 12:42:54 <JyZyXEL> edcba: yeah i did a grep -ri "genesis" on the source
173 2012-10-10 12:43:11 <Luke-Jr> JyZyXEL: it's already made!
174 2012-10-10 12:43:23 <JyZyXEL> Luke-Jr: not for the scamcoin im working on :D
175 2012-10-10 12:43:39 <JyZyXEL> we don't have a testnet blockchain :p
176 2012-10-10 12:44:56 <gmaxwell> JyZyXEL: then get support from the applicable scamcoin developers.
177 2012-10-10 13:44:45 <phantomcircuit> Recently I've seen people questioning the hosting situation for bitcoin.org
178 2012-10-10 13:45:00 <phantomcircuit> I have to say that I agree that hosting it on github is probably not the best choice
179 2012-10-10 13:45:17 <phantomcircuit> (although it isn't a bad choice)
180 2012-10-10 13:45:42 <Luke-Jr> it would probably benefit from international mirrors
181 2012-10-10 13:46:10 <Luke-Jr> but I'm not sure it's that high priority
182 2012-10-10 13:57:09 <gmaxwell> The github hosting is stinky but it works. What we should probably have it seperated hosting that pulls from github only when there are relevan signed tags.
183 2012-10-10 13:57:33 <gmaxwell> This would let us have SSL, ??? worthless as it is.
184 2012-10-10 14:40:07 <jgarzik> sipa: yep
185 2012-10-10 14:40:26 <jgarzik> sipa: -reindex is slow because it is doing a lot of reading from the same disk as it is constantly writing
186 2012-10-10 14:40:42 <jgarzik> sipa: nothing new or intrinsically special to -reindex
187 2012-10-10 14:40:50 <jgarzik> looks like it works :)
188 2012-10-10 14:57:17 <sipa> yes, judging from the code, no reason why it shouldn't
189 2012-10-10 14:58:22 <sipa> i'll try to add consistency checking to ultraprune this evening
190 2012-10-10 15:12:10 <jgarzik> sipa: does any existing -checklevel verify _everything_, including signatures?
191 2012-10-10 15:12:25 <jgarzik> ISTR you posted (on the forums?) that sigs were not verified at any checklevel
192 2012-10-10 15:13:14 <sipa> jgarzik: no
193 2012-10-10 15:13:40 <sipa> signatures are never rechecked
194 2012-10-10 15:17:00 <sipa> and iirc there was a reason while apart from script vrrification, sometimes other consistency checks can't exactly match everything verified during connection
195 2012-10-10 15:18:31 <sipa> oh yes, a transactuin that was overwritten after being fully spent
196 2012-10-10 15:18:45 <sipa> the original spending info is lost for those
197 2012-10-10 15:21:34 <jgarzik> sipa: I think verifying that 99% is better than 0%
198 2012-10-10 15:23:13 <sipa> of course
199 2012-10-10 15:37:05 <gavinandresen> Looking for more gitian signatures on the 0.7.1rc1 release.... if you have time, please build
200 2012-10-10 15:37:30 <sipa> gavinandresen: how did you manage to fix the problem with building miniupnp?
201 2012-10-10 15:37:40 <gavinandresen> I built using my kvm machine
202 2012-10-10 15:37:58 <gavinandresen> (built the deps-0.0.5 dependency)
203 2012-10-10 15:38:01 <sipa> bah
204 2012-10-10 15:38:24 <gavinandresen> agreed.
205 2012-10-10 15:38:40 <gavinandresen> I know what the problem is, but don't know how to fix it
206 2012-10-10 15:38:55 <sipa> what is the problem?
207 2012-10-10 15:39:32 <gavinandresen> installing... ummm... that package that lets you run windows binaries emulated....
208 2012-10-10 15:39:48 <sipa> wine?
209 2012-10-10 15:39:49 <gavinandresen> ... it's getting a permission denied when writing to some system directory under lxc
210 2012-10-10 15:39:55 <sipa> ah
211 2012-10-10 15:40:04 <sipa> ok, back to kvm it is
212 2012-10-10 15:42:20 <sipa> devrandom: ^ ?
213 2012-10-10 15:43:27 <gavinandresen> what is that package name... wine depends on it...
214 2012-10-10 15:44:02 <sipa> binfmt-support ?
215 2012-10-10 15:44:24 <gavinandresen> that's the one!
216 2012-10-10 15:44:43 <gavinandresen> installing it in the VM you'll see the error in the gitian var/install.log
217 2012-10-10 15:44:54 <devrandom> sipa: what's up?
218 2012-10-10 15:45:08 <sipa> devrandom: apparently binfmt-support fails to install under lxc
219 2012-10-10 15:45:20 <devrandom> uh oh
220 2012-10-10 15:46:16 <gavinandresen> https://bugs.launchpad.net/ubuntu/+source/binfmt-support/+bug/917660
221 2012-10-10 15:46:58 <gavinandresen> ... looks like a similar issue
222 2012-10-10 15:49:09 <devrandom> gavinandresen: should we start using precise as the build env, or will that cause other issues?
223 2012-10-10 15:49:54 <sipa> devrandom: for windows builds, moving to a more recent build env only has advantages, imho
224 2012-10-10 15:50:14 <sipa> for ubuntu builds, i'm afraid that something built under precise will not run on pre-precise
225 2012-10-10 15:50:26 <devrandom> good point
226 2012-10-10 15:50:46 <gavinandresen> right, linux builds we want to use old and crusty for max compatibility. Windows builds we can use whatever works.
227 2012-10-10 15:51:17 <sipa> for example, -flto would be nice to reduce the size of binaries, but is currently not possible because of the ancient gcc 4.2 used in mingw under lucid
228 2012-10-10 15:51:19 <devrandom> perhaps install binfmt-support without post-install triggers for now
229 2012-10-10 15:51:30 <sipa> devrandom: will it work, then?
230 2012-10-10 15:51:52 <devrandom> I'll look into this more later today - I have a meeting to go to
231 2012-10-10 15:51:57 <gavinandresen> devrandom: thanks!
232 2012-10-10 15:52:01 <sipa> and in leveldb, i had to backport part of the windows code to run on pre-g++0x
233 2012-10-10 15:52:16 <sipa> which wouldn't be necessary if we could build the windows one on a more recent gcc
234 2012-10-10 15:52:26 <sipa> eh, c++0x
235 2012-10-10 16:05:46 <sipa> anyone having problems fetching from github?
236 2012-10-10 16:05:55 <sipa> oh nevermind, it just succeeded
237 2012-10-10 16:10:29 <edcba> Laszlo Hanyecz, a Florida programmer, conducted what bitcoiners think of as the first real-world bitcoin transaction, paying 10,000 bitcoins to get two pizzas delivered from Papa John???s.
238 2012-10-10 16:10:32 <edcba> muhahaha
239 2012-10-10 16:10:57 <edcba> those pizzas aren't cheaps :)
240 2012-10-10 16:12:12 <sipa> even back then, they weren't cheap
241 2012-10-10 16:12:24 <sipa> but he had to spend his mining income on something :p
242 2012-10-10 16:13:15 <edcba> i wonder if receiver has still the coins...
243 2012-10-10 16:13:45 <edcba> so much fortune lying on hard disks of bored ppl somewhere :)
244 2012-10-10 16:14:10 <midnightmagic> are these instructions still accurate for HEAD gitian build? https://gist.github.com/806265
245 2012-10-10 16:14:51 <edcba> you mean we can build bitcoin with 1 command now ?
246 2012-10-10 16:15:04 <sipa> midnightmagic: wx isn't used anymore since 0.5
247 2012-10-10 16:15:18 <sipa> so i guess not :)
248 2012-10-10 16:15:35 <sipa> edcba: in normal environments, with dependencies installed: yes
249 2012-10-10 16:15:39 <sipa> gitian is a lot harder though
250 2012-10-10 16:15:46 <edcba> with dependancies installed...
251 2012-10-10 16:15:50 <edcba> :)
252 2012-10-10 16:16:03 <edcba> bdb is still there i guess
253 2012-10-10 16:16:17 <sipa> yes, for now
254 2012-10-10 16:16:21 <midnightmagic> sipa: Aside from that, is there an instruction set somewhere for gitian builds which is followed, or does each dev figure it out on his own and cackel manically?
255 2012-10-10 16:16:38 <edcba> can't we switch to something more mainstream ?
256 2012-10-10 16:17:07 <sipa> edcba: if you know a mainstream framework for multi-platform deterministic building... please :)
257 2012-10-10 16:17:23 <edcba> no i meant for the db
258 2012-10-10 16:17:31 <sipa> yes, we'll switch to leveldb in 0.8
259 2012-10-10 16:17:45 <midnightmagic> https://github.com/devrandom/gitian-builder <- readme.md maybe.. hrm
260 2012-10-10 16:18:02 <midnightmagic> :-)
261 2012-10-10 16:19:04 <midnightmagic> just fyi, this page here: https://gitian.org/ links to the wxwidgets based gitian instructions for building bitcoin.
262 2012-10-10 16:20:03 <edcba> ok the sample is bitcoin...
263 2012-10-10 16:21:30 <sipa> gavinandresen: ok, i386 vm built, now amd64 one...
264 2012-10-10 16:22:46 <midnightmagic> edcba: No, it's the wrong sample, with incorrect instructions. So, the main gitian webpage is out-of-date.
265 2012-10-10 16:28:24 <edcba> https://www.iana.org/assignments/uri-schemes/prov/bitcoin
266 2012-10-10 16:28:41 <edcba> now we only need our bitcoin currency character in unicode :)
267 2012-10-10 16:35:17 <sipa> where can i find Qt 4.8.2 ?
268 2012-10-10 16:36:02 <sipa> on the website there is 4.8.3, and in the FTP archive there is 4.8.0
269 2012-10-10 16:37:57 <sipa> ok, took the link for 4.8.3 and replaced the 3 by a 2...
270 2012-10-10 16:42:38 <forrestv> what's the maximum size transaction the satoshi client will generate?
271 2012-10-10 16:45:28 <gmaxwell> 100k.
272 2012-10-10 16:45:51 <gmaxwell> There is some test in the transaction logic that gives up if it gets a result bigger than that.
273 2012-10-10 16:46:08 <devrandom> gavinandresen: confirmed that binfmt-support works with a precise guest
274 2012-10-10 16:46:15 <jgarzik> the max size we accept is MAX_BLOCK_SIZE, though, right?
275 2012-10-10 16:46:37 <jgarzik> ACTION wonders if testnet contains a 1MB, 2-tx block :)
276 2012-10-10 16:47:08 <gmaxwell> jgarzik: I generated some really big ones, but I'm not sure if I hit that limit.
277 2012-10-10 16:47:29 <gmaxwell> forrestv:
278 2012-10-10 16:47:30 <gmaxwell> return false;
279 2012-10-10 16:47:45 <gmaxwell> main.h:static const unsigned int MAX_BLOCK_SIZE_GEN = MAX_BLOCK_SIZE/2;
280 2012-10-10 16:47:46 <forrestv> gmaxwell, thanks
281 2012-10-10 16:47:52 <gavinandresen> devrandom: ok. So if we wanted to switch to compiling on precise, that means everybody would need three make_base_vm's (lucid 32- and 64-bit, and a precise for doing the windows build) ?
282 2012-10-10 16:47:56 <gmaxwell> main.h:static const unsigned int MAX_BLOCK_SIZE = 1000000;
283 2012-10-10 16:48:22 <devrandom> gavinandresen: yes, that is correct
284 2012-10-10 16:48:23 <gmaxwell> forrestv: Might I ask why you're asking?
285 2012-10-10 16:48:29 <devrandom> I didn't test a full build yet
286 2012-10-10 16:48:41 <gmaxwell> forrestv: I suspect but haven't tested, that you an create a larger one with the raw txn API.
287 2012-10-10 16:49:15 <sipa> max level size for a transaction is 1M iirc
288 2012-10-10 16:49:21 <sipa> *legal
289 2012-10-10 16:49:51 <forrestv> gmaxwell, limits for transactions forwarded ahead of time over the p2pool network
290 2012-10-10 16:51:05 <midnightmagic> gavinandresen: what's the commitid you gyus are building to? 0a4e67afadd738151ae1cd4468b5422a21a0eabf ?
291 2012-10-10 16:52:02 <gavinandresen> midnightmagic: tag v0.7.1rc1
292 2012-10-10 16:52:17 <gavinandresen> midnightmagic: git fetch --tags if you don't have it
293 2012-10-10 16:52:44 <nym> Bitfloor's WF account is down
294 2012-10-10 16:54:18 <sipa> WF?
295 2012-10-10 16:58:29 <midnightmagic> gavinandresen: okie dokie.
296 2012-10-10 17:04:40 <sipa> gavinandresen: ok, Qt built
297 2012-10-10 17:04:42 <sipa> now deps...
298 2012-10-10 17:15:05 <nym> sipa: wells fargo
299 2012-10-10 17:16:00 <Luke-Jr> devrandom: everything seems to work fine on an older gitian, fyi
300 2012-10-10 17:36:48 <gmaxwell> hah. I had a bitcoind hit a peak logging rate of 2316 lines in a second.
301 2012-10-10 17:37:00 <gmaxwell> 2316 10/04/12 17:26:51
302 2012-10-10 17:37:28 <chmod755> lol
303 2012-10-10 17:38:04 <gmaxwell> 7 other seconds with over 2k messages per second in that one's node.
304 2012-10-10 17:38:05 <gmaxwell> er log
305 2012-10-10 17:38:23 <chmod755> gmaxwell, what are you trying to find in your logs?
306 2012-10-10 17:38:47 <gmaxwell> I'm not trying to find anything.
307 2012-10-10 17:51:46 <jgarzik> colored coins python script, https://bitcointalk.org/index.php?topic=117630.0 -> http://pastebin.com/h103Yj3N (bitpaint.py)
308 2012-10-10 17:52:01 <jgarzik> uses RPC interface
309 2012-10-10 17:52:28 <jgarzik> raw transaction API, specifically
310 2012-10-10 17:54:36 <jgarzik> sigh. by hesitating on pybond, everyone will wind up using stuff less optimally implemented, harder to track stuff
311 2012-10-10 17:55:02 <nym> gmaxwell: do you run any sites?
312 2012-10-10 17:56:46 <gmaxwell> jgarzik: welp ultraprune is gonna break that one pretty good
313 2012-10-10 17:56:49 <gmaxwell> nym: hm? Why?
314 2012-10-10 18:03:11 <nym> gmaxwell: i'm trying to network :)
315 2012-10-10 18:03:23 <gmaxwell> Ah. Okay. No.
316 2012-10-10 18:03:56 <nym> ACTION just checked out your github
317 2012-10-10 18:04:16 <gmaxwell> Oh I only use github for staging to bitcoin.
318 2012-10-10 18:04:42 <nym> ah
319 2012-10-10 18:04:48 <nym> so you're a contributor?
320 2012-10-10 18:05:16 <gmaxwell> Yes.
321 2012-10-10 18:05:23 <nym> ++!
322 2012-10-10 18:05:26 <Diablo-D3> hes a developer.
323 2012-10-10 18:05:58 <nym> gmaxwell: I really like this: https://en.bitcoin.it/wiki/Agents
324 2012-10-10 18:06:41 <MC1984> no ones ever gonna do anything with that agents thing
325 2012-10-10 18:06:54 <gmaxwell> I think TD (Mike) wrote most of that page, though at least I inspired some of it.
326 2012-10-10 18:07:05 <MC1984> which is a shame cos im reading a book where humanity is usurped by autonomous corporate entities
327 2012-10-10 18:07:06 <Diablo-D3> http://arstechnica.com/gaming/2012/10/wing-commanders-chris-roberts-returns-to-gaming-with-ambitious-mmo/
328 2012-10-10 18:07:07 <gmaxwell> MC1984: people will, but it has to become easy enough first.
329 2012-10-10 18:07:09 <Diablo-D3> do want all over the place
330 2012-10-10 18:07:19 <gmaxwell> MC1984: Daemon?
331 2012-10-10 18:07:33 <MC1984> accelerando
332 2012-10-10 18:08:41 <gmaxwell> hm. I am almost certant that I've read that, but I don't recall it at all.
333 2012-10-10 18:09:23 <gmaxwell> MC1984: you should read Daemon and Freedom by Daniel Suarez if you think the agents idea is fun. (Jeff recommended it to me after the storj post and I enjoyed them a lot)
334 2012-10-10 18:09:23 <MC1984> manfred manx, his queen of a jovian moon daughter and the cat
335 2012-10-10 18:09:25 <jgarzik> Thinking about agents inevitable boils down to, for me, provable execution by untrusted parties
336 2012-10-10 18:09:41 <jgarzik> *inevitably
337 2012-10-10 18:09:46 <MC1984> gmaxwell gotcha
338 2012-10-10 18:10:08 <gmaxwell> jgarzik: Yea, you can't have that. But whats really provable in this world? Making failure tracable and having recourse is often more pratical.
339 2012-10-10 18:10:08 <jgarzik> People will surely know they are a really part of an autonomous entity
340 2012-10-10 18:10:26 <jgarzik> And that makes the autonomous entity not so autonomous
341 2012-10-10 18:10:45 <gmaxwell> No man is an island.
342 2012-10-10 18:10:56 <gmaxwell> Why should the robots be any different? :P
343 2012-10-10 18:10:58 <jgarzik> ACTION was trying to think along the lines of a bytecode engine that would execute signed code
344 2012-10-10 18:11:14 <MC1984> thats why they will organise themselves into a hive mind
345 2012-10-10 18:11:16 <amiller> nym, i strongly dislike the agents page, i've pitched a fit about it before (see the talk page)
346 2012-10-10 18:11:16 <jgarzik> do that in parallel, and at least you prove the inputs and outputs
347 2012-10-10 18:12:02 <gmaxwell> amiller: I think you assumed a much stronger reading of it than almost anyone else would. Even as someone familar with the same science as you in that space I didn't take that reading. :P
348 2012-10-10 18:12:34 <gmaxwell> Though I do agree with your concern that people are too quick to believe things like usefully provable/tamperproof computation is possible.
349 2012-10-10 18:12:48 <amiller> nym if you have a moment i'm going to interview you in a pm and see what you think the page is talking about
350 2012-10-10 18:12:56 <nym> k
351 2012-10-10 18:14:47 <jgarzik> ACTION races off to grab ProvablyFairSociety.{com,org,net} and Daemon2016.com
352 2012-10-10 18:14:54 <jgarzik> j/k
353 2012-10-10 18:15:18 <MC1984> join the venus project
354 2012-10-10 18:15:21 <MC1984> or something
355 2012-10-10 18:15:42 <gmaxwell> amiller: neat idea.
356 2012-10-10 18:16:13 <MC1984> how long could an agent hide itself?
357 2012-10-10 18:16:27 <MC1984> moving around between vps and such
358 2012-10-10 18:16:35 <jgarzik> MC1984: well, an agent cannot start without being bootstrapped by humans
359 2012-10-10 18:16:47 <MC1984> ofc
360 2012-10-10 18:16:49 <jgarzik> MC1984: you are limited by the automated APIs, and mechanical turks
361 2012-10-10 18:17:25 <jgarzik> MC1984: patterns of behavior quickly become obvious: the same text posting to the same set of forums, soliciting bids for hosting API plugins
362 2012-10-10 18:17:57 <jgarzik> even varying each of those variables, a pattern of resource consumption is inevitable
363 2012-10-10 18:18:09 <gmaxwell> I proposed one solution to that, which is that the human who bootstraps an agent must basically provide a bond for 100% of the startup costs. The agent will pay back with interest if successful, with the idea of making cheating unprofitable.
364 2012-10-10 18:18:23 <MC1984> what about polymorphism
365 2012-10-10 18:18:44 <MC1984> learning as it goes
366 2012-10-10 18:18:52 <jgarzik> or simple bot bug fixing
367 2012-10-10 18:19:07 <jgarzik> who sees the source code? who approves changes? who compiles the changes?
368 2012-10-10 18:21:21 <jgarzik> that's why I continue to flirt with bytecode engine approaches. the DNA, the brain is the bytecode stored in the entity's memory. the only thing a host needs to run is the base bytecode engine.
369 2012-10-10 18:21:33 <gmaxwell> jgarzik: yea, my solution to that was to follow the biological model. You don't fix bots. They have children which are potentially fixed. The code is fixed at birth with only limited flexiblity.
370 2012-10-10 18:21:59 <MC1984> isnt that what it says on the agents page
371 2012-10-10 18:22:10 <MC1984> genetic algorithms
372 2012-10-10 18:22:20 <gmaxwell> MC1984: Thats nice, but I didn't write that.
373 2012-10-10 18:22:29 <gmaxwell> 'genetic algorithms' ?
374 2012-10-10 18:22:39 <gmaxwell> Utterly useless for this in my opinion.
375 2012-10-10 18:23:12 <gmaxwell> By biological model I simply mean that the primary code is fixed for an orginism (as it is in all higher order life on earth)
376 2012-10-10 18:23:24 <jgarzik> given that computing resources are potentially long lived, you either want some concept of self-healing (upgrading) or parasitic (start child, commit suicide)
377 2012-10-10 18:23:30 <MC1984> oh
378 2012-10-10 18:24:00 <jgarzik> e.g. VPS and hosting accounts will live longer than processes
379 2012-10-10 18:24:10 <jgarzik> and will exist longer than parents _should_, in times of upgrade
380 2012-10-10 18:24:47 <jgarzik> I like the model of each process as a biological cell
381 2012-10-10 18:24:49 <MC1984> is this agents thing actualy describing sort of weak AI then or am i confused
382 2012-10-10 18:24:52 <jgarzik> those pop in and out of existence rapidly
383 2012-10-10 18:25:03 <jgarzik> cells have DNA (bytecode corpus)
384 2012-10-10 18:25:40 <jgarzik> MC1984: basically correct. I think the term is "narrow AI" -- something more scripted and dumb, not thinking or reasoning
385 2012-10-10 18:25:58 <jgarzik> like AI players in computer games
386 2012-10-10 18:26:12 <gmaxwell> Sufficiently advanced software is indistinguishable from intelligent. :P
387 2012-10-10 18:26:18 <MC1984> but if they can self select good traits
388 2012-10-10 18:27:22 <MC1984> whats the chances of packing an agent off into the internet with 10btc in its pocket and coming back in a few decades to find skynet
389 2012-10-10 18:27:39 <gmaxwell> MC1984: with the kind of stuff we're talking about, about zero.
390 2012-10-10 18:27:57 <MC1984> aw
391 2012-10-10 18:28:13 <jgarzik> MC1984: helpfully, the US military is already building skynet anyway ;p
392 2012-10-10 18:28:21 <MC1984> but it would be a great leap in AI research anyway right
393 2012-10-10 18:28:46 <gmaxwell> It's not bad news. Non-human intelligence will very likely be very very unfriendly from our perspective. It would be bad if that could happen in an uncontrolled way.
394 2012-10-10 18:28:51 <MC1984> ive always thought one way or another AIs would build themselves rather than us constructing them
395 2012-10-10 18:29:00 <MC1984> we just provide time and processing
396 2012-10-10 18:29:10 <edcba> maybe we will construct them
397 2012-10-10 18:29:14 <edcba> ...by mistake :)
398 2012-10-10 18:29:32 <jgarzik> Daniel Suarez' latest book _Kill Decision_ is about swarms of drones, built based upon biological models of behavior. One example used is weaver ants, individually unintelligent, but with very aggressive emergent/hive behaviors.
399 2012-10-10 18:29:36 <MC1984> if you take the universe as a simulation, then thats how we came about anyway
400 2012-10-10 18:29:41 <MC1984> even if you dont take it as a sim
401 2012-10-10 18:29:59 <jgarzik> The book posits weaver ant behavior in a colony of military drones-gone-wild
402 2012-10-10 18:30:02 <gmaxwell> edcba: optimization processes are prone to unexpected outcomes. Dangerous stuff if you let them have real world results.
403 2012-10-10 18:30:37 <jgarzik> HFT has plenty of unintended real world outcomes
404 2012-10-10 18:31:04 <MC1984> hmm hive mind
405 2012-10-10 18:31:11 <jgarzik> and with the Financialization of Everything being quite inevitable thanks to general march of technology, fun implications
406 2012-10-10 18:31:40 <edcba> financialization is optimisation
407 2012-10-10 18:31:46 <gmaxwell> edcba: an example I like to give is??? for example, one time I was working on an antenna design and I made a parameterized model??? you give it the dimensions it simulates it, returns the performance. Then I stuck that in an optimizer and it went along tweaking the parameters looking for improvements...
408 2012-10-10 18:32:31 <gmaxwell> edcba: later I returned and was quite excited to see it had made enormous improvements .. only to find out that it had managed to make the lengths of some parts _negative_. "oops"
409 2012-10-10 18:32:56 <edcba> ACTION goto buy some negative length antennas
410 2012-10-10 18:32:57 <MC1984> cool subspace antenna
411 2012-10-10 18:33:24 <gmaxwell> almost every single _simple_ optimization problem I've worked with has exposed holes in the objective function, I can't imagine how hard it would be to get the objective functions of an AI system right so that it doesn't decide that is isn't best to kill everyone.
412 2012-10-10 18:33:37 <edcba> haha
413 2012-10-10 18:34:16 <edcba> remind me of some guy who thought the future of programming would be to evolve programs through AI...
414 2012-10-10 18:34:35 <MC1984> gmaxwell we cant even make it so that the original AI (us) doesnt decide thats a good course of action
415 2012-10-10 18:34:36 <gmaxwell> "Maximize average happiness Oh Multivac" {multivac kills everyone except one guy, and keeps him constantly on morphine}
416 2012-10-10 18:34:58 <Eliel_> gmaxwell: could even be that the more you try to prevent that decision, the harder it becomes :D
417 2012-10-10 18:35:00 <gmaxwell> MC1984: it's worse, not only will computers do evil, they'll do insane evil.
418 2012-10-10 18:35:04 <edcba> but that guy must be really happy !
419 2012-10-10 18:35:09 <Diablo-D3> gmaxwell: erm
420 2012-10-10 18:35:13 <Diablo-D3> Ive seen that joke somewhere
421 2012-10-10 18:35:22 <Diablo-D3> but it was god, not multivac
422 2012-10-10 18:35:38 <jgarzik> or to take a scene from _Daemon_, a directive to "end spam" results in the machine gunning of all spammers ;p
423 2012-10-10 18:35:46 <Diablo-D3> man
424 2012-10-10 18:35:47 <MC1984> gmaxwell reminds me of a historical account of a billion years of galactic evolution
425 2012-10-10 18:35:50 <Diablo-D3> you know what we should be doing?
426 2012-10-10 18:35:51 <gmaxwell> jgarzik: some insanity is a good idea.
427 2012-10-10 18:35:54 <MC1984> pdf thing i read
428 2012-10-10 18:36:00 <edcba> jgarzik: may work finally !
429 2012-10-10 18:36:03 <jgarzik> hehehe
430 2012-10-10 18:36:14 <Diablo-D3> trying to advance real AI using distributed computing
431 2012-10-10 18:36:23 <MC1984> some species uploaded and proceeded to do things like blockade the stars of other species start systems and such
432 2012-10-10 18:36:25 <MC1984> nice stuff
433 2012-10-10 18:36:26 <Diablo-D3> an open source Daemon
434 2012-10-10 18:36:37 <jgarzik> Diablo-D3: would be fun :)
435 2012-10-10 18:36:46 <jgarzik> Diablo-D3: as long as it's not written in Java
436 2012-10-10 18:36:48 <Diablo-D3> yes, especially when I sudo a sandwich
437 2012-10-10 18:37:06 <jgarzik> ACTION has been looking at Parrot VM
438 2012-10-10 18:37:14 <Diablo-D3> man
439 2012-10-10 18:37:16 <Diablo-D3> if Im going to do a vm
440 2012-10-10 18:37:17 <jgarzik> though really, a purpose-built VM would probably be needed
441 2012-10-10 18:37:18 <Diablo-D3> Ill write my own
442 2012-10-10 18:37:31 <jgarzik> built from the ground up to sign/verify code
443 2012-10-10 18:37:36 <Diablo-D3> vms try too hard to actually be machines instead of dumb assemblers
444 2012-10-10 18:38:25 <gmaxwell> MC1984: A biological example I've seen is the expirements in group selection. Artifically select flour beetle populations that maintain small size... Do they evolve to restrain their birthrates? No. They evolve to become canibals that eat offspring of others of their kind.
445 2012-10-10 18:40:08 <jgarzik> Definition: virtual machine cell... a simple virtual machine, simple key/value store limited to X size, and basic crypted network I/O ability. plus a list of parent's public keys.
446 2012-10-10 18:40:44 <jgarzik> maybe a single local fs file, as sandboxed disk storage
447 2012-10-10 18:43:50 <jgarzik> "cell death" == process ends, due to bug, remote command or system event
448 2012-10-10 18:44:21 <jgarzik> "cell birth" == process starts, associates itself with a group of cells, and downloads initial instructions
449 2012-10-10 18:45:19 <jgarzik> the program code could remain static during cell lifetime, if that is added protection
450 2012-10-10 18:47:10 <MC1984> what i want to know is if strong AI is possible where are they
451 2012-10-10 18:47:35 <jgarzik> MC1984: a strong AI would be indistinguishable from a human, wouldn't it?
452 2012-10-10 18:47:46 <MC1984> where is the vanguard invasion fleet of von neumann machines eating jupiter, why arnt the stars going out as dyson shells go around them
453 2012-10-10 18:48:08 <jgarzik> how do you know Ben Bernanke isn't _already_ our robotic overlord?
454 2012-10-10 18:48:18 <MC1984> fffffffffffffffffffffff
455 2012-10-10 18:49:13 <MC1984> scarier thought, perhaps we really are the first evolution of intelligent life within a very wide radius since the big bang
456 2012-10-10 18:49:23 <MC1984> and thus we really do only have each other for comapny
457 2012-10-10 18:49:41 <MC1984> the sun is only in the second gen of stellar formation right?
458 2012-10-10 18:50:47 <MC1984> so if there is to be a self replicator phage on this universe, it has to be us that kicks it off lol
459 2012-10-10 18:51:22 <MC1984> i wouldnt put it past our species at all
460 2012-10-10 18:51:36 <MC1984> the general misanthropy you get after reading lots of sci fi
461 2012-10-10 18:52:21 <jgarzik> or maybe a skynet would decide that humanity all needs large penii, recreational pharma and additional reality TV shows
462 2012-10-10 18:53:09 <gmaxwell> jgarzik: in the uniform space of all ideas there are a great many really bad ones than mildly bad ones! :P
463 2012-10-10 18:53:47 <gmaxwell> so your AI need to be selected out of a very specific subset of all possible AI's in order to have objectives that are merely mildly bad. :P
464 2012-10-10 18:54:10 <MC1984> funny because it seems reality TV has gotten far enough where you can watch a pretty young blonde drink a quart of donkey cum for money on TV
465 2012-10-10 18:54:17 <MC1984> perhaps that is skynets doing
466 2012-10-10 18:54:20 <gmaxwell> E.g. there is an infinite number of possible AI's who only high level goal is to convert the whole universe into X (for any possible X).
467 2012-10-10 18:54:39 <gmaxwell> s/who/whos/
468 2012-10-10 18:56:01 <edcba> i doubt it's infinite
469 2012-10-10 18:56:13 <edcba> it epends on how much information is inside our universe :)
470 2012-10-10 18:56:32 <MC1984> i beleive theres an estimate for that somewhere
471 2012-10-10 18:57:01 <gmaxwell> "Good news, we've found alien 'life'. Bad-news, it's only interested in converting all our mass into flemdars." "Whats a flemdar?" "oh it's some kind of psychic sex toy for a species that died out a billion years ago (they were all converted into flemdars by an over powered factory that sent these ships)"
472 2012-10-10 18:58:37 <sipa> hmm, i wonder... given a block's added txouts... how many blocks do you need after it to be +- certain one of them gets spent?
473 2012-10-10 18:58:42 <sipa> i suppose hardly more than 1
474 2012-10-10 18:59:00 <edcba> ?
475 2012-10-10 18:59:24 <edcba> ACTION doesn't understand the question
476 2012-10-10 18:59:53 <sipa> because of how i intend to make ultraprune's consistency check work: roll back N blocks, and check that the undo data exactly matches the utxo set state
477 2012-10-10 19:00:04 <Diablo-D3> [04:57:01] <gmaxwell> "Good news, we've found alien 'life'. Bad-news, it's only interested in converting all our mass into flemdars." "Whats a flemdar?" "oh it's some kind of psychic sex toy for a species that died out a billion years ago (they were all converted into flemdars by an over powered factory that sent these ships)"
478 2012-10-10 19:00:15 <Diablo-D3> but I already get women off.
479 2012-10-10 19:00:44 <sipa> edcba: i want to know: does every block spend outputs from the previous block
480 2012-10-10 19:01:07 <sipa> probably not
481 2012-10-10 19:01:08 <sipa> if not
482 2012-10-10 19:01:21 <sipa> does every block spend outputs from one of the two previous blocks
483 2012-10-10 19:01:44 <edcba> i don't understand what you mean by 'spent' then
484 2012-10-10 19:01:50 <sipa> redeemed
485 2012-10-10 19:01:51 <sipa> consumed
486 2012-10-10 19:01:55 <sipa> used as input
487 2012-10-10 19:02:25 <edcba> i don't see why they would need to be spent in any given time frame ?
488 2012-10-10 19:03:19 <sipa> at startup, i intend to make ultraprune (my rewrite of the validation engine) verify its state, by trying to undo/disconnect a few of the last blocks
489 2012-10-10 19:04:00 <sipa> if every block always spends outputs from the previous block, then any inconsistency caused by a partial write will always result in a failure during that disconnection
490 2012-10-10 19:04:40 <MC1984> http://www.theconservativefoundation.co.uk/index.php?page=who&win=board&item=9
491 2012-10-10 19:04:41 <MC1984> lol
492 2012-10-10 19:04:45 <MC1984> leejin
493 2012-10-10 19:05:16 <edcba> sorry i don't understand your assumption of block spending outputs
494 2012-10-10 19:05:27 <sipa> it's not an assumption
495 2012-10-10 19:05:31 <sipa> it's a question
496 2012-10-10 19:05:51 <sipa> ok, i'll try to explain
497 2012-10-10 19:05:58 <sipa> assume there are 3 blocks, A-B-C
498 2012-10-10 19:06:06 <sipa> somewhere in the chain
499 2012-10-10 19:06:11 <edcba> a block validates some tx, receivers then may or may not spent them ?
500 2012-10-10 19:06:27 <sipa> you know how a bitcoin transaction works?
501 2012-10-10 19:06:36 <edcba> yeah i think i remember :)
502 2012-10-10 19:06:52 <edcba> some inputs some outputs some scripts
503 2012-10-10 19:07:03 <sipa> and in particular, what does an input contain?
504 2012-10-10 19:08:12 <edcba> ok you mean total(inputs) = total(outputs) ?
505 2012-10-10 19:08:26 <sipa> that's not true
506 2012-10-10 19:08:40 <sipa> total(inputs) >= total(outputs)
507 2012-10-10 19:08:47 <sipa> the difference becomes fee
508 2012-10-10 19:09:04 <edcba> hmm
509 2012-10-10 19:09:07 <sipa> but that's not what i mean; what exactly, in the bytes of a transaction, does an input consist of?
510 2012-10-10 19:09:20 <sipa> you can guess :)
511 2012-10-10 19:09:36 <edcba> should be the input of script but i don't remmeber looking at that
512 2012-10-10 19:09:55 <sipa> it does contain a script yes, but that's not the important part
513 2012-10-10 19:10:11 <sipa> it contains the hash of a previous transaction, and a numer referring to one of its outputs
514 2012-10-10 19:10:22 <sipa> *number
515 2012-10-10 19:10:28 <edcba> ok
516 2012-10-10 19:10:43 <sipa> so one way to look at it is that a transaction has some outputs, which i'll call "coins" now
517 2012-10-10 19:10:48 <edcba> 'address' of previous output
518 2012-10-10 19:10:54 <sipa> NO!
519 2012-10-10 19:10:58 <sipa> *not* an address
520 2012-10-10 19:11:09 <edcba> 'pointer' of previous output
521 2012-10-10 19:11:13 <edcba> reference :)
522 2012-10-10 19:11:17 <sipa> right, it's some kind of address
523 2012-10-10 19:11:26 <sipa> just wanted to stress that it's certainly not a bitcoin address
524 2012-10-10 19:11:51 <edcba> yeah i tried to avoid a tech term and failed at bitcoin term :)
525 2012-10-10 19:11:58 <sipa> each coin gets assigned to an output script (which typically corresponds to a bitcoin address, but not necessarily)
526 2012-10-10 19:12:38 <sipa> and transaction consume coins from previous transactions (by proving ownership via a crypto signature), melt them together, split them again, and re-assign the outputs
527 2012-10-10 19:12:43 <edcba> but if C points to output A.1 it won't necessarily consumes A.2
528 2012-10-10 19:12:48 <sipa> exactly
529 2012-10-10 19:12:54 <sipa> but it always consumes an entire output
530 2012-10-10 19:13:07 <sipa> not an entire transaction, but always one or more integral outputs of it
531 2012-10-10 19:13:18 <sipa> (or outputs from multiple separate transactions)
532 2012-10-10 19:13:20 <edcba> yes since we want it to recreate outputs
533 2012-10-10 19:13:26 <edcba> (even if same address)
534 2012-10-10 19:13:29 <sipa> right
535 2012-10-10 19:13:42 <sipa> the point is that this way, bitcoin internally doesn't need to keep track of balances
536 2012-10-10 19:14:04 <sipa> it only needs to check that no output is spent twice, and that the value of inputs always is enough to cover the outputs
537 2012-10-10 19:15:40 <sipa> so in practice what you need to verify whether a transaction is valid, is just a list of unspent transaction outputs
538 2012-10-10 19:15:56 <sipa> (and a bit more, but that's the most important piece of information)
539 2012-10-10 19:16:32 <sipa> a block, being a list of transaction, can be seen as a "patch" to this set of unspent outputs - its transactions consume some outputs, and produce some new ones
540 2012-10-10 19:16:35 <sipa> ok?
541 2012-10-10 19:17:30 <edcba> hmm yes
542 2012-10-10 19:18:17 <sipa> the current bitcoin validation engine doesn't work this way
543 2012-10-10 19:18:36 <sipa> it keeps a concatenation of all blocks around, and an index with pointers into it for every transaction
544 2012-10-10 19:18:47 <sipa> and for each of those transactions, where they get spent
545 2012-10-10 19:19:12 <sipa> the problem with this is that it needs access to the entire history + that entire index the whole time
546 2012-10-10 19:20:09 <edcba> i remember there was mention of pruning the tx tree somewhere in paper and maybe in code too (TODO ? :)
547 2012-10-10 19:20:41 <sipa> yes, but that's not exactly what i'm talking about
548 2012-10-10 19:21:19 <edcba> ACTION downloads source code...
549 2012-10-10 19:21:56 <edcba> 13MB .zip ?
550 2012-10-10 19:22:04 <edcba> wtf happened :)
551 2012-10-10 19:22:48 <gmaxwell> edcba: the zip has binaries.
552 2012-10-10 19:23:03 <gmaxwell> nice fat static ones that include boost and such.
553 2012-10-10 19:23:12 <gmaxwell> if you just want the source you should use git.
554 2012-10-10 19:23:30 <gmaxwell> sipa: the replay for the consistency check is a nice idea.
555 2012-10-10 19:24:32 <sipa> it won't detect random corruptions, but undoing a few thousand transactions' worth of blocks (in memory only) should detect any partial write or recent data corruption at least
556 2012-10-10 19:25:10 <sipa> not that i expect that to do match... leveldb's data blocks are checksummed, and logs consist of atomically-written blocks
557 2012-10-10 19:25:34 <sipa> but still, application-level consistency checks will certainly make people feel better
558 2012-10-10 19:29:50 <amiller> sipa, i want to ask you about the goals and security model for logdb
559 2012-10-10 19:30:08 <amiller> my main question is why use a hash chain rather than error correction, since for storing wallet keys i think you care more about recoverability rather than integrity
560 2012-10-10 19:31:32 <sipa> i don't think it's the job of the application to provide error recovery at that level
561 2012-10-10 19:31:57 <sipa> hard drives very rarely have random corruptions, since they do checksumming and error correction by themself already these days
562 2012-10-10 19:32:15 <sipa> if there's a problem with a hard drive, you want to have a backup
563 2012-10-10 19:33:02 <sipa> and the logdb model makes sure you always have a consistent wallet, but potentially an older state
564 2012-10-10 19:33:26 <sipa> combined with a key pool or deterministic keys, that shouldn't cause monetary loss
565 2012-10-10 19:34:06 <amiller> what's an inconsistent wallet?
566 2012-10-10 19:34:19 <sipa> a missing transaction, for example
567 2012-10-10 19:34:29 <sipa> of an incorrect privkey-pubkey pair
568 2012-10-10 19:34:48 <sipa> or who-knows-what arbitrary data
569 2012-10-10 19:37:38 <amiller> hm, i hadn't thought about the role of wallet integrity in terms of deterministic keys so i should probably think about that more, i bet that's a good reason to have the hash chain
570 2012-10-10 19:38:08 <sipa> i suppose having error-correction on top of that is not a bad idea
571 2012-10-10 19:38:30 <sipa> but i consider a wallet to be valuable, more than just its private keys
572 2012-10-10 19:38:42 <edcba> ok so you want to modify accept/GetOutputFor/FetchInputs ?
573 2012-10-10 19:38:46 <amiller> is there a reason why application level integrity would be appropriate but application level error correction is not?
574 2012-10-10 19:39:08 <sipa> it's just say that integrity has higher priority
575 2012-10-10 19:39:21 <sipa> needing to rescan a recent part of the chain isn't that bad
576 2012-10-10 19:39:54 <sipa> in case of corruption, you know nothing anymore and if you assume the worst, you'll need to try extracting keys and rescan the chain
577 2012-10-10 19:40:07 <sipa> i want to avoid that at all costs, as it means losing transaction metaday
578 2012-10-10 19:40:29 <amiller> maybe the wallet is actually several parts - like the priv keys might be stored with application level error correction, and the transaction metadata with application level consistency?
579 2012-10-10 19:40:30 <edcba> but the integrity should be the responsibility of DB
580 2012-10-10 19:40:47 <sipa> amiller: i suppoe, yes
581 2012-10-10 19:40:59 <edcba> or if corruption happens in memory just divides by 0 since you can't assume anything
582 2012-10-10 19:41:06 <sipa> but if you want application-level error correction you'll need a model of how disk corruption works
583 2012-10-10 19:41:25 <sipa> those are usually just a few sectors that become unreadable
584 2012-10-10 19:41:42 <sipa> not random bits that get flipped, for example
585 2012-10-10 19:41:45 <edcba> still it is responsibility of DB or OS or whatever you want
586 2012-10-10 19:41:53 <amiller> i think i agree with midnightmagic that tahoe-lafs' model is the right one for recovery
587 2012-10-10 19:42:03 <sipa> being?
588 2012-10-10 19:42:04 <amiller> you'd want to split the private key in m-of-n style, over multiple sectors or multiple files or w/e
589 2012-10-10 19:42:20 <edcba> or multiple disks...
590 2012-10-10 19:42:25 <edcba> or multiple clouds...
591 2012-10-10 19:42:31 <edcba> or multiple universes
592 2012-10-10 19:42:33 <amiller> scattered across the land
593 2012-10-10 19:42:37 <edcba> oh no we can't do that
594 2012-10-10 19:42:40 <amiller> like in dragon ball z
595 2012-10-10 19:42:51 <sipa> yeah, i think at that points i don't think it's within the scope of what we're trying to do
596 2012-10-10 19:43:09 <amiller> yeah, likely not :p
597 2012-10-10 19:43:10 <sipa> i'd rather have the program make automated encrypted backups
598 2012-10-10 19:43:36 <sipa> which is far easier, and also provides protection against a much more likely threat: humans
599 2012-10-10 19:43:40 <edcba> the only thing we need to assure is we can backup files and reuse it some time later :)
600 2012-10-10 19:43:42 <amiller> well i made my main point which is that recovery rather than integrity is what i think matters for priv keys, i'm not sure about the other parts of the wallet though
601 2012-10-10 19:43:48 <edcba> without losing all your coins :)
602 2012-10-10 19:43:58 <edcba> damn epic fail :)
603 2012-10-10 19:44:14 <sipa> i think integrity is essential, and something you cannot expect from the OS already
604 2012-10-10 19:44:33 <sipa> as you have partial writes that can be interrupted
605 2012-10-10 19:44:56 <sipa> (though some filesystems do provide atomic transactions, i don't think we can rely on those)
606 2012-10-10 19:45:08 <edcba> you have to know what needs to be transactionnaly written
607 2012-10-10 19:45:16 <sipa> that's not a problem
608 2012-10-10 19:45:27 <edcba> but i'd still prefer a deterministic wallet...
609 2012-10-10 19:45:33 <sipa> yeah, me too
610 2012-10-10 19:45:47 <edcba> even if it decreases security
611 2012-10-10 19:46:36 <edcba> maybe not tied to a password yet but still a backupable wallet for eternity
612 2012-10-10 19:47:18 <sipa> hmm, never thought about that
613 2012-10-10 19:47:33 <sipa> you could have a deterministic wallet reset its seed every time you make a backup
614 2012-10-10 19:47:44 <sipa> gmaxwell: ^
615 2012-10-10 19:48:48 <sipa> gavinandresen: linux qt build doesn't match
616 2012-10-10 19:49:29 <gavinandresen> sipa: sigh
617 2012-10-10 19:50:01 <sipa> what's the link to your qt/deps?
618 2012-10-10 19:50:05 <sipa> i'l retry with those
619 2012-10-10 19:50:16 <gavinandresen> skypaint.com/bitcoin
620 2012-10-10 19:50:16 <sipa> win32 build matches
621 2012-10-10 19:50:18 <sipa> and sigs pushed
622 2012-10-10 19:51:29 <sipa> oh, wait... those aren't used for linux
623 2012-10-10 19:51:50 <sipa> gavinandresen: where are your binaries?
624 2012-10-10 19:52:12 <gavinandresen> I uploaded to sourceforge
625 2012-10-10 19:52:15 <sipa> ok, got them
626 2012-10-10 19:52:43 <gavinandresen> I'm starting a kvm build, to see if I get a mismatch kvm versus lxc ....
627 2012-10-10 19:53:10 <sipa> oh, i used kvm entirely now
628 2012-10-10 19:57:19 <sipa> decompiling the file gives entire the same result
629 2012-10-10 19:58:27 <sipa> both 32-bit and 64-bit
630 2012-10-10 20:06:16 <sipa> gavinandresen: got it!
631 2012-10-10 20:06:25 <sipa> it's a header that differs
632 2012-10-10 20:07:04 <sipa> the version of gcc
633 2012-10-10 20:07:42 <sipa> wait no
634 2012-10-10 20:07:51 <sipa> there are two headers in different order
635 2012-10-10 20:07:55 <sipa> swapped
636 2012-10-10 20:08:52 <gavinandresen> ... so all the individual files are the same, but the master checksum is different?
637 2012-10-10 20:09:15 <sipa> no, a header inside the binary
638 2012-10-10 20:09:27 <sipa> about the GCC and library versions used
639 2012-10-10 20:09:42 <sipa> your binary has them in one order, mine in the opposite order
640 2012-10-10 20:46:06 <gmaxwell> kjj_: it's not clear to me that that guy knows pools are made of people.
641 2012-10-10 21:08:10 <slush> gmaxwell: what guy? (I read the channel history and still don't understand)
642 2012-10-10 21:10:13 <gmaxwell> slush: sorry, crossing streams with the forum.
643 2012-10-10 21:10:16 <slush> ok
644 2012-10-10 21:10:57 <gmaxwell> Someone concerned with the consolidation of hash rate into pools, https://bitcointalk.org/index.php?topic=117638.0 the thread has become rather acrimonious.
645 2012-10-10 21:13:24 <slush> gmaxwell: oh this. thanks for link