1 2012-09-29 00:04:19 <gmaxwell> nothing.
  2 2012-09-29 02:25:56 <Impaler> Hey maaku
  3 2012-09-29 02:28:49 <denisx> why is the number of transaction hashes in buildmerkletree not in sync with the poolsz?
  4 2012-09-29 02:32:12 <Impaler> Anyone else interested in discussing demurrage, I've got some new ideas on how to implement it
  5 2012-09-29 03:19:23 <BlueMattBot> Project Bitcoin build #84: FAILURE in 5 hr 13 min: http://jenkins.bluematt.me/job/Bitcoin/84/
  6 2012-09-29 04:10:43 <jgarzik> Stuck download report: https://bitcointalk.org/index.php?topic=113613.0
  7 2012-09-29 04:10:56 <jgarzik> "Around ~196700 or so, it stops downloading"
  8 2012-09-29 04:57:55 <kjj_> jgarzik: let this be a lesson to you.  if you are going to feed the trolls, feed them shit
  9 2012-09-29 05:27:19 <Luke-Jr> kjj_: lol
 10 2012-09-29 05:30:09 <kjj_> seems like such a simple thing, right?
 11 2012-09-29 05:46:50 <jgarzik> kjj_: a "bitcoin for technical people" introduction or boot camp would be nice
 12 2012-09-29 05:47:13 <jgarzik> target: not programmers, but anyone who can grok the concept of crypto hashes and crypto keypairs
 13 2012-09-29 05:47:51 <jgarzik> not necessarily what you were looking for, but that is one idea of mine
 14 2012-09-29 05:48:18 <jgarzik> "bitcoin is a digitally timestamped notary service for small messages
 15 2012-09-29 05:48:44 <jgarzik> and we will shoot you if you use it as simply a timestamped small messaging service"
 16 2012-09-29 05:49:09 <jgarzik> "it's a currency, not IM"
 17 2012-09-29 05:59:16 <Luke-Jr> jgarzik: I almost wonder if the technical side should be given a different name, since it's so vastly different from the high-level view most users have
 18 2012-09-29 05:59:40 <Luke-Jr> a lot of people starting to get into the technical side seem to have preconceptions that just don't work
 19 2012-09-29 05:59:57 <jgarzik> Luke-Jr: "under the hood, we use TonalCoin(tm) technology"
 20 2012-09-29 06:00:11 <Luke-Jr> jgarzik: aww, then we'd need to make it actually tonal-based ;)
 21 2012-09-29 06:00:38 <Luke-Jr> which doesn't really make sense for low-level stuff anyway
 22 2012-09-29 06:01:24 <amiller> the word "mining" is pretty misleading
 23 2012-09-29 06:01:26 <jgarzik> In an imaginary world where compat is no problem, I would redo all the RPC APIs to simply denominate in integer satoshis
 24 2012-09-29 06:01:29 <jgarzik> amiller: agreed
 25 2012-09-29 06:01:37 <jgarzik> 
ACTION prefers "transaction verification"

 26 2012-09-29 06:01:42 <amiller> i've run into a lot of people who really think that the winning block, the thing you obtain with your gpu, is itself the bitcoin
 27 2012-09-29 06:02:05 <jgarzik> the block is the notary's seal
 28 2012-09-29 06:02:48 <Luke-Jr> jgarzik: actually, it might be a not-so-bad idea to plan for some version before 1.0 where we explicitly break RPC compat to fix everything that's inconsistent or stupid?
 29 2012-09-29 06:03:43 <jgarzik> I dunno if there will ever be a point where we will be willing to break _everything_ simply for the sake of cleanliness
 30 2012-09-29 06:03:53 <c_k> the word "notary" is missing from the average new zealanders vocabulary
 31 2012-09-29 06:03:57 <c_k> probably australian too
 32 2012-09-29 06:04:05 <jgarzik> c_k: equivalent?
 33 2012-09-29 06:04:36 <amiller> unfortunately a notary implies a particular set of behaviors too, like checking someone's ID card, and timestamping, but not necessarily validating a document in any other sense
 34 2012-09-29 06:04:53 <c_k> well, we have "justices of the peace" aka JP's that do what notaries seem to do
 35 2012-09-29 06:05:04 <amiller> sometimes i say it's a cross between an election and a lottery, which puts it in the class of (arguably) useful social institutions
 36 2012-09-29 06:05:10 <jgarzik> a link to http://en.wikipedia.org/wiki/Notary will do ;p
 37 2012-09-29 06:05:15 <c_k> A person authorized to perform certain legal formalities, esp. to draw up or certify documents.
 38 2012-09-29 06:05:21 <jgarzik> though yes, calling it a "notary" is an imperfect analogy
 39 2012-09-29 06:05:28 <c_k> ah
 40 2012-09-29 06:05:31 <c_k> well in that case
 41 2012-09-29 06:05:32 <jgarzik> as are all analogies
 42 2012-09-29 06:05:35 <c_k> we call them lawyers
 43 2012-09-29 06:06:27 <maaku> coming up with a better metaphor than mining would go a long ways
 44 2012-09-29 06:06:30 <amiller> yeah, decentralized notary is about the best to expect
 45 2012-09-29 06:11:19 <Impaler> hey their
 46 2012-09-29 06:11:30 <Impaler> "minting"
 47 2012-09-29 06:13:31 <Impaler> dosnt have the same conotation as printing
 48 2012-09-29 06:13:49 <Impaler> and it is a 'coin' so mint makes a lot more sense
 49 2012-09-29 06:15:33 <Graet> the minting bit is a by product of initial distribution of bitcoins, the real thing being done is transaction verification
 50 2012-09-29 06:19:40 <jgarzik> "transaction verification" connects the value of the process to the user in a clear way
 51 2012-09-29 06:20:26 <jgarzik> you understand you need that, and being asked to give up $0.001 for the service is reasonable
 52 2012-09-29 06:20:37 <jgarzik> (hopefully)
 53 2012-09-29 06:20:42 <amiller> i hate the initial block reward thing because it is the dominating force right now, but makes it difficult to understand bitcoin in the steady state
 54 2012-09-29 06:20:52 <jgarzik> yep
 55 2012-09-29 06:20:57 <jgarzik> anyway, bedtime *poof*
 56 2012-09-29 06:21:12 <Impaler> yes we have to wait a long time too see if its really going to work long term
 57 2012-09-29 06:21:16 <amiller> 'hate's not the right word of course since it's useful, but still confusing
 58 2012-09-29 06:22:11 <Impaler> if you look at bitcoin as a protocol development project the design->test->redesign cycle is terribly long
 59 2012-09-29 06:22:12 <amiller> i think the initial block reward is effectively a timer that's running out, we have to solve all the biggest problems before then
 60 2012-09-29 06:23:36 <Impaler> thats another way of looking at it
 61 2012-09-29 06:24:29 <amiller> maybe the bitcoin foundation will be able to get a lot more people to contribute in development
 62 2012-09-29 06:24:59 <amiller> i hope they reach out to a bunch of famous cryptography and distributed protocols people and get them to do substantive review
 63 2012-09-29 06:25:24 <Impaler> yea more developers would help but I worry that the BIG fixes will be resisted by users
 64 2012-09-29 06:26:17 <Impaler> raw security like that will be accepted, but the economic contradictions seem insurmountable
 65 2012-09-29 07:03:13 <lucasleiva> bitcoin for sale cheap pm me
 66 2012-09-29 07:03:48 <BlueMattBot> Project Bitcoin build #85: STILL FAILING in 1 hr 2 min: http://jenkins.bluematt.me/job/Bitcoin/85/
 67 2012-09-29 07:05:22 <midnightmagic> "economic contradictions"? :)
 68 2012-09-29 08:21:47 <sipa> TD: regarding that error, when was that datadir built? i changed the database format recently
 69 2012-09-29 08:22:02 <TD> a couple of days ago, it was post inclusion of the version number
 70 2012-09-29 08:22:08 <TD> i haven't synced my client since
 71 2012-09-29 08:22:14 <TD> er, merged with master i mean
 72 2012-09-29 08:27:52 <BlueMattBot> Project Bitcoin build #86: STILL FAILING in 1 hr 1 min: http://jenkins.bluematt.me/job/Bitcoin/86/
 73 2012-09-29 08:29:59 <sipa> TD: strange, i'll try to re-import everything from scratch
 74 2012-09-29 08:30:21 <TD> this is when starting a client that was running previously. i wonder if there's some issue with hard shutdown or
 75 2012-09-29 08:30:37 <TD> well, imean, i don't remember how i shut it down
 76 2012-09-29 08:30:51 <TD> but it seems to have got itself stuck now
 77 2012-09-29 08:31:53 <sipa> well, there is a potential issue if the block and coin databases do get flushed to disk, but the block itself doesn't
 78 2012-09-29 08:32:02 <sipa> and that could explain what you got
 79 2012-09-29 08:32:33 <sipa> every once in a while all three are sync'ed to disk
 80 2012-09-29 08:32:39 <sipa> in the correct order
 81 2012-09-29 08:32:39 <TD> hmm
 82 2012-09-29 08:32:51 <sipa> but doin that for every block (during IBD) is terribly slow
 83 2012-09-29 08:32:58 <TD> IBD had finished
 84 2012-09-29 08:33:32 <TD> yeah, the error does imply it's trying to read a block that does not exist on disk
 85 2012-09-29 08:33:36 <sipa> indeed
 86 2012-09-29 08:34:11 <sipa> i thought about adding a thread that is responsible for writing stuff to databases and files, and keeping blocks in memory while it's doing that
 87 2012-09-29 08:34:14 <TD> i guess this is fundamental to the design of having block data stored outside the database
 88 2012-09-29 08:34:27 <TD> i wonder how much it'd cost to have the blocks also written to the leveldb
 89 2012-09-29 08:34:55 <sipa> i don't mind writing it to a database, but certainly not to the same one
 90 2012-09-29 08:35:04 <TD> if it's not the same one, it wouldn't resolve the issue
 91 2012-09-29 08:35:05 <sipa> which would have the same problem
 92 2012-09-29 08:35:25 <TD> what's wrong with writing it to the same one? if the keyspace is chosen properly it'd essentially be in a separate set of tablets
 93 2012-09-29 08:35:45 <TD> sorry, not tablets, ssts
 94 2012-09-29 08:37:00 <sipa> well, there are 3 datasets now: blocks, block index, coin set
 95 2012-09-29 08:37:21 <sipa> having an entry in the block index that isn't in the blocks is a problem
 96 2012-09-29 08:37:40 <sipa> but having the coin set be further ahead than the block index is a problem too
 97 2012-09-29 08:37:57 <TD> i wonder how leveldb performance would be affected if blocks were stored in the block index db under keys like block:000000200403:hash -> blockdata
 98 2012-09-29 08:38:09 <TD> they'd essentially be sorted in the database in height order
 99 2012-09-29 08:38:25 <TD> so there should be minimal movement of data during compaction. then block index and block data could be done in the same write batch
100 2012-09-29 08:39:10 <TD> irritating.
101 2012-09-29 08:39:17 <TD> i guess i need to resync the chain from scratch
102 2012-09-29 08:39:29 <TD> how did the old bdb code do this? did it just flush the blkdat file after every write?
103 2012-09-29 08:39:43 <sipa> no
104 2012-09-29 08:39:53 <sipa> you should have experiences the exact same problem with bdb
105 2012-09-29 08:40:29 <sipa> but i've rarely seen those
106 2012-09-29 08:40:41 <TD> hm
107 2012-09-29 08:41:24 <TD> what's the best way to recover? delete the coins and blktree directories?
108 2012-09-29 08:41:28 <TD> it'll rebuild from the files in blocks/ ?
109 2012-09-29 08:41:38 <sipa> no
110 2012-09-29 08:41:53 <sipa> you can delete the coins, and it will rebuild if the blocks+blktree are intact
111 2012-09-29 08:41:59 <sipa> but that's not your problem here
112 2012-09-29 08:42:21 <TD> what's in blktree?
113 2012-09-29 08:42:34 <sipa> the block index
114 2012-09-29 08:42:42 <sipa> basically a blockhash -> diskpos map
115 2012-09-29 08:42:49 <sipa> with some metadata about the block files
116 2012-09-29 08:43:06 <sipa> like how large the files are, how many blocks in them, which heights, which dates
117 2012-09-29 08:43:18 <sipa> and where the undo data is
118 2012-09-29 08:43:27 <TD> ah
119 2012-09-29 08:43:30 <TD> why "tree" ?
120 2012-09-29 08:43:42 <sipa> because it represents the block tree
121 2012-09-29 08:43:56 <sipa> as opposed to data about the active chain in the tree, which is in coin
122 2012-09-29 08:44:03 <TD> ok
123 2012-09-29 08:44:43 <TD> so there's no way to recover from this beyond deleting coins, blocks, blktree and then resyncing from a remote node
124 2012-09-29 08:46:16 <sipa> i'm afraid so
125 2012-09-29 08:46:23 <sipa> well, there's -loadblock
126 2012-09-29 08:48:37 <TD> it might be worth experimenting with using one leveldb for everything, and see how it affects performance
127 2012-09-29 08:50:31 <TD> having situations which are known to cause nodes to get stuck and need reinitializing isn't great, even if it only happens once in a blue moon
128 2012-09-29 08:50:40 <TD> (not that bitcoinj is any better, sigh)
129 2012-09-29 08:50:43 <sipa> i certainly agree
130 2012-09-29 08:51:22 <sipa> but storing blocks inside leveldb would also imply that all block data gets written twice (once to the log, once to the database, and perhaps a few times when rewriting the tablets)
131 2012-09-29 08:52:38 <sipa> bleh, just reproduced
132 2012-09-29 08:52:49 <sipa> i yanked the power cable and battery from my laptop during IBD
133 2012-09-29 08:52:56 <TD> :(
134 2012-09-29 08:53:09 <TD> yes, it would definitely increase throughput a lot
135 2012-09-29 08:53:24 <TD> OTOH read bandwidth is fairly cheap, disk seeks are the costly part
136 2012-09-29 08:53:34 <TD> that's pretty much the rationale behind the leveldb design
137 2012-09-29 08:53:41 <TD> and it's all done on a separate thread
138 2012-09-29 08:53:47 <sipa> yes, i know
139 2012-09-29 08:53:50 <TD> so, in the end it may or may not make much difference, i don't have a good feel for it
140 2012-09-29 08:54:13 <sipa> for data that's frequently read but infrequently written, it makes sense
141 2012-09-29 08:54:23 <sipa> but blocks are read almost never
142 2012-09-29 08:54:41 <sipa> then again, i'm just guessing
143 2012-09-29 08:55:40 <sipa> wait a second
144 2012-09-29 08:55:46 <TD> yeah, i don't know of any way to avoid this situation though other than flushing the blocks after every write. and even then maybe all you do is shrink the size of the race
145 2012-09-29 08:55:53 <sipa> this is not ultraprune or leveldb!
146 2012-09-29 08:56:22 <sipa> TD: sure there is a solution; do all leveldb writes and block writes synchronously
147 2012-09-29 08:56:28 <sipa> but do it for many blocks at once
148 2012-09-29 08:56:37 <sipa> in a separate thread
149 2012-09-29 08:56:48 <TD> hmm
150 2012-09-29 08:57:08 <TD> you mean just build up a giant writebatch that covers hundreds of blocks, then set the leveldb option that makes it flush everything with every write
151 2012-09-29 08:57:11 <TD> yeah i guess so
152 2012-09-29 08:57:40 <sipa> currently i do an fsync + sync leveldb write every time >5000 transactions changed
153 2012-09-29 08:57:53 <sipa> which is a few hundred blocks initially, and a few blocks now
154 2012-09-29 08:58:17 <sipa> oh no, the coin set write is non-sync, but that's not a problem
155 2012-09-29 08:58:37 <sipa> as it can lag behind on the block data
156 2012-09-29 09:00:36 <sipa> anyway, this "reproduced" error i got from yanking the power from my laptop was just git head bdb
157 2012-09-29 09:01:37 <TD> oh
158 2012-09-29 09:01:44 <TD> i guess we can live with it for now then
159 2012-09-29 09:01:57 <sipa> better recovery is certainly nice
160 2012-09-29 09:02:25 <sipa> like automatically rebuilding blktree if it doesn't match the stored blocks
161 2012-09-29 09:05:14 <sipa> so what i envision: a global block+undo data cache, which tracks which block entries/block data/undo data is "dirty", and loops: { wait_for_something_dirty(); write_dirty_block_data(); write_dirty_index_entries(); }
162 2012-09-29 09:05:51 <sipa> and then the block connection logic can issue a sync() to that cache before writing a coin set updates
163 2012-09-29 09:06:12 <sipa> ").fixGrammar();
164 2012-09-29 09:06:48 <TD> wouldn't that be equivalent to doing a sync after every new block ?
165 2012-09-29 09:07:29 <sipa> coin set updates aren't written after every block
166 2012-09-29 09:07:50 <sipa> (right now, after every 5000 modified transactions)
167 2012-09-29 09:08:32 <TD> ok
168 2012-09-29 09:08:36 <TD> sorry yeah you said
169 2012-09-29 09:09:18 <sipa> which might even speed things up, because right now block connection is paused while writing block/undo data
170 2012-09-29 09:11:09 <TD> so -loadblock=blk0001.dat,blk0002.dat works?
171 2012-09-29 09:13:46 <TD> (using old blkdat files, of course)
172 2012-09-29 09:14:43 <sipa> you need to specify -loadblock two times
173 2012-09-29 09:15:03 <TD> ok
174 2012-09-29 09:15:06 <TD> thanks
175 2012-09-29 09:18:55 <TD> btw welcome to zurich. hope you weren't expecting sunshine today ....
176 2012-09-29 09:19:58 <sipa> the weather was nice yesterday :)
177 2012-09-29 09:22:30 <sipa> TD: just did the same with ultraprune+leveldb; no problems
178 2012-09-29 09:22:52 <sipa> i know, it's just anecdotecal evidence right now
179 2012-09-29 09:23:21 <TD> doh
180 2012-09-29 09:23:24 <TD> -loadblock failed
181 2012-09-29 09:23:25 <TD> hearn-macbookpro:gitrepo hearn$ ./Bitcoin-Qt.app/Contents/MacOS/Bitcoin-Qt -loadblock=/Users/hearn/Library/Application\\ Support/Bitcoin/blk0001.dat -loadblock=/Users/hearn/Library/Application\\ Support/Bitcoin/blk0002.dat
182 2012-09-29 09:23:40 <TD> last 5 lines of log
183 2012-09-29 09:23:41 <TD> ProcessBlock: ACCEPTED
184 2012-09-29 09:23:52 <sipa> and that's entirely in a clean setup?
185 2012-09-29 09:23:55 <sipa> from scratch?
186 2012-09-29 09:24:02 <TD> that was after deleting coins, blocks, blktree
187 2012-09-29 09:24:20 <TD> and using blk0001.dat/0002.dat that were in that directory. perhaps left over from a pre-leveldb run
188 2012-09-29 09:24:29 <TD> pre-ultraprune i mean
189 2012-09-29 09:24:32 <sipa> doesn't matter
190 2012-09-29 09:24:46 <sipa> you should be able to pipe /dev/urandom into -loadblock without problems
191 2012-09-29 09:26:18 <sipa> yo're on OSX?
192 2012-09-29 09:26:35 <sipa> maybe a platform-specific problem somewhere
193 2012-09-29 09:40:38 <TD> alright
194 2012-09-29 09:40:41 <TD> re-sync from a remote node it is
195 2012-09-29 10:59:48 <denisx> can someone explain to me why the count of transactions in buildmerkletree is not in par with the poolsz value I see in debug.log?
196 2012-09-29 11:00:17 <denisx> on par
197 2012-09-29 11:06:27 <TD> poolsz is all pending transactions
198 2012-09-29 11:06:38 <TD> buildmerkletree includes only pending transactions that meet your policy, i'd think
199 2012-09-29 11:35:22 <sipa> what TD said
200 2012-09-29 12:43:46 <lucasleiva> anyone buying
201 2012-09-29 12:57:09 <Hasimir> lucasleiva, wrong channel, this is dev and you probably want otc
202 2012-09-29 13:09:33 <Crazy_man> Check out my new Bitcoin project idea: http://pastebin.com/1vNP6CRP Does anyone wanna participate?
203 2012-09-29 13:13:23 <Luke-Jr> Crazy_man: sounds self-contradicting
204 2012-09-29 13:15:30 <Crazy_man> how so
205 2012-09-29 13:17:11 <Crazy_man> Luke-Jr: you dont like the id of anonymous donations through bitcoins
206 2012-09-29 13:17:33 <Luke-Jr> Crazy_man: anonymous generally means laundering, which is a crime
207 2012-09-29 13:18:00 <Luke-Jr> so it doesn't make sense to have a rule against illegal activity
208 2012-09-29 13:18:12 <Crazy_man> for it to be laundering, money has to be dirty
209 2012-09-29 13:18:20 <Crazy_man> in the first place
210 2012-09-29 13:18:33 <Luke-Jr> nope
211 2012-09-29 13:18:34 <Crazy_man> like money from drug sales
212 2012-09-29 13:18:48 <Luke-Jr> laundering is concealing the source of it, period, AFAIK
213 2012-09-29 13:19:11 <sipa> still, anonimity is not a crime
214 2012-09-29 13:19:17 <sipa> privacy is a very basic right
215 2012-09-29 13:19:52 <Luke-Jr> sipa: the line between privacy and laundry is blurry :<
216 2012-09-29 13:19:53 <Crazy_man> thats only for >$10,000
217 2012-09-29 13:19:57 <Crazy_man> banks have to report
218 2012-09-29 13:20:08 <Crazy_man> transactions
219 2012-09-29 13:20:29 <sipa> if you pay with your own legally-earned cash in a shop, the source is concealed
220 2012-09-29 13:20:38 <sipa> it's perfectly legal though
221 2012-09-29 13:20:53 <Crazy_man> luke is right about banks having to report >$10,000
222 2012-09-29 13:21:18 <Crazy_man> transactions
223 2012-09-29 13:21:22 <Crazy_man> but this is not a bank
224 2012-09-29 13:21:23 <Luke-Jr> sipa: the source is you :p
225 2012-09-29 13:21:32 <Crazy_man> more so an escrow
226 2012-09-29 13:21:43 <sipa> Luke-Jr: nobody knows where it came from
227 2012-09-29 13:21:47 <sipa> probably not even you
228 2012-09-29 13:22:04 <Luke-Jr> sipa: I'd think all that matters is who is paying
229 2012-09-29 13:22:13 <Crazy_man> Luke-Jr: $10,000 reporting law only applies to cash transactions not bitcoins
230 2012-09-29 13:22:13 <sipa> certainly not
231 2012-09-29 13:22:18 <Luke-Jr> eg, if A gives money to B to pay C, it's really A paying C, but C doesn't know that
232 2012-09-29 13:22:32 <Luke-Jr> but so long as A isn't involved, C doesn't need to know..
233 2012-09-29 13:22:57 <Luke-Jr> Crazy_man: I think you need to consult a lawyer before you act on that opinion
234 2012-09-29 13:23:11 <Luke-Jr> I don't think any of these laws are "only for cash"
235 2012-09-29 13:23:14 <Diablo-D3> unless the government declares bitcoin a legal and regulated monetary instrument
236 2012-09-29 13:23:19 <Diablo-D3> then it does.
237 2012-09-29 13:23:52 <Luke-Jr> Crazy_man: btw, IANAL, or anyone else here AFAIK ???
238 2012-09-29 13:23:53 <Crazy_man> i guess i will have to get peoples' information for $10k+ transactions
239 2012-09-29 13:23:57 <Crazy_man> if they ever happen
240 2012-09-29 13:24:00 <sipa> ianal, but i see no reason why a transaction worth $10000 or more, when the medium is goldfish, wouldn't need to be reported
241 2012-09-29 13:24:13 <Luke-Jr> 
ACTION ponders $10k worth of goldfish

242 2012-09-29 13:24:21 <Matt_von_Mises> Hello. Does anyone understand RAND_add's entropy argument?
243 2012-09-29 13:24:56 <Crazy_man> but i dont think i have to report the person's name
244 2012-09-29 13:24:57 <sipa> yes, it points to the random data
245 2012-09-29 13:25:04 <Crazy_man> i ust have to report
246 2012-09-29 13:25:08 <Crazy_man> the transaction
247 2012-09-29 13:25:13 <sipa> well, not necessarily entirely random, just data that contains some entropy
248 2012-09-29 13:25:34 <Luke-Jr> Crazy_man: considering the Patriot Act?
249 2012-09-29 13:25:43 <sipa> Matt_von_Mises: oh, you mean that double passed in?
250 2012-09-29 13:25:53 <sipa> Matt_von_Mises: that's the amount of entropy in the data passed in
251 2012-09-29 13:25:56 <Matt_von_Mises> sipa: Yes the double
252 2012-09-29 13:26:18 <sipa> Matt_von_Mises: what are you adding?
253 2012-09-29 13:26:53 <Crazy_man> Luke-jr: patriot act doesnt have a money laundering provision as far as i know
254 2012-09-29 13:27:10 <Matt_von_Mises> So it's the amount of bytes in the data which are supposedly random? So if I get someone to type something random in the terminal (eg. sfugweo9h32f;ewG'EWGEWG\\wptw9gfs) the value should be the length of the string if I try to add that?
255 2012-09-29 13:27:20 <sipa> Matt_von_Mises: stop
256 2012-09-29 13:27:24 <Crazy_man> Luke-Jr: would a bitcoin donation website be hard to develop
257 2012-09-29 13:27:32 <TD> patriot act did change AML provisions in US law, notably around hawala networks. however, AML laws are both vague and vary by jurisdiction
258 2012-09-29 13:27:33 <sipa> Matt_von_Mises: data != entropy
259 2012-09-29 13:27:38 <TD> so it's not worth trying to discuss it in the general case.
260 2012-09-29 13:27:44 <TD> for the US you'd need to refer to state-level regulations, even
261 2012-09-29 13:28:06 <Luke-Jr> Crazy_man: Patriot Act is why banks are required to get photo ID
262 2012-09-29 13:28:19 <sipa> Matt_von_Mises: entropy is the degree of randomness; if every byte was fully independent from all the others, and each had a 1/256 chance for being any byte, then yes
263 2012-09-29 13:28:30 <sipa> Matt_von_Mises: but keyboard input is FAR from as random as that
264 2012-09-29 13:28:53 <Matt_von_Mises> sipa: That's what I thought, the text wont even include every value since it will only include a range of characters from a keyboard input.
265 2012-09-29 13:29:10 <Matt_von_Mises> sipa: Yes, so what I want to know is how to calculate the value.
266 2012-09-29 13:29:33 <sipa> Matt_von_Mises: you're better off not using the typed data at all, but recording the microsecond timestamps of the keypresses, and using the sub-millisecond values of those as entropy source
267 2012-09-29 13:29:59 <sipa> Matt_von_Mises: people might be pressing the 'a' key the whole time, in which case the text contains almost no entropy
268 2012-09-29 13:30:19 <sipa> in general, don't ask people for entropy, because humans typically misunderstand what it means
269 2012-09-29 13:30:30 <sipa> just let them perform a task to generate it
270 2012-09-29 13:30:51 <TD> Luke-Jr: that's the "banking secrecy act", actually
271 2012-09-29 13:30:57 <TD> patriot act just ramped up the aggressiveness of the system not created it
272 2012-09-29 13:31:10 <TD> (BSA created the concept of money laundering)
273 2012-09-29 13:31:29 <Matt_von_Mises> sipa: OK, I suppose I could just use dev/random and ask for the user to move the cursor around because apparently moving the cursor provides entropy to dev/random (I read that somewhere)?
274 2012-09-29 13:32:37 <sipa> Matt_von_Mises: if your system has a /dev/random, definitely use it
275 2012-09-29 13:32:47 <sipa> Matt_von_Mises: but don't exhaust it
276 2012-09-29 13:35:24 <Matt_von_Mises> sipa: Ok thanks. I'm reading it once on start-up.
277 2012-09-29 14:11:21 <sipa> BlueMatt, Luke-Jr: pushed an update to bitcoin-seeder; finding onion peers should work now
278 2012-09-29 14:24:54 <yellowhat> Crazy_Man i have briefly read your text. if you want to do a "kickstarter for bitcoin" i think its a good idea. if you use multisig correctly you can remove a lot of potential problems and trust issues. also i think its hard work (4-20 weeks of serious full-time coding) do do it really well and you need a marketing budget to get things rolling. good luck :)
279 2012-09-29 14:27:46 <yellowhat> oh he is gone already
280 2012-09-29 14:35:42 <gavinandresen> Before somebody asks: yep, I know that bitcoinfoundation.org is down.  According to RackSpace, looks like a hardware problem not a DDoS
281 2012-09-29 14:48:06 <kreal> I hate what stuff like that happens.
282 2012-09-29 14:48:15 <kreal> more so in the start up period.
283 2012-09-29 14:48:27 <kreal> its awful.
284 2012-09-29 14:53:53 <kreal> on a side note I tried signin up providing WalletBit as companyname, then noticed the bitcoin prices and decided. apparently not for me.
285 2012-09-29 15:08:30 <jgarzik> gavinandresen: BF needs a twitter for status updates like that
286 2012-09-29 15:09:44 <gavinandresen> jgarzik: good idea...
287 2012-09-29 15:09:55 <jgarzik> yellowhat: BTW RE KickStarter, you don't even need CHECK_MULTISIG.  Just create an anyone-can-pay transaction that sends $TargetAmount to $PublicKey.  Send it around, people add signatures as they like.  It remains an invalid transaction until it hits the target amount.
288 2012-09-29 15:11:14 <jgarzik> speaking of...  would be interesting to add a crowdfunding RPC
289 2012-09-29 15:11:24 <jgarzik> add-sig-to-remote-tx
290 2012-09-29 15:11:41 <jgarzik> download tx from $URL, sign it, and re-upload
291 2012-09-29 15:15:59 <gavinandresen> That reminds me... can I get some ACKs on https://github.com/bitcoin/bitcoin/pull/1818
292 2012-09-29 15:22:17 <Luke-Jr> sipa: without tor?
293 2012-09-29 15:23:39 <sipa> Luke-Jr: obviously not
294 2012-09-29 15:23:57 <Luke-Jr> sipa: not obviously :<
295 2012-09-29 15:24:12 <sipa> well you can't connect to tor without tor :)
296 2012-09-29 15:24:20 <sipa> we don't replicate tor functionality inside bitcoin
297 2012-09-29 15:33:08 <sipa> wow, 10 onion peers and 84 ipv6 peers already
298 2012-09-29 15:33:39 <gavinandresen> nice!
299 2012-09-29 17:23:44 <jgarzik> Man.  This hazek-edits-Gavins-post thing is REALLY bugging me.
300 2012-09-29 17:24:07 <kjj_> meh.  seemed to be well intentioned, but misguided
301 2012-09-29 17:24:12 <jgarzik> IMNSHO that is a worse abuse of moderator powers than simply deleting a post from a flamewar opponent.
302 2012-09-29 17:24:28 <jgarzik> And it calls into question -every- past post by gavin/satoshi/me/anyone else.
303 2012-09-29 17:25:09 <kjj_> none of the messages on the forum are signed.  they are just entries in a database
304 2012-09-29 17:25:24 <kjj_> they are in question the second the are posted.
305 2012-09-29 17:25:37 <jgarzik> true enough
306 2012-09-29 17:25:39 <enmaku> @jgarzik waitwait, which post? I'm out of the loop here...
307 2012-09-29 17:25:52 <kjj_> the first post in the foundation thread
308 2012-09-29 17:26:08 <enmaku> oh dear... off I go to research
309 2012-09-29 17:26:11 <jgarzik> but it shouldn't be that easy, without admin rights
310 2012-09-29 17:26:14 <kjj_> the links didn't work for hazek (the moderator of that board) so he changed the links to .com
311 2012-09-29 17:26:42 <enmaku> 
ACTION facepalm

312 2012-09-29 17:26:42 <jgarzik> some mod might decide it is fun to change all of satoshi's old posts, just to win a flame war
313 2012-09-29 17:26:54 <jgarzik> only the physical admin should be able to do that
314 2012-09-29 17:26:56 <kjj_> heh.  I just figured out how to reduce the noise to signal ratio
315 2012-09-29 17:27:26 <kjj_> we can make new forums that require all posts to be GPG signed
316 2012-09-29 17:27:33 <enmaku> well just because the forum doesn't have digital signing built in doesn't mean you can't just sign things yourself, right?
317 2012-09-29 17:27:46 <enmaku> wait, what am I saying, the conversion from bbcode to html would fubar ths signature
318 2012-09-29 17:27:49 <jgarzik> enmaku: yep... and we do for major announcements, download links, etc.
319 2012-09-29 17:27:58 <jgarzik> yeah, no markup
320 2012-09-29 17:28:29 <jgarzik> and it is a fair point, in retrospect, and gavin should have signed this with a pgp key
321 2012-09-29 17:28:38 <jgarzik> s/and/that/
322 2012-09-29 17:28:45 <enmaku> bitcointalk appears to be down for me right now, so I guess no research :(
323 2012-09-29 17:29:05 <jgarzik> enmaku: very, very, very slow but working here
324 2012-09-29 17:29:12 <enmaku> and as soon as I say it, it comes back up. I think I just need to go back to bed today
325 2012-09-29 17:29:16 <jgarzik> enmaku: 2 minutes per link slow
326 2012-09-29 17:30:17 <enmaku> looks like they've all been changed back to .org
327 2012-09-29 17:30:31 <enmaku> still it's a really solid point, this totally should've been signed
328 2012-09-29 17:34:29 <kjj_> I presume that edits are logged.  we'll have to ask Theymos about that (unless someone else knows for sure)
329 2012-09-29 17:35:50 <enmaku> @kjj_ yeah, SMF logs by default, but they log to the same db the posts are held in, there are plenty of ways for someone with db access to do an un-logged edit
330 2012-09-29 17:36:41 <enmaku> it can also be disabled outright
331 2012-09-29 17:36:57 <enmaku> which, stupid as it sounds, might have been done given the kind of traffic and responsiveness problems they've had
332 2012-09-29 17:37:02 <kjj_> yeah, trust is always a factor.  at some point, you are relying on someone
333 2012-09-29 17:37:05 <jgarzik> AFAIK theymos does maintain an admin <-> moderator distinction, where moderators [should not] have direct db access
334 2012-09-29 17:37:27 <enmaku> @jgarzik well that's reassuring at least
335 2012-09-29 17:37:47 <jgarzik> I would rather have a distributed forum with each post signed by ECDSA ;p
336 2012-09-29 17:38:02 <enmaku> @jgarzik oh god, ForumCoin
337 2012-09-29 17:38:07 <enmaku> what did you just do
338 2012-09-29 17:38:14 <lianj> have a forum with a git backend
339 2012-09-29 17:38:47 <kjj_> heh, the foundation could set up a CA, and the forum could reject posts that aren't signed by an approved certificate...
340 2012-09-29 17:38:52 <dangermouse_> ForumCoin
341 2012-09-29 17:43:47 <Eliel> jgarzik: Have you looked at this p2p forum software called Osiris?
342 2012-09-29 17:44:06 <jgarzik> lianj: *ding* 10 points to you ;p
343 2012-09-29 17:44:32 <jgarzik> git backend, signed posts
344 2012-09-29 17:44:52 <jgarzik> extra points for "distributed among untrusted parties on the internet"
345 2012-09-29 17:45:25 <jgarzik> Eliel: this? http://en.wikipedia.org/wiki/Osiris_%28Serverless_Portal_System%29
346 2012-09-29 17:45:34 <enmaku> how transparent is all the backend and signing stuff to the end users though
347 2012-09-29 17:46:06 <enmaku> bc using git for a forum sounds like a great idea to a dev, but if it's not simplified for users it drives away 90% of the community
348 2012-09-29 17:46:34 <sipa> "hey i don't like this discussion; sec, let me fork ik..."
349 2012-09-29 17:46:48 <Eliel> jgarzik: yes, that.
350 2012-09-29 17:47:21 <jgarzik> hehehe
351 2012-09-29 17:47:36 <jgarzik> not quite _that_ distributed.  admins and moderators would have familiar access perms
352 2012-09-29 17:48:10 <jgarzik> though yes, if the community thought the admins went off the rails, they could "elect" a new admin or moderator key
353 2012-09-29 17:48:13 <Eliel> jgarzik: osiris does support reasonably heavy central control too :P
354 2012-09-29 17:48:37 <Eliel> for an individual board
355 2012-09-29 17:50:05 <enmaku> bah the isis PHP gateway is read-only, you actually have to install a client app to post
356 2012-09-29 17:51:40 <enmaku> and by the time you talk the average user through downloading a client and setting up a keypair... we might as well just write ForumCoin
357 2012-09-29 17:51:50 <Eliel> enmaku: it's very transparent. The program is a background service you interact with through your browser.
358 2012-09-29 17:52:10 <Eliel> feels like a pretty normal forum to the user
359 2012-09-29 17:52:21 <lianj> good thing about the git backend is you couldnt change stuff in the history unnoticed without writing the whole commit history
360 2012-09-29 17:52:31 <enmaku> Eliel: once it's set up, yes. it's getting people through the setup process that's gonna suck
361 2012-09-29 17:52:32 <lianj> *rewriting
362 2012-09-29 17:53:01 <Eliel> enmaku: I don't remember how it was but I think it was start installer, next, next, next, next, done.
363 2012-09-29 17:53:10 <Eliel> (for windows version)
364 2012-09-29 17:53:29 <enmaku> Eliel: so keypair management is done transparently?
365 2012-09-29 17:53:39 <enmaku> no user input necessary?
366 2012-09-29 17:54:17 <Eliel> enmaku: I can't remember having to generate them manually, so I think. It's been quite a while since I installed it so I might just have forgotten though.
367 2012-09-29 17:54:34 <kjj_> seems unlikely.  key management is hard to do right, and not because the UI is bad
368 2012-09-29 17:55:14 <Eliel> I think it's abstracted behind account creation. You can create different identities in it.
369 2012-09-29 17:55:57 <kjj_> heh.  if key management is easy, it probably isn't secure.  and if it is secure, it probably isn't easy.
370 2012-09-29 17:56:05 <kjj_> that's been my experience with PKI since forever
371 2012-09-29 17:56:28 <Eliel> kjj_: what aspects of security do you mean?
372 2012-09-29 17:56:53 <enmaku> I think it's probably best we just stick with GPG/ECDSA signing on standard platforms
373 2012-09-29 17:57:11 <enmaku> just let people who want to check signatures do so without causing any static for those that don't
374 2012-09-29 17:58:51 <Eliel> it looks like it's just abstracted as an user-account creation (that you can export and import if necessary)
375 2012-09-29 18:00:19 <Eliel> no direct involvement with the encryption keys that I can see here.
376 2012-09-29 18:01:30 <Eliel> just the usual, set username and password and you're done. you can login or logout.
377 2012-09-29 18:02:04 <Eliel> that it doesn't ask for email address is probably the most glaring oddity there.
378 2012-09-29 18:19:46 <midnightmagic> A javascript post-markup mechanism could be used rather than server-side, and the signature would work fine. Or just a subsection of the message could contain a cleartext attached sig.
379 2012-09-29 18:20:10 <midnightmagic> Also, versioned edits would basically solve the problem too.
380 2012-09-29 18:20:20 <kjj_> if you hit quote on a post, you get the "source".  presumably that is what would be signed and verified
381 2012-09-29 18:20:21 <midnightmagic> Unfortunately SMF doesn't have versioned edits.
382 2012-09-29 18:20:48 <kjj_> Eliel: keeping keys secure, distributing them, deciding which ones to trust and not trust.
383 2012-09-29 18:21:29 <enmaku> midnightmagic: it does keep a timestamped edit log though, yes?
384 2012-09-29 18:21:43 <enmaku> midnightmagic: not necessarily the content of each edit, but a log of who edited when
385 2012-09-29 18:21:47 <midnightmagic> The real solution is a browser-based plugin for managing identity, light signatures, proper secure signatures, and simple capabilities delegation.
386 2012-09-29 18:22:13 <midnightmagic> enmaku: I think the full content should be completely versioned and totally visible without admin intervention.
387 2012-09-29 18:22:24 <kjj_> well, SSL's PKI model blows chunks.  It is seriously the worst thing that ever happened to the internet.  but we all use it because the other models are worse
388 2012-09-29 18:22:36 <jgarzik> kjj_: are there any better models?
389 2012-09-29 18:22:47 <midnightmagic> SSL does suck. something simpler and gnupg-based would probably be better.
390 2012-09-29 18:22:53 <kjj_> jgarzik: not that I've ever heard of
391 2012-09-29 18:23:13 <kjj_> Moxie might be onto something, but his ideas don't seem to pan out when you follow them to the end
392 2012-09-29 18:23:20 <midnightmagic> WoT seems to me to be the only real solution to the issue.
393 2012-09-29 18:23:44 <enmaku> midnightmagic: well, yes, full versioned edits would be ideal, but in their absence it's better to have something than nothing :)
394 2012-09-29 18:23:51 <midnightmagic> convergence you mean?  that will only work if people begin making proper use of WoT, and not the sort of wot that that stupid voting mechanism uses..
395 2012-09-29 18:23:59 <midnightmagic> enmaku: I agree.
396 2012-09-29 18:24:25 <midnightmagic> I think the WoT mechanism I'm talking about is mywot.com
397 2012-09-29 18:24:28 <kjj_> the PGP/GPG model seems to be the least-centralized, but it doesn't really work well in practice
398 2012-09-29 18:24:36 <enmaku> midnightmagic: actually one of the many reasons I prefer StackExchange chat to IRC, though none of the Bitcoin community seems to ever use it :(
399 2012-09-29 18:24:59 <midnightmagic> kjj_: It can if you are willing to rate someone numerically *and* willing to accept a few levels of trust away from you.
400 2012-09-29 18:25:34 <midnightmagic> kjj_: Local WoT introducers for example, would work for the case of families and friends. Then the local tech guy can just point them to the convergence sites that they think are trustworthy.
401 2012-09-29 18:25:36 <kjj_> midnightmagic: which is why no one uses GPG.  (for approximate values of "no one")
402 2012-09-29 18:26:20 <sipa> enmaku: is that group chat?
403 2012-09-29 18:26:45 <midnightmagic> kjj_: You mean except for every secure distribution mechanism for every free OS there is, most cryptographers, every hacker that accepts pseudo-nonymous incoming messages, almost everyone in #bitcoin-otr, and and and? :)
404 2012-09-29 18:27:14 <Eliel> kjj_: there's no way to get most people to bother about those aspects of key security without a real practical need to do so.
405 2012-09-29 18:27:14 <enmaku> sipa: take IRC, add in full logging and versioning on every channel, markdown and oneboxing and you've got SE chat
406 2012-09-29 18:27:15 <midnightmagic> enmaku: What's this stackexchange chat?  Does it have a WoT mechanism of some sort?
407 2012-09-29 18:27:30 <enmaku> oh and a lot more administrative capabilities.
408 2012-09-29 18:27:35 <midnightmagic> Eliel: If you wrap it up into a "never have to sign in to a supported website again" people will.
409 2012-09-29 18:28:24 <midnightmagic> Eliel: openid would've been a nice thing but no sites out there allow custom openid providers.
410 2012-09-29 18:28:36 <Eliel> I've seen a couple :P
411 2012-09-29 18:28:37 <sipa> enmaku: really? i tried to use it once, and all it did was take me to a login page, while i was already logged in
412 2012-09-29 18:28:48 <kjj_> midnightmagic: yeah, that is "no one" in the context of the internet
413 2012-09-29 18:28:56 <enmaku> sipa: that's weird... there's a sandbox room if you want to play with it - http://chat.stackexchange.com/rooms/1/sandbox
414 2012-09-29 18:29:28 <enmaku> sipa: there's also a general Bitcoin chat but as I mentioned it, no one really uses it much - http://chat.stackexchange.com/rooms/1233/bitcoin
415 2012-09-29 18:29:42 <midnightmagic> kjj_: And yet when openid promised to be that thing, suddenly all kinds of major websites started supporting it.
416 2012-09-29 18:30:11 <Eliel> kjj_: anyway, for a forum software, key distribution and trust decisions are not so important. keeping the keys secure, though, is.
417 2012-09-29 18:30:13 <midnightmagic> kjj_: And the idea spread into custom "login with your facebook id" b-s.
418 2012-09-29 18:30:43 <Eliel> kjj_: the keys are used to encrypt private messages, you can use that to verify identity if need be.
419 2012-09-29 18:30:58 <midnightmagic> kjj_: Bastardized and encompassed and stolen and mutated into uselessness by commercial interests who know it's just a countdown until something like what I describe arrives.
420 2012-09-29 18:31:43 <Eliel> kjj_: I expect osiris keeps keys encrypted with a password. It even shows a nice bar that tells you how secure your password is when you type it.
421 2012-09-29 18:32:25 <Eliel> (at least, I can't figure out any other reason for the password that's associated with the user accounts you can create)
422 2012-09-29 18:33:15 <midnightmagic> Eliel: lol those little weird "your password is this secure" bars are so silly.
423 2012-09-29 18:34:09 <enmaku> midnightmagic, Eliel: insert "correct horse battery staple" XKCD comic here
424 2012-09-29 18:34:30 <midnightmagic> enmaku: lol and that was a total misrepresentation of how secure that password actually is!
425 2012-09-29 18:35:04 <Eliel> well, yes, now with the comic so famous, it's one of the most insecure passwords out there :D
426 2012-09-29 18:35:33 <enmaku> Eliel: someone already used sha256("correct horse battery staple") as a brain wallet
427 2012-09-29 18:35:49 <enmaku> not sure if it was their proof of concept or if someone else emptied it, but it's sure empty now
428 2012-09-29 18:36:08 <midnightmagic> Eliel: Even before that.. you don't get to use every letter of a real word as part of your entropy estimation.
429 2012-09-29 18:36:14 <sipa> enmaku: there you go, posted something :)
430 2012-09-29 18:36:45 <Eliel> midnightmagic: that's more of a criticism on the algorithm that calculates the strength.
431 2012-09-29 18:38:50 <Eliel> but yeah, those can't shield against every mistake the user could make.
432 2012-09-29 18:43:24 <enmaku> sipa: just left you a rant on SE chat about how awesome it is and how biased I am ;)
433 2012-09-29 18:43:55 <midnightmagic> Eliel: Yeah that's true. I just meant XKCD's author's description of the entropy in the correct horse battery staple password is incorrect.
434 2012-09-29 18:44:18 <Eliel> oh, yeah, true that
435 2012-09-29 20:01:33 <Eliel> yay! someone is doing something that might work :) http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone
436 2012-09-29 20:05:06 <EvanR2> so bitcoind is not only taking a shit ton of memory, but it also downloads the block chain at super turtle speed
437 2012-09-29 20:05:09 <EvanR2> 0.6.3
438 2012-09-29 20:05:16 <EvanR2> im not behind a nat, what gives
439 2012-09-29 20:13:03 <egecko> yep, just let it run
440 2012-09-29 20:13:56 <Eliel> EvanR2: perfectly normal.
441 2012-09-29 20:14:45 <wizkid057> so, why dont we just use a distributed hash table?
442 2012-09-29 20:14:46 <wizkid057> :D
443 2012-09-29 20:17:18 <sipa> wizkid057: try as you may
444 2012-09-29 20:17:58 <sipa> amiller has done some great work around the idea of how a fully validating bitcoin client could work without trusted storage
445 2012-09-29 20:19:23 <wizkid057> sipa: was just trolling, I know a DHT wouldnt work here very well.  I just recall that many many less technically savy folks have suggested it and it kinda became a running joke. lol
446 2012-09-29 20:19:40 <sipa> haha
447 2012-09-29 20:19:51 <sipa> gmaxwell had this quote
448 2012-09-29 20:21:26 <sipa> #bitcoin-dev-20120416.log:18:15:17< gmaxwell> Someday I'm going to get myself invited to some conference with the president, and while he's talking about some middle east conflict thing??? I'm going to ask if they've considered using a DHT.
449 2012-09-29 20:23:22 <wizkid057> rofl
450 2012-09-29 20:23:33 <wizkid057> thats awesome
451 2012-09-29 20:23:39 <wizkid057> how did I miss that...
452 2012-09-29 20:23:43 <wizkid057> 
ACTION pulls up logs

453 2012-09-29 20:23:59 <sipa> the timestamp is GMT+1 or GMT+2
454 2012-09-29 20:24:22 <wizkid057> found it... haha
455 2012-09-29 21:21:45 <arij> any devs for hire? pm me!
456 2012-09-29 21:31:06 <sipa> TD[gone]: mind if i edit your commit to be in leveldb/ instead of leveldb-1.5.0/ ?
457 2012-09-29 23:19:39 <kreal> arij, what do you need?