1 2012-09-20 00:12:14 <diki> EC point multiplication really sucks up cycles.
2 2012-09-20 00:15:35 <Diablo-D3> isnt that the point?
3 2012-09-20 00:24:18 <gjs278> debug.log should be an damn option
4 2012-09-20 00:24:27 <gjs278> I have mine symlinked to /dev/null too
5 2012-09-20 00:26:59 <gmaxwell> symlinking it to /dev/null is a not unreasonable thing... though, you 'void your warranty' doing that.??? if you have issues and you've canned the log we can't help you
6 2012-09-20 00:45:44 <gmaxwell> Guest7582: give us Jeff back!
7 2012-09-20 00:45:53 <gmaxwell> 19:41 < B0g4r7> man they're tearing into bfl on the forum
8 2012-09-20 00:45:56 <gmaxwell> 0_o
9 2012-09-20 00:46:32 <jgarzik_> yeah, saw that
10 2012-09-20 00:46:48 <jgarzik_> I saw "mail fraud", did not see "mail fraud ponzi"
11 2012-09-20 00:48:35 <gmaxwell> okay, perhaps better then; lots of ways to get a mail fraud conviction...
12 2012-09-20 00:49:25 <gmaxwell> international lottery scam. 0_o
13 2012-09-20 00:49:43 <MrTiggr> ^^ that
14 2012-09-20 00:49:52 <Gladamas> ya, apparently it was that ^^
15 2012-09-20 00:49:52 <MrTiggr> there were 5 or 6 others convicted too
16 2012-09-20 00:49:56 <Gladamas> ya
17 2012-09-20 00:50:07 <MrTiggr> he was held in italy then extradited to usa
18 2012-09-20 00:50:08 <Gladamas> in fact the bfl ceo wasnt the main defendant
19 2012-09-20 00:50:12 <MrTiggr> yerp
20 2012-09-20 00:51:29 <MrTiggr> not really sure that there is anythign to see there .. i have all the d0x i need if things went sour but it seems a bit like frantic panicked d0x-dropping to me - we've seen it all too often on the forums
21 2012-09-20 00:51:42 <gmaxwell> on the flip side, felony convictions really hurt people's employment prospects; I'd expect a higher density of self employed white-collar excons than would otherwise be expected.
22 2012-09-20 00:57:45 <MrTiggr> totally agreed
23 2012-09-20 00:58:04 <MrTiggr> i did wonder if running BFL might be a breach of his Parole/Probation terms
24 2012-09-20 01:00:36 <MrTiggr> "The defendant shall not transfer, sell, give away, or otherwise convey any asset with a fair market value in excess of $500 without approval of the Probation Officer until all financial obligations imposed by the Court have been satisfied in full."
25 2012-09-20 01:02:07 <gmaxwell> MrTiggr: presumably he's off probation, no?
26 2012-09-20 01:06:59 <MrTiggr> well im not sure
27 2012-09-20 01:07:15 <MrTiggr> it was 2009 he was extradited from italy
28 2012-09-20 01:07:26 <MrTiggr> 2 years sentence iirc
29 2012-09-20 01:07:46 <MrTiggr> so its possible that he is still under some sort of probation or Parole
30 2012-09-20 01:08:28 <MrTiggr> but i do NOT know for sure ... seems kinda FUDy to me - i've grabbed all the info i can for later if it goes sour ...moving on
31 2012-09-20 01:09:11 <gmaxwell> er. yea, almost certantly. The two year sentence is somewhat surprising... $19 million. The federal points system makes it unusual for people to get short sentences when they can attach big dollar values.
32 2012-09-20 01:10:17 <MrTiggr> which is why i actually suspect he is out on "good behaviour" and on an extended parole/probo would be my guess ... i bet he would cop shit from his PO for BFL
33 2012-09-20 01:15:13 <crazyMF> Is there any chance of seeing some multi-core action in bitcoind ?
34 2012-09-20 01:23:41 <gmaxwell> MrTiggr: nah, the federal system abolished parole, you have to server ~87% of your sentence, probably got shorter time due to cooperation credits or something. (so says my token ex-con contact I just asked :P)
35 2012-09-20 01:26:14 <MrTiggr> :D /me is not USA ( /me is Aus.) so im not 100% on all that law side of things :D i'd lean towards trusting your ex-con :D
36 2012-09-20 01:27:06 <EasyAt> gmaxwell: I believe that is a state by state thing
37 2012-09-20 01:27:22 <gmaxwell> EasyAt: This is all federal.
38 2012-09-20 01:27:58 <EasyAt> oh wow
39 2012-09-20 01:28:33 <EasyAt> That'll be tough on the already over populated prisons
40 2012-09-20 01:30:19 <EasyAt> I guess it was abolishe din '84. nvm
41 2012-09-20 01:31:12 <gmaxwell> yea "why do you think they're over populated" :P
42 2012-09-20 01:31:46 <EasyAt> ACTION goes back to the kids table :)
43 2012-09-20 01:32:46 <gmaxwell> well, I didn't know any of this before the aformentioned friend ended up going to prison.
44 2012-09-20 01:43:50 <galambo> when i buy specialty hardware i make sure the company management are international felons
45 2012-09-20 01:47:47 <JFK911> galambo: right on. that broadcom guy kicks ass
46 2012-09-20 01:52:11 <MrTiggr> that sounds like one of those posters "i dont often buy specialy hardware .. but when i do: <insert funny>"
47 2012-09-20 01:54:29 <galambo> it would be funny if the guy has totally turned over a new leaf but organizes the company like a scam because that's all he knows (don't use last names, lol)
48 2012-09-20 01:55:31 <MrTiggr> im pretty sure there are/have been some major playahs in bitcoin that operate EXACTLY like that galambo
49 2012-09-20 01:56:55 <gmaxwell> galambo: hah
50 2012-09-20 04:09:17 <jgarzik_> amiller: basic DHT in place, in pybond
51 2012-09-20 08:54:19 <wumpus> can someone send me some testnet3 coins at mrpHsZRyP9nUKcFnokN4xpEtstddJLZDAt ? thanks
52 2012-09-20 08:55:39 <Luke-Jr> wumpus: it seems my build.h hasn't been updated since April despite many different builds since then - any ideas? :/
53 2012-09-20 08:56:00 <otimm> wumpus: http://testnet.freebitcoins.appspot.com/
54 2012-09-20 08:56:29 <wumpus> Luke-Jr: just remove it
55 2012-09-20 08:56:41 <Luke-Jr> wumpus: isn't it supposed to be updated automaticalyl?
56 2012-09-20 08:56:46 <wumpus> not if it already exists
57 2012-09-20 08:56:54 <wumpus> afaik
58 2012-09-20 08:56:57 <Luke-Jr> O.o
59 2012-09-20 08:57:18 <wumpus> not sure though
60 2012-09-20 08:57:31 <wumpus> otimm: thanks
61 2012-09-20 08:58:20 <wumpus> I'm not sure how smart the build.h generator is; if it would write the flie every build, it would result in a make of a few .o files and a link every time
62 2012-09-20 08:59:46 <wumpus> so it should only write the file if something actually changed in the git info it retrieves.. but maybe it does that, I don't know
63 2012-09-20 09:04:03 <Luke-Jr> wumpus: it does, apparently the file moved to a build/ dir, but the compiler is configured to prefer the one in src/ :/
64 2012-09-20 09:06:26 <wumpus> yes to file moved to build to support shadow builds (ie, building in another directory/filesystem that the source is in) with qmake
65 2012-09-20 09:09:21 <Luke-Jr> right, so the include order just needs to be fixed
66 2012-09-20 09:09:38 <Luke-Jr> weird, even with build.h corrected and make clean'd, I can't get bitcoin-qt's debug window to show the right one
67 2012-09-20 09:09:50 <wumpus> maybe, or you should just remove the build.h in the wrong place, after all it shouldn't ever re-make it
68 2012-09-20 09:10:08 <Luke-Jr> I did, but that doesn't fix the include order being wrong :p
69 2012-09-20 09:11:11 <wumpus> there should only be one build.h so include order doesn't matter
70 2012-09-20 10:24:38 <ielo> hey, does anyone know where i can find kisom?
71 2012-09-20 10:24:51 <sipa> who or what is kisom?
72 2012-09-20 10:25:23 <ielo> hi sipa
73 2012-09-20 10:25:31 <ielo> he used to do some bitcoin dev work i think
74 2012-09-20 10:25:50 <ielo> he once sent me a link to a paper and now i cant find it
75 2012-09-20 10:25:53 <ielo> like 2 years ago
76 2012-09-20 10:27:19 <sipa> no such name in the commit history or on the forum
77 2012-09-20 10:27:20 <LolcustBackup> what was the paper about, lelo ?
78 2012-09-20 10:29:21 <ielo> LolcustBackup, there was one paper about a digital currency predating bitcoin, it was hypothetical, it wasn't that old either
79 2012-09-20 10:29:51 <ielo> LolcustBackup, and there was also a bunch of papers about cypherpunk stuff, and those ones all came from one place, like onion routing and stuff
80 2012-09-20 10:31:14 <ielo> sipa, is there any name like kyle isom
81 2012-09-20 10:32:08 <sipa> can't find anything, no
82 2012-09-20 10:32:55 <ielo> sipa, thanks anyhow
83 2012-09-20 11:36:43 <Luke-Jr> Is iso646.h allowed in bitcoind code? :P
84 2012-09-20 11:37:27 <sipa> why...?
85 2012-09-20 11:38:53 <Luke-Jr> sipa: j/w
86 2012-09-20 11:39:06 <Luke-Jr> that's basically what came to mind when I stumbled upon it :P
87 2012-09-20 11:39:11 <Luke-Jr> ("why????")
88 2012-09-20 12:27:00 <sipa> gmaxwell: is the coverage report of the canonical branch up to date>
89 2012-09-20 12:27:02 <sipa> ?
90 2012-09-20 15:10:51 <knotwork> bitcoind died again: bitcoin-rpchand[11220] general protection ip:4dcffa sp:7fea713d4be0 error:0 in bitcoind[400000+3a0000]
91 2012-09-20 15:13:56 <knotwork> gdb says args is undefined command , to use help. help shows nothing seemingly to do with passing args
92 2012-09-20 15:16:04 <knotwork> ok found set args.
93 2012-09-20 15:16:34 <knotwork> but then gdb said detatching from child process so I suspect it actually fired up the daemon in background instead of tracing its exacution
94 2012-09-20 15:19:46 <knotwork> ok killed it and tried leaving out the -daemon arg
95 2012-09-20 15:28:40 <knotwork> ok seems to be running under gdb okay now
96 2012-09-20 17:01:14 <BlueMattBot> Project Bitcoin build #67: FAILURE in 4 hr 10 min: http://jenkins.bluematt.me/job/Bitcoin/67/
97 2012-09-20 17:59:57 <jgarzik> gavinandresen: seen the osx crash reports?
98 2012-09-20 18:00:11 <jgarzik> gavinandresen: there are several reports of problems on osx, specifically
99 2012-09-20 18:01:00 <gavinandresen> I know of two people reporting crash-on-startup running osx 10.5
100 2012-09-20 18:01:06 <gavinandresen> are there more?
101 2012-09-20 18:01:25 <gavinandresen> (and anybody crashing NOT on 10.5 ?)
102 2012-09-20 18:25:38 <jgarzik> gavinandresen: one 10.5, one "leopard", one "lion"
103 2012-09-20 18:25:59 <jgarzik> #1842 is lion
104 2012-09-20 18:26:02 <gavinandresen> leopard == 10.5...
105 2012-09-20 18:26:04 <jgarzik> the first two are forum
106 2012-09-20 18:26:48 <gavinandresen> 1842 I hadn't seen, that's disturbing
107 2012-09-20 18:27:32 <jgarzik> gavinandresen: gcc is the osx compiler? is all the -pthread gadgetry set, for both compile and link steps?
108 2012-09-20 18:28:20 <jgarzik> gavinandresen: any chance of a bitcoin <-> dependent library mismatch causing problems this severe?
109 2012-09-20 18:28:21 <gavinandresen> jgarzik: should be, -pthread in both LIBS and CFLAGS
110 2012-09-20 18:29:14 <gavinandresen> jgarzik: there is a chance, but all the dependent libraries should be shipped as part of the .app bundle
111 2012-09-20 18:29:30 <jgarzik> gavinandresen: compile flag mismatch between bitcoin core and qt?
112 2012-09-20 18:29:51 <gavinandresen> no, I make clean and only compile qt when building releases
113 2012-09-20 18:30:52 <gavinandresen> there IS a newer version of the macports qt4-mac available, I wonder if the version I was compiling against had issues....
114 2012-09-20 18:30:56 <eian> Does anyone know what this page means: http://blockchain.info/charts/cost-per-transaction
115 2012-09-20 18:31:55 <eian> ah, nevermind - found the description
116 2012-09-20 18:32:07 <jgarzik> gavinandresen: I was also thinking of the possibility that _makefiles_ apply differences between core and qt that a 'make clean' might not catch. i.e. like the recent crashes fixed by -DBOOST_SPIRIT_THREADSAFE
117 2012-09-20 18:32:27 <jgarzik> anyway, just throwing spaghetti against a wall, as it were, to see if it sticks
118 2012-09-20 18:32:31 <jgarzik> ACTION looks at diffs
119 2012-09-20 18:32:46 <jgarzik> another common platform-specific source of headaches has been the underlying locking/sync stuff
120 2012-09-20 18:32:57 <gavinandresen> jgarzik: harumph.... looking at the qt-generated makefile I see no -pthread stuff....
121 2012-09-20 18:33:28 <jgarzik> gavinandresen: yeah, that is the sort of problem I was thinking about
122 2012-09-20 18:33:57 <gavinandresen> ACTION wonders how to tell qmake to to the pthread thing....
123 2012-09-20 18:34:09 <jgarzik> make sure link gets it, as well as compile
124 2012-09-20 18:56:11 <sipa> gavinandresen: manually add -pthread to LIBS and to QMAKE_CXXFLAGS ?
125 2012-09-20 18:57:51 <gavinandresen> sipa: I'll compile that way and see if it helps. Wish I understood qmake better, there is a QMAKE_LFLAGS_THREAD setting but I don't know how to tell qmake to use it
126 2012-09-20 18:58:12 <jgarzik> BFL_Sonny - "Hi, my name is Sonny Vleisides" - https://bitcointalk.org/index.php?topic=110868.0
127 2012-09-20 19:00:32 <Luke-Jr> gavinandresen: quick Google suggests CONFIG += thread
128 2012-09-20 19:00:36 <Luke-Jr> wonder why we don't have that already
129 2012-09-20 19:00:43 <Luke-Jr> thread\tThe target is a multi-threaded application or library. The proper defines and compiler flags will automatically be added to the project.
130 2012-09-20 19:01:43 <sipa> ha!
131 2012-09-20 19:01:54 <sipa> that sounds like what we need
132 2012-09-20 19:02:03 <gavinandresen> Luke-Jr: thanks, it works if I do: qmake CONFIG+=thread QMAKE_CFLAGS_THREAD+=-pthread QMAKE_LFLAGS_THREAD+=-pthread ... etc
133 2012-09-20 19:03:37 <Luke-Jr> so I wonder if this is CVE-2012-1910 all over again?
134 2012-09-20 19:09:07 <Luke-Jr> sigh, seems even deleting src/build.h my builds still say 0.6.0-whatever
135 2012-09-20 19:09:45 <xisalty> lol
136 2012-09-20 19:09:50 <xisalty> You ought to fix that
137 2012-09-20 19:10:23 <sipa> Luke-Jr: heh, how is that possible?
138 2012-09-20 19:10:29 <Luke-Jr> sipa: I don't know :/
139 2012-09-20 19:10:36 <sipa> Luke-Jr: or is 0.6.0 the last tag it knows about?
140 2012-09-20 19:10:42 <Luke-Jr> nope
141 2012-09-20 19:10:46 <Luke-Jr> build/build.h is correct
142 2012-09-20 19:10:56 <sipa> and it's a Qt build?
143 2012-09-20 19:11:00 <Luke-Jr> yes
144 2012-09-20 19:11:17 <sipa> not using a build in some alternate directory?
145 2012-09-20 19:11:28 <Luke-Jr> no, just building in the same dir as the .pro file
146 2012-09-20 19:12:18 <sipa> gavinandresen: just CONFIG+=thread is not enough?
147 2012-09-20 19:12:25 <gavinandresen> sipa: no
148 2012-09-20 19:12:45 <sipa> not even if you do it in the .pro file?
149 2012-09-20 19:12:57 <gavinandresen> I'll try that...
150 2012-09-20 19:14:26 <gavinandresen> sipa: no, doesn't work. there is no -pthread stuff in the osx qmake template folder doo-hickey
151 2012-09-20 19:15:01 <gavinandresen> I'm in the middle of upgrading my macports qt4, so I'm going to let that finish before doing more experimenting.
152 2012-09-20 19:31:12 <eian> Is bitcoin considered a synchronous distributed system?
153 2012-09-20 19:31:21 <sipa> it's not synchronous
154 2012-09-20 19:31:39 <eian> I suppose I mean academically
155 2012-09-20 19:31:50 <eian> http://groups.csail.mit.edu/tds/papers/Lynch/pods83-flp.pdf
156 2012-09-20 19:31:59 <Diablo-D3> you mean asynchronous but strictly ordered
157 2012-09-20 19:32:27 <eian> I'm trying to map other literature to what I understand about the protocol
158 2012-09-20 19:32:31 <Diablo-D3> bitcoin trades latency for hard ordering
159 2012-09-20 19:32:31 <sipa> eian: bitcoin does not achieve consensus
160 2012-09-20 19:32:37 <sipa> in the traditional sense of the word
161 2012-09-20 19:32:44 <Diablo-D3> sipa: it "does"
162 2012-09-20 19:32:47 <Diablo-D3> but it uses math to do it
163 2012-09-20 19:32:52 <sipa> Diablo-D3: no, it doesn't
164 2012-09-20 19:32:59 <sipa> nodes do not necessarily agree
165 2012-09-20 19:33:17 <Diablo-D3> I mean, in the sense that all the nodes have to agree blocks are valid
166 2012-09-20 19:33:31 <sipa> the assumption is that the chance that they disagree about things further in the past decreases exponentially
167 2012-09-20 19:33:39 <sipa> but there is always a non-zero chance for reorganisation
168 2012-09-20 19:33:40 <Diablo-D3> sipa: that too
169 2012-09-20 19:33:51 <sipa> which means there is no consensus in the traditional definition
170 2012-09-20 19:34:06 <Diablo-D3> if you strictly use a traditional definition? Ill agree with that
171 2012-09-20 19:34:17 <sipa> eian: if you want to know more about the theory, talk to amiller
172 2012-09-20 19:34:21 <Diablo-D3> but that whole way of thought has sort of fallen to the wayside in quite a few projects in the past few years
173 2012-09-20 19:34:26 <sipa> he did a talk about these things in london
174 2012-09-20 19:34:35 <eian> are the slides up somewhere?
175 2012-09-20 19:34:39 <eian> or video or something?
176 2012-09-20 19:34:41 <sipa> not yet, i think
177 2012-09-20 19:34:44 <eian> :(
178 2012-09-20 19:34:48 <eian> ok thanks - I will do that
179 2012-09-20 19:35:10 <Diablo-D3> btw, sipa, I want your opinion
180 2012-09-20 19:35:17 <Luke-Jr> sipa: fwiw, removing my Makefile seems to have fixed it somehow
181 2012-09-20 19:35:23 <Diablo-D3> you understand what db transactions are, right?
182 2012-09-20 19:36:03 <sipa> yes
183 2012-09-20 19:36:13 <Diablo-D3> would it be worth doing concurrent transactions that name previous transactions like how bitcoin names previous blocks to ensure ordering in a global database store?
184 2012-09-20 19:36:26 <Diablo-D3> (and it can name multiple past transactions)
185 2012-09-20 19:36:40 <Diablo-D3> it'll store a log of the past, oh, 24 hours or whatever
186 2012-09-20 19:37:00 <Diablo-D3> plus the last one that effected an object (row, whatever) in the db
187 2012-09-20 19:38:21 <sipa> hmm, maybe if you want to have several separate candidate timelines, and aren't sure yet which one will be chosen
188 2012-09-20 19:38:40 <Diablo-D3> well, there should only ever be one, but this should stomp out accidental forks
189 2012-09-20 19:38:45 <Diablo-D3> whoever has the longest chain wins
190 2012-09-20 19:39:11 <Diablo-D3> Im trying to think of an eventual consistency setup that isnt retarded
191 2012-09-20 19:39:24 <sipa> like when a transaction would conflict with another, instead of abort and roll back immediately, you allow both to continue without seeing eachother's effects, and choose at some later point which one wins
192 2012-09-20 19:39:30 <sipa> if not, i think there are more efficient solutions
193 2012-09-20 19:39:54 <sipa> Luke-Jr: that makes no sense at all...
194 2012-09-20 19:40:01 <Diablo-D3> not abort and roll back, because that cant be handled efficiently at the sync layer
195 2012-09-20 19:40:14 <Diablo-D3> (that'd have to be handled in the frontend of that code)
196 2012-09-20 19:40:14 <Luke-Jr> sipa: I agree, but it is what it is :/
197 2012-09-20 19:40:23 <Luke-Jr> sipa: I guess the Makefile lacks a rule to remake itself?
198 2012-09-20 19:40:32 <Diablo-D3> but something like, use hashes, pick whichever number is smaller if there isnt a chain that wins
199 2012-09-20 19:40:47 <Diablo-D3> on databases that rarely back to back commit on the same objects, nothing will ever happen
200 2012-09-20 19:41:08 <sipa> why are you trying to write a database engine?
201 2012-09-20 19:41:19 <Diablo-D3> for fun.
202 2012-09-20 19:41:27 <sipa> very good reason :)
203 2012-09-20 19:41:34 <Luke-Jr> lol'
204 2012-09-20 19:41:46 <Diablo-D3> hell, same thing goes with that rather strange base C lib Im writing
205 2012-09-20 19:41:49 <Diablo-D3> for fun
206 2012-09-20 19:42:11 <Diablo-D3> get this, the most primary part of my base lib is an STM impl
207 2012-09-20 19:42:38 <Diablo-D3> everything else in the lib uses it to read and write to memory
208 2012-09-20 19:42:43 <Diablo-D3> including the malloc impl
209 2012-09-20 19:42:56 <Diablo-D3> delicious mandatory thread safety, om nom nom
210 2012-09-20 19:43:13 <Diablo-D3> ACTION goes afk to get food before he eats the lib
211 2012-09-20 20:04:45 <lianj> is 1dice having a hard time atm, seeing chains of unconfirmed ones
212 2012-09-20 20:15:14 <lianj> looks like this one is the origin of the chain, http://blockchain.info/tx-index/70326c78601f000c39f8c5094274777d1ef0dcdce110082825ff657d741905c4
213 2012-09-20 21:16:49 <amiller> <amiller> jgarzik, neat dht
214 2012-09-20 21:16:51 <amiller> <amiller> i don't understand your keyspace partitioning / distance metric
215 2012-09-20 21:16:53 <amiller> <amiller> node.node_id ^ key_num ( https://github.com/jgarzik/pybond/blob/master/dht.py#L223 )
216 2012-09-20 21:16:55 <amiller> <amiller> actually nvm i think that's precisely the key distance used in kademlia
217 2012-09-20 21:16:57 <amiller> <amiller> btw i'm still trying to simplify pynode by replacing httpsrv/asynchat with gevent, but i have to fix my broken bitcoin installation first
218 2012-09-20 21:17:52 <amiller> eian, yeah bitcoin is a synchronous protocol, aka the "probably almost completely reliable" network model http://www.mail-archive.com/cryptography%40metzdowd.com/msg10001.html
219 2012-09-20 21:17:57 <jgarzik> amiller: thanks for the resend
220 2012-09-20 21:18:13 <jgarzik> amiller: I'll switch to your version, if you can get http working
221 2012-09-20 21:18:21 <amiller> eian, i put a bunch of background material in here https://bitcointalk.org/index.php?topic=99631.0
222 2012-09-20 21:18:21 <jgarzik> amiller: that's rather the deal breaker :/
223 2012-09-20 21:18:42 <jgarzik> amiller: the current changes to node.py look 100% OK
224 2012-09-20 21:18:59 <amiller> jgarzik, i'm gonna do it for pynode first
225 2012-09-20 21:19:08 <jgarzik> amiller: cool!
226 2012-09-20 21:19:09 <amiller> jgarzik, that's the only way i'll be able to test that it's working correctly, pretty much like you said
227 2012-09-20 21:19:29 <jgarzik> amiller: if you fix pynode for that, that means I can very easily upgrade pynode to full node status
228 2012-09-20 21:20:36 <jgarzik> amiller: a HUGE problem with asyncore, and another reason it is only a rapid prototyping toy, is that it lacks any concept of timers and timeouts. modern asynchronous P2P has all sorts of background, periodic tasks
229 2012-09-20 21:20:44 <jgarzik> gevent should solve that
230 2012-09-20 21:23:01 <jgarzik> amiller: yes, my DHT is a cheesy Kademlia knock-off
231 2012-09-20 21:23:25 <jgarzik> amiller: though I have done other DHT work as well, such as http://www.nicemice.net/amc/research/tangle/
232 2012-09-20 21:23:37 <jgarzik> (not my research, but I worked quite a bit on the implementation, after its initial release)
233 2012-09-20 21:23:56 <amiller> ah that's very cool, i was gonna say i was totally impressed you hacked out a kademlia that fast
234 2012-09-20 21:24:16 <Luke-Jr> jgarzik: do you plan to support current Python releases btw?
235 2012-09-20 21:24:39 <jgarzik> amiller: the server side is trivial, because it (wisely!) never does any recursive lookups
236 2012-09-20 21:24:58 <jgarzik> amiller: it's all dumb data serving, really
237 2012-09-20 21:25:12 <amiller> jgarzik, what does it take to make such a thing DoS resistant and suitable for public nodes?
238 2012-09-20 21:25:17 <jgarzik> Luke-Jr: I accept python3 patches, if that's what you are asking
239 2012-09-20 21:25:28 <jgarzik> amiller: well...
240 2012-09-20 21:25:45 <Luke-Jr> jgarzik: ok, I'll try to find time one of these weeks..
241 2012-09-20 21:25:47 <jgarzik> amiller: no known DHT technology is DoS/Sybil proof really
242 2012-09-20 21:26:27 <jgarzik> Luke-Jr: ideal is supporting python 2.7 + python 3.x. python 2.[recently] remains a hard requirement for some years to come.
243 2012-09-20 21:26:41 <Luke-Jr> jgarzik: that's slightly harder ;)
244 2012-09-20 21:26:45 <jgarzik> amiller: I was thinking about requiring a node-id generated from a proof of work
245 2012-09-20 21:27:04 <jgarzik> amiller: makes sybil and some other attacks more expensive
246 2012-09-20 21:27:22 <Luke-Jr> today I observe the sad state of how busy I am in terms of reaction time to code change requests; I think it took me 2 weeks to address an issue gmaxwell/gavin brought up :<
247 2012-09-20 21:28:53 <jgarzik> amiller: outside of that... just gotta think of sane hueristics, like limiting total data traffic from each A.B.C.D/24
248 2012-09-20 21:29:36 <jgarzik> amiller: cheap botnets make swiss cheese of even those defenses, so that mainly addresses stupid-software unintentional attacks more than anything else, I think
249 2012-09-20 21:30:09 <jgarzik> proof-of-work can increase the cost to store (and therefore flood) data
250 2012-09-20 21:30:10 <amiller> jgarzik, at least the problem is no harder / easier than what the ordinary bitcoin clients face
251 2012-09-20 21:30:44 <amiller> so yeah, that all sounds right, and i trust your judgment regarding sane heuristics :p
252 2012-09-20 21:31:04 <jgarzik> rofl, who says I have good judgement???
253 2012-09-20 21:31:06 <jgarzik> ACTION runs
254 2012-09-20 21:31:37 <amiller> jgarzik, related to proof of work, another option i thought of once is to require bitcoins in escrow in the blockchain
255 2012-09-20 21:32:17 <jgarzik> amiller: yeah, you could require a signed message from a key owning an unspent txout
256 2012-09-20 21:32:41 <jgarzik> signed messages are out-of-band (with respect to bitcoin), so no blockchain bloat
257 2012-09-20 21:36:56 <jgarzik> ACTION wonders, then, the best way (p2p or rpc) to check and see if a txout is (a) valid and (b) unspent
258 2012-09-20 21:37:28 <sipa> no objection for an RPC to check that
259 2012-09-20 21:43:46 <Luke-Jr> jgarzik: your own, or arbitrary?
260 2012-09-20 21:44:00 <jgarzik> arbitrary
261 2012-09-20 21:46:01 <sipa> jgarzik: imho, the prime purpose of a validation node (one which is not intended for downloading blocks) is exactly tracking which outputs are still available, so it would make sense to have an RPC for that
262 2012-09-20 21:46:51 <jgarzik> sipa: what about P2P? (not disputing an RPC's usefulness... just trying to think it through)
263 2012-09-20 21:48:12 <sipa> i don't see a usecase directly for exposing that information via P2P, but i don't see a reason why it would be a problem
264 2012-09-20 21:55:34 <jgarzik> sipa: agree ^2
265 2012-09-20 22:02:15 <sipa> i really wonder why people mail "i have a business transaction for you", assuming they can get my interest without saying anything...
266 2012-09-20 22:08:09 <diki> sipa:That's called SPAM.
267 2012-09-20 22:10:18 <sipa> i have reason to assume they do actually have something to tell me
268 2012-09-20 22:12:29 <graingert> sipa: I have a buisness transaction for you
269 2012-09-20 22:29:57 <amiller> man, i think i just figured out something grand
270 2012-09-20 22:29:59 <amiller> i've been fussing about what sort of behavior model to use for bitcoin
271 2012-09-20 22:30:20 <amiller> "mostly unconditional-honest" doesn't seem too realistic
272 2012-09-20 22:30:37 <amiller> "rational" seems too difficult though, since i think many people mine at a loss
273 2012-09-20 22:31:05 <amiller> so the right model is the one that results in people playing at lotteries!
274 2012-09-20 22:31:36 <amiller> bitcoin is basically a cross between a lottery and an election, nearly every state has a lottery of some kind that pays out negative-expected value, and the proceeds are used for social benefit
275 2012-09-20 22:32:26 <amiller> in a rational model no one plays the lottery, but in an altruistic model lotteries aren't needed since everyone just donates all the time
276 2012-09-20 22:33:42 <sipa> you should start a field of science called computational sociology
277 2012-09-20 22:34:32 <amiller> sipa, heh.
278 2012-09-20 22:34:37 <amiller> since you mentioned that, i wanted to point out this neat paper http://eli.informatics.indiana.edu/CHI2012_Collapse_CameraReadyFinalClean.pdf
279 2012-09-20 22:34:51 <amiller> many computer science papers begin with some kind of sociological justification, typically moores law
280 2012-09-20 22:35:08 <amiller> "computing power seems to double every couple of years, therefore we're justified in putting our attention on XXXX topic that uses lots of computation"
281 2012-09-20 22:35:10 <jgarzik> amiller: I think there is a fair amount of altruism (we unpaid devs) and a fair amount of deluded self-interest ("I'll get rich!!!!" ... invests in Ponzis)
282 2012-09-20 22:35:55 <amiller> these people suggest a new way to start your paper, it goes something like "we're apparently on the brink of a cataostrophic collapse of civilization, therefore computational topic XXX is of interest..."
283 2012-09-20 22:36:44 <amiller> jgarzik, sweet, so bitcoin apparently harnesses the limitless power of deluded self-interest :p
284 2012-09-20 22:37:16 <jgarzik> just like every politician in existence relies on "low information voters" to stay in power ;p
285 2012-09-20 22:37:24 <jgarzik> </sarcasm>
286 2012-09-20 22:37:50 <amiller> consider a group of individuals who play a losing lottery, $1 for a 0.9 expected return
287 2012-09-20 22:38:03 <amiller> if they would simply cooperate with each other, they could run their own lottery for a higher rate of return
288 2012-09-20 22:38:18 <amiller> the problem is they're deluded, self-interested, and mutually untrusting
289 2012-09-20 22:38:49 <amiller> they're better off mining in that case
290 2012-09-20 22:40:07 <sipa> i wonder whether some long-term-prospect-rational way isn't appropriate: the reason people run a node is because it provides them with the ability to do X (insert something bitcoin offers), and that outweighs the cost
291 2012-09-20 22:40:41 <sipa> mining may some day be the same: people mine at a slight loss because they want to protect the network, as a 51% attack may harm the usefulness of the entire system
292 2012-09-20 22:41:31 <amiller> ACTION digs for an explanation of why people are willing to accept a low-expected-value, high-uncertainty option as long as the potential upside is glorious enough
293 2012-09-20 22:41:48 <jgarzik> it is my fear that the biggest reason why people today run a validating node is... that's the default for the reference software. not out of specific intent.
294 2012-09-20 22:41:54 <sipa> oh, take into account human's inability to estimate chances
295 2012-09-20 22:42:44 <jgarzik> incentives against running a validating node seem like they will increase in the future
296 2012-09-20 22:44:12 <sipa> i wonder what model could explain gambling though...
297 2012-09-20 22:44:14 <amiller> jgarzik, well by changing the proof-of-work puzzle we may at least be able to remove the incentive _against_ validation :/ but i accept that i need to pick my battles and this one is... out there
298 2012-09-20 22:45:15 <jgarzik> come up with an untrusted model for paying $the network
299 2012-09-20 22:45:18 <jgarzik> then it's self-sustaining
300 2012-09-20 22:45:55 <jgarzik> i.e. you are paid some tiny amount for running a validating node
301 2012-09-20 22:46:38 <amiller> jgarzik, how might you go about 'proving' to an observer that you're indeed running a full node
302 2012-09-20 22:46:38 <jgarzik> otherwise it is as you said... people run a node because bitcoin provides them with some ability
303 2012-09-20 22:46:43 <jgarzik> and they get value that way
304 2012-09-20 22:46:52 <jgarzik> amiller: that is The Question ;p
305 2012-09-20 22:47:31 <amiller> yeah and 'RAFT' is roughly the answer http://dspace.mit.edu/openaccess-disseminate/1721.1/72402
306 2012-09-20 22:47:53 <jgarzik> amiller: others may test and sample and observe behavior compared to their own datasets... but collating that data such that everyone is credited work in a cheat-proof manner. hard.
307 2012-09-20 22:48:36 <sipa> use a challenge-response that requires traversing a part of the utxo set?
308 2012-09-20 22:48:50 <amiller> sipa, right!
309 2012-09-20 22:48:54 <amiller> now let miners choose their own challenges
310 2012-09-20 22:49:27 <amiller> and among their successfully answered challenges are a small number of 'winners' that earn them a reward
311 2012-09-20 22:58:01 <gmaxwell> 17:33 < sipa> you should start a field of science called computational sociology
312 2012-09-20 22:58:25 <gmaxwell> I think you want sociology of computation; computational sociology would be Asimov's Foundation universe. :P
313 2012-09-20 22:58:52 <Diablo-D3> hell
314 2012-09-20 22:58:57 <Diablo-D3> we have a company called us robotics
315 2012-09-20 22:59:18 <gmaxwell> amiller: mining as lottery is contraindicated by the existance of pools. :P
316 2012-09-20 23:00:04 <maaku> "computational sociology" sounds like a nice academic phrase for crypto-anarchism :P
317 2012-09-20 23:03:07 <amiller> gmaxwell, i'm pretty sure we're living through asimov's foundation universe
318 2012-09-20 23:03:11 <amiller> it's how i learned to cope with a falling empire
319 2012-09-20 23:03:44 <amiller> and build on the work of a brilliant absentee mathematician
320 2012-09-20 23:04:07 <amiller> i think our first crisis was the botnets
321 2012-09-20 23:08:57 <amiller> maybe satoshi will reveal hidden blockchain messages to us every so often, that would be neat
322 2012-09-20 23:18:50 <Luke-Jr> "computational sociology" aka psychohistory???
323 2012-09-20 23:38:02 <gmaxwell> Wallet encryption ought to tell you that you need to redo your backups after you encrypt your wallet. I talked to someone last night who got burned by <create wallet> <backup> <send funds> <encrypt> <perform a transaction> <laptop stolen>.
324 2012-09-20 23:40:02 <Joric> hows that if address pool stays the same
325 2012-09-20 23:40:27 <sipa> it doesn't
326 2012-09-20 23:40:45 <Joric> it what??
327 2012-09-20 23:41:05 <sipa> encrypting resets the address pool, as otherwise pre-encrypt copies are vulnerable to theft
328 2012-09-20 23:41:49 <sipa> by resetting the pool, you make sure no active address has ever had its keys touch disk in unencrypted form
329 2012-09-20 23:41:51 <gmaxwell> Joric: it retains the keys, but marks them all used, and fills in a whole new set.
330 2012-09-20 23:54:54 <Diablo-D3> sipa: oh fuck
331 2012-09-20 23:54:56 <Diablo-D3> seriously?
332 2012-09-20 23:55:05 <Joric> NACK!
333 2012-09-20 23:55:16 <Diablo-D3> sipa: I think we need an option for that
334 2012-09-20 23:55:24 <Diablo-D3> encrypt, and encrypt meharder
335 2012-09-20 23:55:35 <Diablo-D3> because I always encrypt the actual backups