1 2012-10-23 00:04:44 <gmaxwell> I'm due for a new laptop, but I don't want to lose my 1400x1050 IPS 12.1" display. Hoping that at the end of the year some reasonable vendors will ship some high ppi parts.
2 2012-10-23 00:05:34 <BlueMatt> meh, just go for a 13" and you can get something close
3 2012-10-23 00:06:04 <BlueMatt> my 13 is 1600x900
4 2012-10-23 00:06:10 <BlueMatt> no, 14, sorry
5 2012-10-23 00:06:23 <phantomcircuit> gmaxwell, asus zenbook series
6 2012-10-23 00:06:28 <phantomcircuit> 1920x1080 13.1"
7 2012-10-23 00:06:34 <BlueMatt> ooo
8 2012-10-23 00:07:24 <gmaxwell> I understand samsung will be shipping a 2560x1440 13.3" product this year.
9 2012-10-23 00:08:14 <phantomcircuit> i would buy that in a flash
10 2012-10-23 00:08:32 <gmaxwell> though IIRC their laptops have crap keyboards, so I might be waiting a bit longer.
11 2012-10-23 00:10:58 <BlueMatt> gmaxwell: ooooooo
12 2012-10-23 00:11:19 <BlueMatt> apple will, but a pc...
13 2012-10-23 00:31:33 <BlueMattBot> Project Bitcoin build #114: STILL FAILING in 1 hr 51 min: http://jenkins.bluematt.me/job/Bitcoin/114/
14 2012-10-23 00:32:13 <BlueMatt> :(
15 2012-10-23 00:32:27 <BlueMatt> /mnt/jenkins/jobs/Bitcoin/workspace/src/leveldb/libleveldb.a: No such file or directory
16 2012-10-23 00:32:32 <BlueMatt> i586-mingw32msvc-g++: /mnt/jenkins/jobs/Bitcoin/workspace/src/leveldb/libmemenv.a: No such file or directory
17 2012-10-23 00:32:45 <BlueMatt> sipa: ...
18 2012-10-23 00:35:49 <gmaxwell> woot. has coverage now.
19 2012-10-23 00:36:02 <gmaxwell> https://people.xiph.org/~greg/bitcoin_coverage/coverage.ultraprune.testbitcoin/home/gmaxwell/src/bax/src/index.html < just test_bitcoin for the moment, still need to run the other tests.
20 2012-10-23 00:39:24 <SomeoneWeird> u broke
21 2012-10-23 00:41:21 <gmaxwell> SomeoneWeird: hm?
22 2012-10-23 00:44:42 <SomeoneWeird> how large is the testnet now?
23 2012-10-23 00:46:04 <gmaxwell> SomeoneWeird: define large?
24 2012-10-23 00:46:17 <lianj> MB
25 2012-10-23 00:46:18 <xblitz> I dont know if this has been noticed already.. I havent seen it on the forums.. I have been keeping a fork of bitcoind .. and I recompiled it with the lastest commits (including pruning stuff ) I am using it on the testnet and it seems i am running of a different blockchain fork than blockexplorer.info since block 33596
26 2012-10-23 00:46:44 <gmaxwell> SomeoneWeird: 13mbytes
27 2012-10-23 00:46:50 <BlueMatt> xblitz: wat does your fork change
28 2012-10-23 00:47:02 <xblitz> but I mined a couple of blocks.. and other miners too .. so im not alone on it
29 2012-10-23 00:47:12 <gmaxwell> xblitz: you're not running the latest ultra prune code. I intentionally broke the old stuff there.
30 2012-10-23 00:47:22 <SomeoneWeird> oh, small as again
31 2012-10-23 00:47:23 <SomeoneWeird> nice
32 2012-10-23 00:47:24 <Diablo-D3> yay gmaxwell
33 2012-10-23 00:47:26 <xblitz> nothing special.. juste options to display more stuff on the rpc...
34 2012-10-23 00:47:43 <xblitz> gmaxwell : oh!
35 2012-10-23 00:47:49 <BlueMatt> oh, yea, there was a bug in ultraprune, fixed today, that (accidentally) changed block acceptance rules
36 2012-10-23 00:47:51 <gmaxwell> xblitz: and thanks for running testnet!
37 2012-10-23 00:47:58 <xblitz> :)
38 2012-10-23 00:48:05 <gmaxwell> (and git master!)
39 2012-10-23 00:48:19 <gmaxwell> and actually paying attention. I'm shocked.
40 2012-10-23 00:48:22 <gmaxwell> :P
41 2012-10-23 00:48:29 <gmaxwell> There is hope for the future.
42 2012-10-23 00:48:35 <xblitz> oh yeah.. i see the new commits now ;)
43 2012-10-23 00:48:43 <BlueMatt> gmaxwell: agreed
44 2012-10-23 00:48:46 <gmaxwell> _and_ we even got your report in less than 24 hours. :P
45 2012-10-23 00:48:55 <BlueMatt> its quite amazing
46 2012-10-23 00:48:59 <xblitz> haha
47 2012-10-23 00:49:08 <gmaxwell> 24 minutes would have been better, but hey it's a start.
48 2012-10-23 00:49:17 <BlueMatt> xblitz: its surprisingly rare
49 2012-10-23 00:49:44 <gmaxwell> We need to discover fork creating bugs on mainnet within probably ~20 minutes to avert a mess.
50 2012-10-23 00:50:15 <xblitz> So i guess were a couple running on this "bad" fork were at block 33613 while the "real" fork is at 33630
51 2012-10-23 00:50:16 <BlueMatt> true, but, again, getting reports of fork-creating bugs in the first place is nice...
52 2012-10-23 00:50:35 <gmaxwell> (I have mainnnet nodes here that will page me if they get blocks they reject, but thats not really enough.)
53 2012-10-23 00:50:41 <xblitz> which split at 33596
54 2012-10-23 00:50:48 <gmaxwell> xblitz: yep.
55 2012-10-23 00:51:14 <BlueMatt> gmaxwell: Id like to see warning lights all over the place if any node gets far off...
56 2012-10-23 00:51:15 <gmaxwell> xblitz: do you know what transaction caused it?
57 2012-10-23 00:51:22 <xblitz> ok.. so pulling in bitcoin:master will put it back to normal? will i have to do a rescan?
58 2012-10-23 00:51:35 <xblitz> I m really not sure.. could even be one of mine
59 2012-10-23 00:51:41 <gmaxwell> BlueMatt: well, considering how high difficulty is??? a hearing single invalid block is probably worth waking people up over.
60 2012-10-23 00:52:01 <BlueMatt> gmaxwell: ack
61 2012-10-23 00:52:10 <gmaxwell> (cause I can't expect someone burning a few hundred dollars just to wake some developers up)
62 2012-10-23 00:52:19 <BlueMatt> speaking of which...I really need to set up a bitcoinj node to do the same
63 2012-10-23 00:52:29 <gmaxwell> xblitz: no, it's mine. You shold look in your debug.log
64 2012-10-23 00:52:40 <BlueMatt> gmaxwell: (and if they do, Ill wake up to congratulate them...)
65 2012-10-23 00:52:49 <gmaxwell> BlueMatt: hahah
66 2012-10-23 00:53:16 <xblitz> gmaxwell: humm... debug.log is full of stuff haha.. you really want me to dig in it?
67 2012-10-23 00:53:33 <xblitz> gmaxwell: if it would help though Im up for it
68 2012-10-23 00:53:37 <gmaxwell> xblitz: No well, I was just suggesting you do for your own education and gratification. I know what it is.
69 2012-10-23 00:53:47 <xblitz> ok ok
70 2012-10-23 00:53:57 <gmaxwell> xblitz: Whenever there is a rejected block the node will log a reason why. So if you ever see that again you can check the log for it.
71 2012-10-23 00:54:12 <xblitz> yeah ok ill go look then!
72 2012-10-23 00:55:22 <gmaxwell> https://people.xiph.org/~greg/bitcoin_coverage/coverage.ultraprune.testbitcoin-testnet/ < coverage now with testnet (note that the covered percentages are high in these reports because it's counting source files which weren't run at all)
73 2012-10-23 00:56:04 <xblitz> debug.log should be timestamped
74 2012-10-23 00:56:08 <xblitz> :\\
75 2012-10-23 00:56:16 <Luke-Jr> xblitz: it is, if you use -logtimestamps!
76 2012-10-23 00:56:23 <xblitz> doh!
77 2012-10-23 00:56:40 <xblitz> rephrase: debug.log should be timestamped by default! haha
78 2012-10-23 00:56:58 <BlueMatt> its a (theoretical) privacy issue if it is
79 2012-10-23 00:57:33 <xblitz> really that bad?
80 2012-10-23 00:58:18 <gmaxwell> xblitz: meh. It's not good. It would make it valuable to steal nodes' logs to do timing analysis for transactions.
81 2012-10-23 00:58:37 <Luke-Jr> xblitz: it is, if some evil organization wants to hack a lot of PCs to mine info
82 2012-10-23 00:58:57 <gmaxwell> Not likely, but it's better for people if the motivation isn't there.
83 2012-10-23 00:59:07 <xblitz> ok
84 2012-10-23 01:00:02 <BlueMatt> gmaxwell: can you run it again with just test_bitcoin+the bitcoinj tool (ie what is auto-run by jenkins/pull-tester)
85 2012-10-23 01:01:08 <gmaxwell> I'm happy to see CTxOutCompressor::CompressAmount with 100% branch coverage.
86 2012-10-23 01:03:19 <gmaxwell> BlueMatt: I will. I just burned so much time getting lcov working that I haven't had a chance yet.
87 2012-10-23 01:03:37 <BlueMatt> ah, ok
88 2012-10-23 01:03:58 <BlueMatt> tanks
89 2012-10-23 01:04:54 <xblitz> gmaxwell: received block 00000000dc3a9cee3f1b \\ ERROR: CheckInputs() : tried to spend coinbase at depth 99
90 2012-10-23 01:05:04 <xblitz> you sneaky bastard! :P
91 2012-10-23 01:05:10 <BlueMatt> thats te one
92 2012-10-23 01:05:19 <xblitz> trying to spend illigal stuff
93 2012-10-23 01:05:23 <xblitz> illegal
94 2012-10-23 01:05:24 <gmaxwell> xblitz: well, it lies! I'm spending at 100. That was the bug, that test was off by one.
95 2012-10-23 01:05:33 <gmaxwell> And for the record, BlueMatt is the sneaky one.
96 2012-10-23 01:05:43 <BlueMatt> shhh
97 2012-10-23 01:05:47 <gmaxwell> he wrote a tool that makes fake blocks that test all these boundary conditions.
98 2012-10-23 01:05:51 <BlueMatt> I didnt mine a block to break it
99 2012-10-23 01:06:13 <gmaxwell> I'm still astonished that there were none at 100 in testnet or mainnet.
100 2012-10-23 01:06:30 <BlueMatt> as you pointed out, the code only allowed easy access to 101, so...
101 2012-10-23 01:07:06 <gmaxwell> _our code_.
102 2012-10-23 01:07:23 <xblitz> so if I rebase to master Will i jump automatically to the good fork?
103 2012-10-23 01:07:50 <xblitz> or i need a rescan
104 2012-10-23 01:09:52 <BlueMatt> xblitz: it should (in theory)
105 2012-10-23 01:09:56 <BlueMatt> xblitz: please report if it doesnt
106 2012-10-23 01:10:02 <xblitz> compiling now
107 2012-10-23 01:11:35 <BlueMatt> gmaxwell: (and if you do get around to writing in more tests, adding the tests noted at http://pastebin.com/cfXyYqCa (thanks to Matt_von_Mises for the list))
108 2012-10-23 01:11:46 <BlueMatt> would be much appreciated ;)
109 2012-10-23 01:14:04 <gmaxwell> heh. I gave him a couple of those. :P
110 2012-10-23 01:21:19 <xblitz> cant seem to actualy pull upstream:master to my fork:master ... but i dont have any modifications to my fork:master
111 2012-10-23 01:21:31 <xblitz> it talks about conflict on multiples files
112 2012-10-23 01:22:09 <xblitz> any ideas?
113 2012-10-23 01:22:53 <xblitz> i just want to put back myfork:master = bitcoin:master
114 2012-10-23 01:22:54 <deedeedeedee> oh god atlas is going on a rant again against development
115 2012-10-23 01:24:19 <BlueMatt> gmaxwell: I know Ive got most of them, but there are a few still missing
116 2012-10-23 01:25:19 <gmaxwell> deedeedeedee: again today or from two days ago?
117 2012-10-23 01:26:53 <gmaxwell> Ah I see again.
118 2012-10-23 01:27:44 <gmaxwell> I think the development community should reject the forums if the forums won't reject persistant irrational assholes like that.
119 2012-10-23 01:27:59 <xblitz> ok.. i fixed my stuff.. back to bitcoin:master :)
120 2012-10-23 01:28:09 <BlueMatt> gmaxwell: I thought the development community rejected the forums ~a year and a half ago...
121 2012-10-23 01:28:25 <BlueMatt> maybe Im just a year ahead of the development community :P
122 2012-10-23 01:28:25 <Luke-Jr> BlueMatt++
123 2012-10-23 01:28:40 <Luke-Jr> I pretty much only look at the forums when someone posts a link
124 2012-10-23 01:28:46 <BlueMatt> same
125 2012-10-23 01:29:18 <gmaxwell> BlueMatt: we still make release announcements and discuss technical things there. (well perhaps not you, but it's not effective if only a few boycott it)
126 2012-10-23 01:30:27 <gmaxwell> seriously, in the US it I'm pretty sure would be a violation of employment law for an employer to subject employees to a customer that acted like that guy. He isn't just insane, ignorant, and rude. He's soul sucking.
127 2012-10-23 01:31:02 <BlueMatt> gmaxwell: I know, it was a joke...still, I generally agree that spending time discussing things on the forum is not generally a good time-use...Id kinda like to see the forums as "its been discussed on #bitcoin-dev/ml and most devs are in general agreement that we should... what does the community think?"
128 2012-10-23 01:31:44 <gmaxwell> ::nods::
129 2012-10-23 01:31:46 <BlueMatt> then you get fewer threads devs have to read and can ignore/highly moderate those few threads
130 2012-10-23 01:33:13 <BlueMatt> anywhoo early class and behind on sleep...see yall tomorrow
131 2012-10-23 01:34:56 <Luke-Jr> ttyl
132 2012-10-23 01:35:55 <MC1984> isnt there some sort of extraordinary blackballing procedure tht can be used against atlas
133 2012-10-23 01:36:50 <gmaxwell> MC1984: he was banned and the forum improved considerably while that was the case.
134 2012-10-23 01:37:16 <gmaxwell> I think theymos is actually performing a crossover expirement on the broken window theorem as applied to online communities.
135 2012-10-23 01:37:25 <MC1984> the what now
136 2012-10-23 01:37:42 <gmaxwell> http://en.wikipedia.org/wiki/Crossover_study
137 2012-10-23 01:38:38 <Luke-Jr> in the meantime, Atlas controls the Wikipedia page for Bitcoin and the mods there of course don't care; so bitcointalk = bitcoin experts and bitcoin = silk road currency
138 2012-10-23 01:40:05 <zveda> really atlas controls wiki ?
139 2012-10-23 01:40:07 <xblitz> oh my.. the current testnet block index is my bank account number!!
140 2012-10-23 01:40:16 <MC1984> just ban him again
141 2012-10-23 01:40:17 <xblitz> wonder if it means something
142 2012-10-23 01:40:34 <MC1984> busting dev balls is too important a job for wingnut like that
143 2012-10-23 01:40:38 <zveda> as in en.bitcoin.it/wiki/Main_Page ? or wikipedia.org ?
144 2012-10-23 01:40:41 <Luke-Jr> zveda: yeah, look at the edit history
145 2012-10-23 01:40:44 <Luke-Jr> en.wikipedia.org
146 2012-10-23 01:40:49 <zveda> ic
147 2012-10-23 01:41:07 <zveda> but cant we out-edit him ? or he's editing maniacally ?
148 2012-10-23 01:42:05 <gmaxwell> zveda: he'll get himselve banned eventually but it requires several thoughtful and patient people do just keep fixing it while he sets himself on fire.
149 2012-10-23 01:42:49 <zveda> strange.. how can 1 crazy but determined person cause so much trouble lol
150 2012-10-23 01:42:51 <MC1984> howardstrong?
151 2012-10-23 01:43:08 <gmaxwell> I might try; (I edited the article extensively a year and change ago)??? but if I touch it now he'll be all over me alleging conflict of interest and I apparently have about zero tolerance for his nonsense.
152 2012-10-23 01:43:12 <gmaxwell> MC1984: yup.
153 2012-10-23 01:43:17 <gmaxwell> thats one of his many identities.
154 2012-10-23 01:43:24 <Luke-Jr> gmaxwell: he'll edit war and try to get you banned
155 2012-10-23 01:43:41 <gmaxwell> well he can't get me banned.
156 2012-10-23 01:43:53 <MC1984> this is the problem with wikipedia
157 2012-10-23 01:44:24 <gmaxwell> MC1984: nah, I don't really think wikipedia is doing anything horribly wrong. It's an open thing, sometimes nutbags screw it up. Thats just the breaks.
158 2012-10-23 01:44:43 <MC1984> yes nutbags are the problem with wikipedia
159 2012-10-23 01:45:05 <MC1984> people who own certain articles as far as they are concerned
160 2012-10-23 01:45:42 <MC1984> as i said one of the only time i contribute was some minor spelling corrections and it got reverted
161 2012-10-23 01:46:31 <zveda> i was annoyed when somebody from pharmaceutical company pwned the wiki page for the buteyko method
162 2012-10-23 01:46:39 <deedeedeedee> what did atlas do to the article?
163 2012-10-23 01:46:40 <zveda> i tried editing it but was no match
164 2012-10-23 01:46:49 <gmaxwell> of the past 500 edits he's made 337.
165 2012-10-23 01:47:01 <deedeedeedee> 0_o
166 2012-10-23 01:47:11 <gmaxwell> deedeedeedee: generally crapped on it in many ways.
167 2012-10-23 01:47:13 <SomeoneWeird> wow
168 2012-10-23 01:48:13 <MC1984> i noticed its basically long winded bullshit which is a shame
169 2012-10-23 01:48:19 <gmaxwell> Here is how it was when I last touched it:
170 2012-10-23 01:48:20 <gmaxwell> http://en.wikipedia.org/w/index.php?title=Bitcoin&oldid=433926110
171 2012-10-23 01:48:25 <gmaxwell> which still kinda sucked
172 2012-10-23 01:48:31 <gmaxwell> but I think it was a lot more actually informative.
173 2012-10-23 01:48:36 <MC1984> cos wiki is the first place people go when they have a wtf is that moment
174 2012-10-23 01:48:42 <gmaxwell> The current version doesn't actually tell you anything about how bitcoin _works_ anymore.
175 2012-10-23 01:49:18 <gmaxwell> Which is kinda important, since the distributed consensus algorithim in bitcoin solves a hard problem which otherwise would make it impossible for bitcoin to be secure.
176 2012-10-23 01:50:07 <Luke-Jr> gmaxwell: if it said too much of how it worked, it would expose Atlas's ignorance ;)
177 2012-10-23 01:51:02 <deedeedeedee> bitcoin mining has some stuff
178 2012-10-23 01:51:03 <gmaxwell> It's also full of inaccuracies. It says, for example, that the proof of work prevents double spending (this is the only mention of that whole issue in the current article, just that statement), which isn't true.
179 2012-10-23 01:51:21 <gmaxwell> deedeedeedee: yea, it talks about mining in irrelevant detail because it's something atlas understands.
180 2012-10-23 01:51:49 <gmaxwell> Who cares about how you increment the nonce? It should link to [[Hashcash]] which explains how to use a hash to create a message bound proof of work
181 2012-10-23 01:52:04 <gmaxwell> or even [[Proof-of-work system]]
182 2012-10-23 01:52:06 <Hasimir> tried contacting a wikipedia admin to get it locked down and stop Atlas
183 2012-10-23 01:52:08 <Hasimir> ?
184 2012-10-23 01:52:46 <deedeedeedee> hes readin this :(
185 2012-10-23 01:52:53 <gmaxwell> of course he is.
186 2012-10-23 01:53:03 <gmaxwell> He's probably got three or four socks in the channel.
187 2012-10-23 01:53:14 <SomeoneWeird> ofc
188 2012-10-23 01:53:45 <gmaxwell> Hasimir: Thats just not how wikipedia works. His misbehavior will get him banned eventually, but the process is slow??? and slower when reasonable men sit quietly while he craps all over it.
189 2012-10-23 01:53:46 <Hasimir> who cares? I'd say the same thing in front of him.
190 2012-10-23 01:53:52 <gmaxwell> Hasimir: Exactly.
191 2012-10-23 01:54:24 <Hasimir> might use a cattle prod, though, you never know how the paranoids are going to react. ;)
192 2012-10-23 01:54:45 <Luke-Jr> gmaxwell: well, despite giving up eventually, I think I can say I tried a little XD
193 2012-10-23 01:55:38 <zveda> this part relaly doesnt sound fair: https://en.wikipedia.org/wiki/Bitcoin#As_an_investment
194 2012-10-23 01:57:30 <jgarzik> orphan mapsz 194, mempool poolsz 1012
195 2012-10-23 01:57:33 <jgarzik> fixes looking better
196 2012-10-23 01:57:59 <deedeedeedee> he just edited it again
197 2012-10-23 02:04:04 <Hasimir> gmaxwell, maybe just create a bot to revert changes from known troll accounts, like Atlas and all his sock puppets when they're identified
198 2012-10-23 02:04:31 <Hasimir> there are plenty of bots on wikipedia already
199 2012-10-23 02:04:43 <pangaearian> Hi. Can someone help me with maybe a newbie question... Has the bitcoind daemon a memory leak and therefore has to be restarted regularly (eg. daily) ?
200 2012-10-23 02:05:58 <gmaxwell> pangaearian: Not that we know of. We'd consider that a fairly serious issue. How are you observing this?
201 2012-10-23 02:06:30 <pangaearian> On a Debian ARM box
202 2012-10-23 02:07:03 <gmaxwell> Hasimir: Wikipedia doesn't know that atlas is mentally ill and unable to cooperate with others yet. Once it realizes this it will agressively and effectively ban him, and the article can be healed.
203 2012-10-23 02:07:04 <Arnavion> pangaearian That thing you linked in #bitcoin is not a memory leak
204 2012-10-23 02:07:05 <pangaearian> Bitcoind is drawing constantly more memory
205 2012-10-23 02:07:20 <Arnavion> It's Linux's file cache
206 2012-10-23 02:07:42 <gmaxwell> Arnavion: thanks, I'd missed the #bitcoin traffic.
207 2012-10-23 02:07:42 <Hasimir> gmaxwell, good point, but I don't think he'll be the last
208 2012-10-23 02:08:35 <gmaxwell> Hasimir: nah, most people aren't as much of a problem as him. Regular editing can take care of the rest. E.g. someone that goes and adds some silly stuff can just be quickly undone, and existing bots will do some of that.
209 2012-10-23 02:08:35 <pangaearian> hm ok. You're right. My fault
210 2012-10-23 02:08:36 <Hasimir> gmaxwell, there's always more crazy
211 2012-10-23 02:08:54 <Hasimir> true
212 2012-10-23 02:09:01 <trudel_> hey everyone
213 2012-10-23 02:09:11 <Hasimir> don't spam the channel again
214 2012-10-23 02:09:33 <trudel_> Sorry for anyone who has already heard this, but I'm writing a handout to promote bitcoins and am looking for people who can review it for errors
215 2012-10-23 02:09:34 <trudel_> http://piratepad.net/x1h9j4w2Ed
216 2012-10-23 02:09:39 <trudel_> there it is^
217 2012-10-23 02:09:41 <Hasimir> ACTION sighs
218 2012-10-23 02:09:48 <deedeedeedee> whats atlases goal?
219 2012-10-23 02:10:27 <trudel_> Is there anyone here who has read the source code?
220 2012-10-23 02:10:31 <Hasimir> deedeedeedee, Atlas? <shrug> ;)
221 2012-10-23 02:10:34 <gmaxwell> deedeedeedee: he's mentally ill (Not me being rude, he's posted about it) and it makes him obess over stuff.
222 2012-10-23 02:10:46 <gmaxwell> er obsess*
223 2012-10-23 02:11:12 <trudel_> hey is there a good forum or something to find people who might be willing to help me make a handout?
224 2012-10-23 02:11:17 <Luke-Jr> trudel_: ??? isn't the right symbol :P
225 2012-10-23 02:11:28 <Luke-Jr> trudel_: #bitcoin would be more appropriate IMO
226 2012-10-23 02:11:31 <trudel_> Well it's the closest there is in the ascii char set
227 2012-10-23 02:11:42 <Luke-Jr> trudel_: ??? isn't in the ASCII char set either
228 2012-10-23 02:11:47 <Hasimir> ACTION gets popcorn
229 2012-10-23 02:11:54 <trudel_> Well whatever
230 2012-10-23 02:12:22 <trudel_> I welcome the suggestion though and am interested in more
231 2012-10-23 02:12:24 <Hasimir> for a printed doc you can use other fonts to get the right symbol
232 2012-10-23 02:12:28 <gmaxwell> ??? is a bit unfortunate because its an established currency symbol.
233 2012-10-23 02:12:34 <trudel_> For what?
234 2012-10-23 02:12:38 <Hasimir> baht
235 2012-10-23 02:12:38 <Luke-Jr> Thai Baht
236 2012-10-23 02:12:44 <trudel_> oh interesting
237 2012-10-23 02:12:54 <trudel_> It should have two lines through it for bitcoin?
238 2012-10-23 02:13:01 <Hasimir> yep
239 2012-10-23 02:13:06 <Luke-Jr> "The bitcoins system is the collection of all protocols, software and computers, that work together over the internet to perform the useful tasks that a bank or government in the incumbent currency systems perform." <-- this sounds wrong, but I don't know how to rephrase it
240 2012-10-23 02:13:12 <trudel_> I thoguht as much
241 2012-10-23 02:13:24 <trudel_> Well how would you answer the question luke?
242 2012-10-23 02:13:28 <Luke-Jr> trudel_: I fixed it for you, but you might need to check fonts to print it properly
243 2012-10-23 02:13:29 <Hasimir> I posted an alternative yesterday
244 2012-10-23 02:13:46 <gmaxwell> Luke-Jr: when in doubt go back to the original post for inspiration: http://p2pfoundation.ning.com/forum/topics/bitcoin-open-source
245 2012-10-23 02:13:50 <Luke-Jr> trudel_: I would answer too technically for average people
246 2012-10-23 02:14:32 <gmaxwell> It's always best to describe bitcoin in those terms because (1) it's very well written, and (2) it's the official word on what bitcoin was supposted to accomplish.
247 2012-10-23 02:14:54 <Luke-Jr> this webpage keeps losing its connection
248 2012-10-23 02:14:54 <trudel_> I've never been much one for officialdom
249 2012-10-23 02:15:06 <Hasimir> Oct 22 14:18:27 <Hasimir>\ttrudel, here: Bitcoin is a software protocol based on cryptographic principles which provides transparent, pseudonymous and decentralised transactions. The name is also used to refer to the unit of currency and the main software client.
250 2012-10-23 02:15:35 <Hasimir> that was my alternate second para
251 2012-10-23 02:15:43 <Luke-Jr> the main software client shouldn't be called Bitcoin though
252 2012-10-23 02:15:48 <gmaxwell> "The bitcoin system is a collection of protocols, software, and infrastructure which provide an alterative money system designed to eliminate the trust required by traditional banking and currency systems."
253 2012-10-23 02:15:56 <trudel_> yeah the first one was called bitcoind
254 2012-10-23 02:16:30 <trudel_> Yeah we now have 4 meanings of the term bitcoin, client, system, currency and the "coin" which is a chain of crypto signatures
255 2012-10-23 02:16:35 <gmaxwell> trudel_: thats not correct.
256 2012-10-23 02:16:36 <trudel_> oh well
257 2012-10-23 02:16:45 <trudel_> well at least 4
258 2012-10-23 02:16:49 <gmaxwell> trudel_: you're repeating one of wikipedia's _many_ factually incorrect claims.
259 2012-10-23 02:16:58 <trudel_> which?
260 2012-10-23 02:17:03 <gmaxwell> No no, I'm saying "< trudel_> yeah the first one was called bitcoind" is incorrect.
261 2012-10-23 02:17:13 <Luke-Jr> I bet there's more than 4 now
262 2012-10-23 02:17:18 <trudel_> I read that in a pdf from satoshi, not wikipedia
263 2012-10-23 02:17:24 <gmaxwell> trudel_: you didn't.
264 2012-10-23 02:17:25 <trudel_> the last one
265 2012-10-23 02:17:41 <trudel_> Which I assume is what you're taking issuw with
266 2012-10-23 02:17:50 <Luke-Jr> the first one was called Bitcoin, but that one doesn't exist really anymore (though the core code is used in bitcoind and Bitcoin-Qt)
267 2012-10-23 02:17:57 <gmaxwell> Bitcoin was originally a windows gui app only, bitcoind was largely gavin's work I believe, but in any case it came much later.
268 2012-10-23 02:18:55 <trudel_> When a transaction is submitted, the whole value of an address is decanted out, divided into the outputs of the transaction, rather than merely being deducted, right? Thus it is conceptualized as "spending" a whole coin, which is then divided by the system... isn't it?
269 2012-10-23 02:19:24 <trudel_> anyway
270 2012-10-23 02:19:37 <trudel_> Yes the wikipedia article is a big problem though
271 2012-10-23 02:19:51 <trudel_> That is really the thing that is most important to focus on, not a handout
272 2012-10-23 02:20:11 <Hasimir> heh
273 2012-10-23 02:20:13 <trudel_> The bitcoin foundation should put on it's site an expert-reviewed version of ti.
274 2012-10-23 02:20:25 <Luke-Jr> trudel_: for a handout, it might be neat to put some small coin on it via a private key
275 2012-10-23 02:20:29 <trudel_> what hasimir?
276 2012-10-23 02:20:32 <trudel_> Yes I did that
277 2012-10-23 02:20:35 <trudel_> 0.02 coin
278 2012-10-23 02:20:44 <trudel_> not much just enough to test things with
279 2012-10-23 02:20:53 <trudel_> I used instawallet.com
280 2012-10-23 02:20:58 <Hasimir> trudel_, we were discussing the wiki page just before you entered the channel
281 2012-10-23 02:21:02 <Luke-Jr> MtGox has a handy import function for private keys
282 2012-10-23 02:21:06 <Luke-Jr> hopefully Bitcoin-Qt will someday too <.<
283 2012-10-23 02:21:10 <trudel_> oh well it's an important topic as I said
284 2012-10-23 02:21:25 <Hasimir> trudel_, you missed the criticisms of that page literally by seconds
285 2012-10-23 02:21:26 <deedeedeedee> trudel_: what do you want to change about wiki?
286 2012-10-23 02:21:30 <trudel_> so what is being done about the wikipedia article then?
287 2012-10-23 02:21:52 <trudel_> Well there should be a page on the bitcoin.org which is like the wikipedia page, but expert reviewed
288 2012-10-23 02:22:17 <trudel_> That would be the bees knees. Handouts could be easily compiled from it.
289 2012-10-23 02:22:36 <trudel_> And curious laymen could easily understand it.
290 2012-10-23 02:23:49 <Hasimir> well, there is the bitcoin wiki
291 2012-10-23 02:24:18 <Luke-Jr> http://luke.dashjr.org/tmp/code/blankTBC.pdf <-- my handouts :P
292 2012-10-23 02:24:38 <trudel_> The wiki is highlyinferior to an encyclopedia entry for the purpose though
293 2012-10-23 02:24:47 <trudel_> will check that out luke
294 2012-10-23 02:25:14 <Luke-Jr> I'm promoting Tonal just as much as Bitcoin with those of course ;)
295 2012-10-23 02:25:35 <trudel_> Hm, like mt gox is hard to get an account on I hear
296 2012-10-23 02:25:40 <trudel_> You must provide id and so forth
297 2012-10-23 02:25:54 <trudel_> that will be too much bother for new users
298 2012-10-23 02:26:10 <trudel_> due to money laundering legislation.
299 2012-10-23 02:26:49 <deedeedeedee> is bitcoin really a currency?
300 2012-10-23 02:27:00 <trudel_> Blockchain.info supposedly has some way to import a new wallet, but not to merge wallets. The ability to merge secret keys is included in the most recent bitcoin client apparently, but getting it onto a page and back is still hard.
301 2012-10-23 02:27:03 <trudel_> deedee, yes.
302 2012-10-23 02:27:07 <trudel_> I would say so.
303 2012-10-23 02:27:16 <trudel_> Why not?
304 2012-10-23 02:27:29 <deedeedeedee> its weird to me
305 2012-10-23 02:27:39 <trudel_> rolleyes, this is bitcoin-dev
306 2012-10-23 02:27:55 <trudel_> no offense
307 2012-10-23 02:27:59 <trudel_> anyway
308 2012-10-23 02:28:22 <trudel_> so about the wikipedia article, are there any devs that are willing to expert review it?
309 2012-10-23 02:28:38 <trudel_> If we could get the document made we can certainly get it up somewhere.
310 2012-10-23 02:29:06 <Luke-Jr> trudel_: importing private keys is a bad idea generally
311 2012-10-23 02:29:10 <trudel_> Bitcoin.org people will notice and either copy it directly to their site, or it will become well indexed by google, or
312 2012-10-23 02:29:12 <deedeedeedee> its biggest problem is a lack of info
313 2012-10-23 02:29:30 <trudel_> just highly useful in bringing the equilibrium of the wikipedia article towards a more accurate state.
314 2012-10-23 02:29:34 <trudel_> Through referencing.
315 2012-10-23 02:30:13 <trudel_> Yeah, Luke, instawallet works arguably better for various reasons.
316 2012-10-23 02:30:27 <trudel_> however about the encyclopedic article.
317 2012-10-23 02:30:57 <trudel_> I see ther eis a long list of people in this channel.
318 2012-10-23 02:31:12 <trudel_> Are they contactable in some way?
319 2012-10-23 02:31:19 <trudel_> Why are they even here?
320 2012-10-23 02:32:04 <trudel_> also can anyone recommend a forum to find people with authoritative knowledge on bitcoin?
321 2012-10-23 02:32:09 <Luke-Jr> trudel_: new to IRC? :p
322 2012-10-23 02:32:17 <deedeedeedee> authoritative?
323 2012-10-23 02:32:28 <trudel_> well like read the source code and so forth
324 2012-10-23 02:32:29 <Luke-Jr> asking questions here is probably the closest you'll get to that
325 2012-10-23 02:32:46 <trudel_> It doesn't seem to be working too hot, is there a better time of day perhaps?
326 2012-10-23 02:33:15 <Luke-Jr> there's also a development mailing list
327 2012-10-23 02:33:22 <Luke-Jr> a few hours ago was more active
328 2012-10-23 02:33:23 <trudel_> where?
329 2012-10-23 02:33:26 <Luke-Jr> maybe check out the channel logs (see topic)
330 2012-10-23 02:33:27 <trudel_> ok
331 2012-10-23 02:33:35 <Luke-Jr> (for times the channel is most busy)
332 2012-10-23 02:33:48 <trudel_> good idea
333 2012-10-23 02:35:10 <gmaxwell> 21:28 < trudel_> so about the wikipedia article, are there any devs that are willing to expert review it?
334 2012-10-23 02:35:33 <gmaxwell> I've already reviewed it, it is wrong in many places, incomplete almost everywhere else, and shouldn't be used for pretty much anything.
335 2012-10-23 02:35:53 <gmaxwell> Go look at a 9 month old version, which will be much more accurate and complete.
336 2012-10-23 02:35:59 <trudel_> Well it still contains a template for the experts to fill in.
337 2012-10-23 02:36:20 <trudel_> hm, ok, but I am really looking for something more authoritative
338 2012-10-23 02:37:32 <trudel_> In any case, the format and so forth is a very good one to make it easy for laymen to gain enough confidence to use the system. confidence requires knowledge of the system, and right now
339 2012-10-23 02:37:45 <trudel_> they are not able to get it in a atimely manner, so they give up.
340 2012-10-23 02:38:03 <trudel_> I mean try thinking of how you would persuade your local convenience store owner to accept bitcoins.
341 2012-10-23 02:38:17 <trudel_> Not doable with the current state of documentation.
342 2012-10-23 02:38:47 <trudel_> We also should have a set of guidelines on how many confirmations to wait for, as that is a real sticking point that scares a lot of people away.
343 2012-10-23 02:38:56 <trudel_> They think they *have* to wait for at least one.
344 2012-10-23 02:39:32 <gmaxwell> trudel_: the reference software embodies the recommendations I'd give.
345 2012-10-23 02:39:35 <trudel_> Even though there have never been any reported sucessful double spends even with unconfirmed payments, have there?
346 2012-10-23 02:39:42 <gmaxwell> sure there have been.
347 2012-10-23 02:39:44 <trudel_> What?
348 2012-10-23 02:39:59 <trudel_> Where gmaxwell?
349 2012-10-23 02:40:16 <gmaxwell> Mybitcoin for example, was doublespendable through their shopping cart interface which accepted unconfirmed transactions.
350 2012-10-23 02:40:30 <trudel_> and it was sucessfully exploited?
351 2012-10-23 02:40:34 <gmaxwell> And its utterly trivial to rip people off if they accept unconfirmed transactions.
352 2012-10-23 02:40:49 <trudel_> Well timing needs to be fairly good
353 2012-10-23 02:40:59 <gmaxwell> trudel_: yes, according to them. I mean, there was only like a million dollars in bitcoin taken but hey, no need to wait.
354 2012-10-23 02:41:03 <trudel_> and secondly it is harder if the victim is well connected to thenetwork
355 2012-10-23 02:41:04 <gmaxwell> trudel_: no, not really.
356 2012-10-23 02:41:08 <gmaxwell> No, thats not true.
357 2012-10-23 02:41:20 <trudel_> How good does timing have to be then?
358 2012-10-23 02:41:38 <gmaxwell> I can trivially construct a transaction that won't confirm for months.
359 2012-10-23 02:42:36 <gmaxwell> And as a result there is no great timing required for it.
360 2012-10-23 02:42:45 <trudel_> oh really
361 2012-10-23 02:42:49 <trudel_> how?
362 2012-10-23 02:43:04 <gmaxwell> By setting the locktime on the transaction in the future.
363 2012-10-23 02:43:18 <trudel_> Well can a client detect yo uar edoing that?
364 2012-10-23 02:43:47 <gmaxwell> Sure but nothing does as far as I know.
365 2012-10-23 02:43:58 <trudel_> Hopefully that hole could be plugged by alerting the user with the client in the futre at least.
366 2012-10-23 02:44:19 <gmaxwell> And connectivity doesn't help for three reasons??? no software actually exists that detects and does something useful with the existance of conflicting transactions (which happen innocently too); because conflicting transactions aren't relayed so your own transaction will likely block you from hearing it; and because there is no need for an attacker to announce the conflict in any case.
367 2012-10-23 02:45:05 <gmaxwell> And because of the last point, it's possible for me try to rob you with every single transaction and have you not know. So even if I'm only successful 10% of the time I can still rob you blind if you're accepting automatically.
368 2012-10-23 02:45:17 <trudel_> well there are some attack approaches that do not require announcement
369 2012-10-23 02:45:26 <gmaxwell> There is a reason that every serious bitcoin service waits several confrmations.
370 2012-10-23 02:45:38 <gmaxwell> trudel_: no attack requires announcement of the conflict.
371 2012-10-23 02:45:39 <trudel_> Not all do.
372 2012-10-23 02:46:02 <trudel_> Well not of the conflict, but of the conflicting transaction, yes.
373 2012-10-23 02:46:16 <gmaxwell> trudel_: the conflicting transaction does not need to be announced.
374 2012-10-23 02:46:20 <trudel_> Which if propagated properly would be an effective alert to the user
375 2012-10-23 02:46:33 <trudel_> For the race attack it does.
376 2012-10-23 02:46:40 <trudel_> And that is by far the most important attack.
377 2012-10-23 02:46:47 <gmaxwell> trudel_: If you don't have any recourse and you accept at zero confirms you are vulnerable to theft, it's as simple as that. If there is some natural ratelimit or recourse then that might be acceptable.
378 2012-10-23 02:46:53 <trudel_> As it is by far the cheapest and easiest
379 2012-10-23 02:47:22 <gmaxwell> trudel_: I think you should not be presenting on bitcoin to anyone. Your understanding of the system is weak, and you're just arguing with me when I'm trying to educate you some.
380 2012-10-23 02:47:22 <trudel_> Yes but the question is how vulnerable.
381 2012-10-23 02:47:28 <trudel_> That is the important question here.
382 2012-10-23 02:48:00 <trudel_> Rolleyes
383 2012-10-23 02:48:24 <trudel_> Do you know what a race attack is?
384 2012-10-23 02:48:32 <gmaxwell> ...
385 2012-10-23 02:49:06 <trudel_> Look are you just after a little ego boost or are you interested in helping here
386 2012-10-23 02:49:28 <gmaxwell> trudel_: I was trying to help you, but you don't want to be helped. You want to argue. I don't want to argue.
387 2012-10-23 02:49:35 <trudel_> rolleyes
388 2012-10-23 02:50:36 <gmaxwell> The system simply doesn't work the way you think it does. Someone can mine a transaction without ever announcing it to anyone else.
389 2012-10-23 02:50:38 <trudel_> Don't pretend you know more than you do man, we all want to encourage bitcoin use here, and you'll end up throwing a wrench in the works if you're not careful.
390 2012-10-23 02:50:51 <gmaxwell> trudel_: I'm not pretending anything. :(
391 2012-10-23 02:51:08 <trudel_> Look I gotta go, thanks for what little help I did get here, good night everyone, no hard feelings gmaxwell.
392 2012-10-23 02:51:18 <gmaxwell> but ... I have hard feelings.
393 2012-10-23 02:51:29 <gmaxwell> :P
394 2012-10-23 02:52:39 <deedeedeedee> that guy sounds like atlas
395 2012-10-23 02:53:13 <gmaxwell> He was too polite.
396 2012-10-23 02:53:19 <gmaxwell> I feel bad for being a jerk there.
397 2012-10-23 02:56:58 <gmaxwell> ACTION needs a vacation
398 2012-10-23 03:31:09 <DrHaribo> gmaxwell: 1tx blocks are either from long poll (like I believe Eligius does), or because people are using old buggy miners that ddos the pool
399 2012-10-23 03:31:38 <gmaxwell> DrHaribo: I mentioned earlier that Eligius does that, but it's been fairly few times that it's happened.
400 2012-10-23 03:31:51 <gmaxwell> hm. Miners ignoring the LP and hopping perhaps. What payout scheme do you use?
401 2012-10-23 03:31:56 <DrHaribo> gmaxwell: old versions of cgminer and bfgminer can suddenly ddos the pool, 1tx work is slightly cheaper to keep them fed
402 2012-10-23 03:32:01 <DrHaribo> PPLNS
403 2012-10-23 03:32:11 <DrHaribo> it's probably old cgminer users making those blocks
404 2012-10-23 03:32:26 <DrHaribo> I have tried asking people to upgrade, but they keep running old miners
405 2012-10-23 03:32:29 <gmaxwell> oh you're actually continually giving 1tx work to those nodes? kinda sucks. :(
406 2012-10-23 03:32:33 <DrHaribo> suddenly they bombard the server with work requests
407 2012-10-23 03:32:40 <DrHaribo> yes, but I have to keep my server running smoothly
408 2012-10-23 03:32:54 <gmaxwell> sure. I recommend 100% fees. :P
409 2012-10-23 03:33:00 <DrHaribo> not fun being ddosed by your own users :D
410 2012-10-23 03:33:41 <gmaxwell> Which pool software are you using?
411 2012-10-23 03:33:47 <DrHaribo> my own
412 2012-10-23 03:34:21 <DrHaribo> I spent a lot of time lately to build defenses against these buggy clients so they wouldn't lag down the pool anymore
413 2012-10-23 03:34:33 <gmaxwell> You should be able to precompute many tx work in the background and only increment extranonce.. it saves you _1_ sha256 operation. to only have 1 tx there. This shouldn't make a performance difference.
414 2012-10-23 03:35:29 <DrHaribo> per request yes.. these guys send requests as quickly as the network will carry them
415 2012-10-23 03:35:40 <DrHaribo> they have bugs where they throw away all the work you give them and ask for more
416 2012-10-23 03:37:50 <gmaxwell> Dunno about you, but my cpu can run sha256 about 24 million times per second. :P
417 2012-10-23 03:38:17 <gmaxwell> In any case... how can we help you stop dos attacking the network? You're producing a conspicious amount of 1tx transactions.
418 2012-10-23 03:38:40 <gmaxwell> If we get gavin to post something yelling at people to upgrade miners do you think it will help?
419 2012-10-23 03:39:38 <DrHaribo> I don't know.. I've had slow downs on the server with high reject rates on work.. I've explained why and asked people to upgrade, but nothing..
420 2012-10-23 03:39:42 <Luke-Jr> I don't think that's related.
421 2012-10-23 03:39:53 <Luke-Jr> the miners which are DDoSing the pool *aren't* finding blocks with the work
422 2012-10-23 03:40:15 <DrHaribo> not with the ddosing work, but they do find blocks when they are behaving properly
423 2012-10-23 03:40:27 <Luke-Jr> DrHaribo: oh, you're always giving them neutered work? :/
424 2012-10-23 03:40:31 <DrHaribo> yep
425 2012-10-23 03:40:52 <DrHaribo> I also put the old miners on a special work serving queue that uses only 1 cpu core
426 2012-10-23 03:41:26 <DrHaribo> I'm quite happy with how smoothly the server us running now
427 2012-10-23 03:41:35 <gmaxwell> If you're identifying them, seriously, why not increase fees for those miners. If they're indifferent they'll not care and you'll gain more income. And the ones who are indifferent will upgrade..
428 2012-10-23 03:41:40 <Luke-Jr> DrHaribo: what I do for Eloipool is just keep a queue full of merkleroots, and another one with blanks for backup
429 2012-10-23 03:41:44 <DrHaribo> Better that they produce a few 1tx blocks, than having the server running badly
430 2012-10-23 03:42:21 <Luke-Jr> DrHaribo: better for whom, though? certainly not better for Bitcoin
431 2012-10-23 03:42:23 <DrHaribo> yeah, I also queue up work when the cpu is idle
432 2012-10-23 03:42:27 <gmaxwell> DrHaribo: Well, not better for people whos transactions are left waiting. Your empty blocks ultimately deny room to pools that can keep up.
433 2012-10-23 03:43:17 <Luke-Jr> DrHaribo: my point is that you only need to feed out those blank merkleroots when they're *actually* using up the real ones
434 2012-10-23 03:43:49 <DrHaribo> I always feed from the cache if there is work left, it gives faster response times
435 2012-10-23 03:44:07 <DrHaribo> Look, it's not that I waant to produce 1tx blocks and harm bitcoin... or lose the income from tx fees
436 2012-10-23 03:44:41 <DrHaribo> I'm just trying to keep things running smoothly, and it hasn't been running smoothly before I made these changes
437 2012-10-23 03:44:50 <gmaxwell> ::sigh:: so is this how it has to be? You're going to do whatever greedy thing will work, so to get you not be antisocial we're going to have to make bitcoin nodes delay forwarding of 1txn blocks when there is stuff in the memory pool?
438 2012-10-23 03:45:24 <DrHaribo> why is it greedy? I'm giving up tx fees on some blocks to keep the server afloat
439 2012-10-23 03:45:25 <Luke-Jr> I'm not sure I understand what's going on then
440 2012-10-23 03:45:35 <gmaxwell> As a short term measure you do what you need to do, but it's not acceptable to leave it in a state where you're accidentally dos attacking for a long time.
441 2012-10-23 03:45:56 <gmaxwell> DrHaribo: I meant greedy in the optimization sense. It's a first solution.
442 2012-10-23 03:46:35 <DrHaribo> Sure, it's not the way I want things to stay
443 2012-10-23 03:46:46 <gmaxwell> DrHaribo: In any case, I'm glad to help and to get other people to help.
444 2012-10-23 03:46:48 <DrHaribo> It's more like an emergency measure to keep the server from going down
445 2012-10-23 03:47:30 <gmaxwell> DrHaribo: hm. perhaps stop LPing those nodes?
446 2012-10-23 03:47:35 <DrHaribo> dunno what you can do there... newer cgminer and bfgminer have bugfixes for this kind of thing, but people just aren't upgrading
447 2012-10-23 03:48:10 <gmaxwell> Well??? it's not like you're the only pool that has miners on olde cgminer and bfgminer. :P
448 2012-10-23 03:48:20 <gmaxwell> Luke-Jr: what bug causes this?
449 2012-10-23 03:48:33 <DrHaribo> true, but most bigger pools don't run on 1 server like I do
450 2012-10-23 03:49:07 <Luke-Jr> gmaxwell: when multiple pools are configured, and a secondary pool returns a longpoll, and --failover-only is used, the miner discards all work from its primary pool and hammers it until it gets one with the new block
451 2012-10-23 03:49:10 <gmaxwell> Do you need help setting up a second server? :P
452 2012-10-23 03:49:15 <Luke-Jr> DrHaribo: Eligius does, fwiw
453 2012-10-23 03:49:30 <gmaxwell> Luke-Jr: to be fair, you have less than a quarter of the hashrate.
454 2012-10-23 03:49:40 <DrHaribo> I need help getting enough donations to pay for a few servers, sure
455 2012-10-23 03:50:39 <Luke-Jr> gmaxwell: not when stress testing with GPUMAX months ago
456 2012-10-23 03:50:40 <gmaxwell> Luke-Jr: ah, it keeps throwing it away because it's the old block. ..awesome.
457 2012-10-23 03:52:33 <Luke-Jr> hmm
458 2012-10-23 03:52:48 <Luke-Jr> the older miners had another bug too that might work to effectively bypass this problem
459 2012-10-23 03:52:57 <Luke-Jr> but it would be semi-antisocial too
460 2012-10-23 03:53:05 <gmaxwell> give them a fake new block?
461 2012-10-23 03:53:10 <Luke-Jr> yes and no
462 2012-10-23 03:53:23 <Luke-Jr> they considered the block version to be part of the prevblock header ??? ;)
463 2012-10-23 03:53:53 <gmaxwell> hm. well at the moment, you could do the old version without any major harm except lowering the vote count.
464 2012-10-23 03:54:05 <gmaxwell> Which doesn't harm anything unless we're near the threshold.
465 2012-10-23 03:54:18 <Luke-Jr> or a newer version and risk alerting 0.7.x clients
466 2012-10-23 03:54:27 <gmaxwell> yea, I don't think oging to a new version is a good idea.
467 2012-10-23 03:54:28 <Luke-Jr> but at the same time
468 2012-10-23 03:54:39 <gmaxwell> but there is no harm in every other repeated getwork giving version 0.
469 2012-10-23 03:54:49 <Luke-Jr> using this hack requires being very careful or you'll escalate the problem
470 2012-10-23 03:54:57 <sipa> BlueMatt: why the ...? in gitian it builds
471 2012-10-23 03:55:08 <gmaxwell> understating your support for height in coinbase is harmless I think.
472 2012-10-23 03:55:23 <gmaxwell> Luke-Jr: yea, you'll trigger it.. hm I wonder if someone isn't already doing that.
473 2012-10-23 03:55:24 <Luke-Jr> gmaxwell: until we cross the threshold, and then you lose the blocks
474 2012-10-23 03:55:57 <gmaxwell> Luke-Jr: e.g. loadbalancing a v0 and a newer pool...
475 2012-10-23 03:55:58 <Luke-Jr> gmaxwell: even if they don't do it, some using v1 and some using v2 will trigger it :/
476 2012-10-23 03:56:14 <Luke-Jr> and for that case, you'd have to use v3
477 2012-10-23 03:56:22 <gmaxwell> uugh. yea. no.
478 2012-10-23 03:56:46 <gmaxwell> DrHaribo: if you don't want to up fees on miners that are costing you more to support, perhaps delay their payouts?
479 2012-10-23 03:56:58 <gmaxwell> these people are costing you, and they're making you cost everyone.
480 2012-10-23 03:56:58 <Luke-Jr> but in reality, there are STILL cases that cause pool flooding besides this, just rarer weirder ones :/
481 2012-10-23 03:57:05 <Luke-Jr> which the hack won't help
482 2012-10-23 03:58:25 <DrHaribo> yeah, I suppose it's something to consider
483 2012-10-23 03:59:13 <gmaxwell> DrHaribo: presumably your pool reports stales?
484 2012-10-23 04:00:03 <gmaxwell> and people already recognize stales as a problem with miners? you could report artifically high stales for those users with a link to an explination.
485 2012-10-23 04:00:29 <gmaxwell> but unless those users are most of your pool, I suspect you have multiple issues.
486 2012-10-23 04:01:11 <DrHaribo> 204505 and 204512 were made by the same user
487 2012-10-23 04:01:12 <Luke-Jr> yeah, you could send "broken miner" as reject reasons for good shares 8/10 times I think
488 2012-10-23 04:01:24 <DrHaribo> he also made a block a couple days ago
489 2012-10-23 04:01:32 <DrHaribo> he has no email address registered
490 2012-10-23 04:01:52 <DrHaribo> (I contacted some miners by email and got them to upgrade)
491 2012-10-23 04:02:34 <DrHaribo> yeah, I have been considering sending messages through the reject message mechanism
492 2012-10-23 04:03:45 <gmaxwell> it's probably too much coding. ... but you could totally stop paying them and queue their payments, until they log into the webpage to release them.. and that would let you display a message.
493 2012-10-23 04:03:54 <DrHaribo> I am trying to get stuff ready for ASICs.. I'm testing var diff now and next I want to add GBT
494 2012-10-23 04:04:07 <DrHaribo> I didn't want to get bogged down by this, I already spent too much time on it
495 2012-10-23 04:06:53 <DrHaribo> I would like to put some "upgrade your miner" message in the reject messages though.. can always be useful in the future too
496 2012-10-23 04:36:23 <MC1984> why dont dupe txns get relayed again?
497 2012-10-23 04:37:35 <MC1984> are unconfirmed txn only so trivially double spendable because basic things are not done to atleast alert to the fact?
498 2012-10-23 04:37:54 <Diablo-D3> I dont understand
499 2012-10-23 04:38:40 <MC1984> like relaying dupe txnx so that your mark will probably hear about your attempted theft
500 2012-10-23 04:39:00 <MC1984> or some simple check for this locktime months in the future thing, whatever the fuck that is
501 2012-10-23 04:39:54 <MC1984> is it that while 6 conf remains and always will be the gold standard, stugg could be done to make zero conf much less risky?
502 2012-10-23 05:16:28 <gmaxwell> as usual, the latest atlas thread is picking up not crazy users too. ::sigh:: I've posted a response which I think will at least give some food for thought: https://bitcointalk.org/index.php?topic=119885.msg1291680#msg1291680
503 2012-10-23 05:28:01 <_dr> how do people even come up with these ideas?
504 2012-10-23 06:16:39 <zveda> gmaxwell: are you atlas' arch nemesis?
505 2012-10-23 06:24:45 <zveda> looks like he submitted
506 2012-10-23 06:33:03 <sipa> how so?
507 2012-10-23 06:34:03 <sipa> oh, i see... his account must have been hacked!
508 2012-10-23 06:51:47 <sturles> Wow! https://bitcointalk.org/index.php?topic=119885.msg1291740#msg1291740
509 2012-10-23 06:52:11 <sturles> Oh, I'm late..
510 2012-10-23 06:54:39 <Joric> whoa Atlas's ignore label went gold
511 2012-10-23 07:24:04 <MC1984> I will not questions things any further. I submit. I am gone.
512 2012-10-23 07:24:10 <MC1984> i will eat a golfball if true
513 2012-10-23 07:46:37 <NaruFGT> Just out of curiousity, which bitcoin wallet software/site do the developers in #bitcoin-dev trust?
514 2012-10-23 07:48:37 <t7> none
515 2012-10-23 08:00:39 <sipa> NaruFGT: obviously the software we develop :)
516 2012-10-23 08:17:03 <keshav> I am trying to create a php based website and fetch data from ecoinpool and couchdb can anyone please tell me how to find out the current hash rate
517 2012-10-23 08:18:44 <t7> i dont even trust the official client
518 2012-10-23 08:18:51 <t7> i havnt got time to read every line of code
519 2012-10-23 08:19:06 <t7> and i still might miss something nefarious
520 2012-10-23 08:26:13 <NaruFGT> @t7: of course, that kindof review is difficult for anybody
521 2012-10-23 08:26:27 <NaruFGT> the general idea is that peer review should be sufficient, but yes I'm sceptical as well.
522 2012-10-23 08:27:13 <NaruFGT> I need a hash comfirmation if somebody doesn't mind taking the time: SHA1 (bitcoin-0.7.1-win32-setup.exe) = 1e6fc42dedb44df4bc0f6e84d20e5f83e971f6ea
523 2012-10-23 08:33:02 <jurov> NaruFGT yes: http://sourceforge.net/projects/bitcoin/files/Bitcoin/bitcoin-0.7.1/SHASUMS.asc/download
524 2012-10-23 08:34:42 <NaruFGT> Thanks! couldn't find them :)
525 2012-10-23 08:40:07 <zveda> whats the point of everyone being a full node if nobody reads the whole code
526 2012-10-23 08:40:48 <zveda> how about peer review of transactions then
527 2012-10-23 08:42:31 <t7> zveda: yes. With wax seals on letters delivered by courier
528 2012-10-23 08:43:04 <t7> might take a while to buy stuff but no one buys anything with bitcoin these days, its all about hoarding
529 2012-10-23 08:43:43 <Ferroh> bitcoind can receive M of N transactions but cannot send them, correct?
530 2012-10-23 08:44:19 <NaruFGT> @t7: snoop around the OTC, there are alot of purchases going on :P
531 2012-10-23 08:54:45 <sipa> Ferroh: it can send to p2sh multisig addresses, but can not generally spend funds received on such addresses
532 2012-10-23 08:55:13 <sipa> the raw transaction interface can do everything, though
533 2012-10-23 08:56:19 <Ferroh> sipa: I see. So BIP0016 is implemented then, right? Is https://en.bitcoin.it/wiki/BIP_0019 implemented?
534 2012-10-23 08:56:21 <t7> sipa did you guys have any more ideas about sending transactions around without using p2p network?
535 2012-10-23 08:58:16 <sipa> Ferroh: yes
536 2012-10-23 08:58:35 <sipa> oh, 18
537 2012-10-23 08:58:43 <Ferroh> yes, both are implemented?
538 2012-10-23 08:59:32 <Ferroh> er, bip 18 is implemented...?
539 2012-10-23 08:59:38 <sipa> wait
540 2012-10-23 08:59:46 <sipa> bip16 is implemented
541 2012-10-23 08:59:49 <Ferroh> ok
542 2012-10-23 08:59:52 <Ferroh> and BIP 19?
543 2012-10-23 09:00:11 <sipa> bip19 does not specify anything new
544 2012-10-23 09:00:38 <sipa> it just suggests using a particular script as standard multisig trandaction
545 2012-10-23 09:00:59 <sipa> it's not considered standard by the reference client, but it is valid
546 2012-10-23 09:01:18 <Ferroh> I see, okay thanks :)
547 2012-10-23 09:01:31 <sipa> t7: elaborate?
548 2012-10-23 09:02:23 <t7> i remember you saying that transactions should be sent over email or something, then verified on the blockchain
549 2012-10-23 09:02:31 <t7> maybe im getting the wrong end of the stick
550 2012-10-23 09:07:21 <NaruFGT> @t7: transactions can be sent over email, or over regular mail, they just need to be interpereted by some client of sorts.
551 2012-10-23 09:07:59 <NaruFGT> And until they are interpereted and written into the block chain, they can be invalidated by a double-spend.
552 2012-10-23 09:08:22 <NaruFGT> at least that's the way I understand it
553 2012-10-23 09:09:14 <Ferroh> sipa: do we know when spending from multisig addresses is planned to be added? Currently we can't use M of N at all, since we can't spend from multisig addresses, right?
554 2012-10-23 09:09:23 <sipa> Ferroh: use the raw transaction API
555 2012-10-23 09:10:44 <Ferroh> wait, you mean the bitcoind API can do M of N transactions now?
556 2012-10-23 09:11:25 <Ferroh> oh, no, you can do it using the _raw_ API
557 2012-10-23 09:12:04 <Ferroh> which I have never used. So there is essentially no "easy" way to use M of N yet, correct?
558 2012-10-23 09:13:00 <sipa> no, it's somewhat intended for third-party applications
559 2012-10-23 09:13:26 <sipa> complex transactions are much harder to consider "received money", especially if you're not the only one who can spend it
560 2012-10-23 09:13:54 <sipa> and many of the more interesting applications require some interaction with the sender or other participators
561 2012-10-23 09:16:04 <Ferroh> that makes sense
562 2012-10-23 09:16:34 <Ferroh> what I dont understand,
563 2012-10-23 09:16:47 <Ferroh> is that you said that you can send funds to a multisig address, but can't spend from that address
564 2012-10-23 09:16:55 <Ferroh> do you just mean that you can't do that in the UI?
565 2012-10-23 09:17:10 <Ferroh> You can do this by using the raw transaction API, right?
566 2012-10-23 09:26:24 <sipa> sending to a P2SH multisig address works in the GUI and in the normal sending RPCs
567 2012-10-23 09:26:41 <sipa> consuming such funds requires the raw transaction API
568 2012-10-23 09:29:26 <zveda> but why is peer review good enough for bitcoin-qt code but not good enough for the blockchain
569 2012-10-23 09:29:41 <zveda> suppose everybody verifies random portion of transactions
570 2012-10-23 09:30:02 <sipa> zveda: because for source code, peer review (and automated tests, deterministic builds, ...) is the best we can do
571 2012-10-23 09:30:26 <zveda> true but if that's the weakest link in the chain, why make the other links stronger
572 2012-10-23 09:31:01 <sipa> you're talking about a different level
573 2012-10-23 09:31:20 <sipa> well bitcoin is a consensus system - we agree to follow some rules, have software that verifies it for us, and all we have to do is be sure that the software actually follows those rules
574 2012-10-23 09:31:58 <zveda> hm ok
575 2012-10-23 09:32:09 <sipa> humans can't do crypto, you know
576 2012-10-23 09:32:14 <zveda> yeh I guess it's different
577 2012-10-23 09:36:20 <NaruFGT> Humans can do crypto, even primative public/private keys can be done on paper :P
578 2012-10-23 09:36:35 <NaruFGT> symetric keys are very easy to do on paper.
579 2012-10-23 09:37:11 <sipa> doing an ECDSA verification would take me hours, if not days
580 2012-10-23 09:37:58 <NaruFGT> yes but doing a very very small RSA key would only be a few hours or even a few minutes
581 2012-10-23 09:39:03 <NaruFGT> and we do checksums soo seamlessly that we don't even realize we're doing it.
582 2012-10-23 09:39:26 <_dr> true
583 2012-10-23 09:39:26 <NaruFGT> When you 'recognize' somebody's face or their voice, that's a checksum.
584 2012-10-23 09:39:31 <sipa> sure, but small-key RSA that doesn't provide any security, in a world where massive electronic computing power is available
585 2012-10-23 09:40:03 <_dr> when i'm entering my bank account number i always suffix it with its sha256 hash to make sure i entered it correctly into my online banking form
586 2012-10-23 09:40:55 <_dr> NaruFGT: actually that's what grandfather neurons are for, our brain doesn't implement it using checksums
587 2012-10-23 09:41:01 <NaruFGT> I wonder if there are examples of real asymetric signatures that we use intuatively
588 2012-10-23 09:41:29 <sipa> ok, assume w estart doing that
589 2012-10-23 09:41:41 <NaruFGT> no... assume we already do that
590 2012-10-23 09:41:55 <sipa> at some point, people will start writing software to do the crypto for them, as they trust their own code, and computers are way faster at doing crypto
591 2012-10-23 09:41:57 <_dr> not in the strict mathematical sense
592 2012-10-23 09:41:58 <NaruFGT> and look for elements of our interactions that suggest we do so
593 2012-10-23 09:42:13 <sipa> i'm talking about doing manual blockchain verification
594 2012-10-23 09:42:21 <NaruFGT> Ohh that
595 2012-10-23 09:42:33 <NaruFGT> that's why we have development teams, lol.
596 2012-10-23 09:42:38 <NaruFGT> that's what this is for
597 2012-10-23 09:43:26 <NaruFGT> and as long as multiple implementations exist (not just forks of one project) the network should be fairly trustworthy
598 2012-10-23 09:46:01 <sipa> i wish there were multiple fully-validating node implementations that were used in production, but it's exceedingly hard to validate whether they all work identically in every case
599 2012-10-23 09:46:13 <NaruFGT> If you can trust the routines you've written, and you trust the platform it's being run on, then that's the closest thing to manually verifying the blockchain you can get I would think.
600 2012-10-23 09:46:59 <sipa> and the problem with consensus systems such as bitcoin, is that following the same rules is more important than the rules themselves in many cases
601 2012-10-23 09:47:23 <NaruFGT> Well there are enough nodes that have the core functionality that I don't understand why it's important, the protocol is fault tolerant.
602 2012-10-23 09:47:53 <sipa> ok, imagine someone writes a full node, publishes it, and people start mining on it
603 2012-10-23 09:48:12 <sipa> however, he never realized that the original node software does not allow the genesis block output to be spent
604 2012-10-23 09:48:34 <sipa> at some point, around 50% is using original code, the other half uses this new software
605 2012-10-23 09:48:57 <sipa> now satoshi pops up, and spends the genesis block output
606 2012-10-23 09:49:07 <NaruFGT> as long as there is a client used for the majority of all transactions, protocol mis-implementations are a threat, and can manipulate (whether on purpose or by mistake) the functionality
607 2012-10-23 09:49:23 <sipa> at this point, the network splits in two
608 2012-10-23 09:49:31 <sipa> one which accepts the spend, one that doesn't
609 2012-10-23 09:49:35 <NaruFGT> but hopefully with time we have enough diversity that there is no majority.
610 2012-10-23 09:49:58 <sipa> the same problem exist with 3 implementation that all have around 33%
611 2012-10-23 09:50:08 <NaruFGT> and yes, that would be a temporary problem, an implementation bug causing denial of service.
612 2012-10-23 09:50:16 <sipa> not denial of service
613 2012-10-23 09:50:18 <sipa> theft
614 2012-10-23 09:50:38 <sipa> every coin that existed in the chain before the split can be spent within each fork once
615 2012-10-23 09:50:50 <sipa> and the forks won't know about this about eachother
616 2012-10-23 09:51:25 <sipa> the bottom line is that to participate in the same, you have to agree to follow its rules *exactly*
617 2012-10-23 09:51:26 <NaruFGT> eventually one of the branches would die
618 2012-10-23 09:51:36 <sipa> a lot of damage can be done by then
619 2012-10-23 09:52:04 <NaruFGT> I mean implementation differences resolve with IRL consensus, people change which client they use to reflect their views on the issue.
620 2012-10-23 09:52:04 <sipa> that's somewhat of a worst-case scenario for me, and i think it would instantly kill the economic value
621 2012-10-23 09:52:21 <sipa> there is no 'views', the network rules are not a democracy
622 2012-10-23 09:52:28 <NaruFGT> they are though.
623 2012-10-23 09:52:35 <sipa> no, they're not!
624 2012-10-23 09:52:43 <sipa> the only thing mining is about, is deciding which *valid* chain wins
625 2012-10-23 09:53:00 <sipa> but which chain is considered valid is never voted upon
626 2012-10-23 09:53:03 <NaruFGT> and a change in protocol would result in a fork in the chain
627 2012-10-23 09:53:18 <sipa> yes, that's why changing the protocol cannot be done (unless *everyone* agrees)
628 2012-10-23 09:53:25 <NaruFGT> unless the majority agree
629 2012-10-23 09:53:28 <sipa> NO
630 2012-10-23 09:53:29 <sipa> NO
631 2012-10-23 09:53:30 <sipa> NO
632 2012-10-23 09:54:09 <sipa> i just gave you an example: one minor difference in validity of a chain will result in a chain split, with coins being double spent
633 2012-10-23 09:54:18 <NaruFGT> ok.. unless the majority of clients implement the protocol in that manor?
634 2012-10-23 09:54:21 <sipa> NO
635 2012-10-23 09:54:41 <sipa> to change the validity rules, *everyone* has to agree, not just a majority
636 2012-10-23 09:54:44 <NaruFGT> Yeah and in the short term, that example leads to a fork in the block chain.
637 2012-10-23 09:55:43 <sipa> bitcoin is a combination of two systems: one part that keeps track of a block of trees, which are fully validated by every full node by every client - no discussion - a chain is valid or it isn't, and everyone agrees about this
638 2012-10-23 09:56:33 <NaruFGT> I could change my client to not 'agree' that a block is valid
639 2012-10-23 09:56:59 <NaruFGT> and it won't result in a permanant branch in the network
640 2012-10-23 09:57:08 <NaruFGT> even if I run it on 1000 machines
641 2012-10-23 09:57:08 <sipa> if you mine, it will
642 2012-10-23 09:57:20 <sipa> you'll start living in your own world
643 2012-10-23 09:57:41 <sipa> as soon as the world creates a block you don't agree with
644 2012-10-23 09:57:44 <NaruFGT> if the other miners acknowledge my blocks as valid...
645 2012-10-23 09:57:49 <sipa> not enough
646 2012-10-23 09:58:01 <NaruFGT> then there's a problem with their implementation
647 2012-10-23 09:58:08 <sipa> if they have more hashpower, and you disagree with a block of them, your client wil ignore every block they create
648 2012-10-23 09:58:25 <sipa> and you build upon the last block your node considered valid
649 2012-10-23 09:58:41 <sipa> but you'll never outpace the rest of the world if they have more hashpower
650 2012-10-23 09:58:42 <Luke-Jr> sipa: but if all the miners disagree with you, you are left vulnerable to attacks
651 2012-10-23 09:59:05 <NaruFGT> and the coins I have generated won't be recognized by people outside of my view of the world
652 2012-10-23 09:59:09 <Luke-Jr> well, I guess if the economic majority disagrees with the miners, that risk is reduced
653 2012-10-23 09:59:50 <NaruFGT> and thus I'll have to patch my client or implementation to behave as the rest of the world does to actually use the network
654 2012-10-23 10:00:03 <sipa> simplified: the only thing miners vote about is which *valid* transactions end up in the chain and when - they cannot vote about which transactions are valid
655 2012-10-23 10:00:36 <NaruFGT> they can ignore transactions, that's sortof deeming them not valid, dos
656 2012-10-23 10:00:50 <sipa> they can, that boils down to delaying indefinitely, and that is possible
657 2012-10-23 10:01:11 <sipa> but when another miner puts a valid transaction in the chain, you have to accept it
658 2012-10-23 10:01:16 <NaruFGT> and that's the fork that you described about genesis spending vs not spending the genesis block
659 2012-10-23 10:01:45 <sipa> depends whether you're talking about not accepting into the memory pool, or not accepting when already mined into a block
660 2012-10-23 10:01:55 <sipa> the first is possible, the second isn't (without hard fork risk)
661 2012-10-23 10:01:56 <NaruFGT> and there is no 50/50, there will be a majority of clients implementing or not implementing the spending of those blocks
662 2012-10-23 10:03:06 <NaruFGT> also there are other forms of denial of service, a merchant can refuse to acknowledge coins from a certain block.
663 2012-10-23 10:03:20 <NaruFGT> maybe even send it to a btc address that doesn't exist/has no private key
664 2012-10-23 10:03:36 <sipa> well that'd be stupid :)
665 2012-10-23 10:03:57 <NaruFGT> it'd be a simple way for a majority of merchants to say "I don't care what you say, this isn't real money"
666 2012-10-23 10:04:14 <NaruFGT> and if most of the merchants agree, they will disregard that as normal behavior
667 2012-10-23 10:06:23 <NaruFGT> Seeing how in real life, racism, classful population division, and xenophobia can drive people to do "stupid" things, I can imagine it happening in bitcoin.
668 2012-10-23 10:06:43 <NaruFGT> "I don't like you and your money is no good here" isn't a new idea.
669 2012-10-23 10:15:59 <asciilifeform> NaruFGT: "take your money elsewhere" isn't a bug, it's a feature. (Potentially.) See "Shitcoin": http://www.loper-os.org/?p=988
670 2012-10-23 10:19:59 <NaruFGT> Of course it's a feature, there are no such things as bugs ;)
671 2012-10-23 10:21:44 <NaruFGT> but I still think it's a cool idea to simply delete (send the coins to a non-existant address, or with some mathematical proof, an address that is proven to have no private key[is that possible?]) the coins. Sortof like burning USD only unlike USD, there is no re-issue or replacement, just more deflation :)
672 2012-10-23 10:28:17 <asciilifeform> NaruFGT: that one's easy. Just wipe your wallet.
673 2012-10-23 10:28:36 <sipa> it's not much a statement if you can't proof you've deleted it :)
674 2012-10-23 10:28:52 <NaruFGT> If you can prove there is no private key, it'd mean a whole lot
675 2012-10-23 10:29:01 <conman> no way of doing that
676 2012-10-23 10:29:01 <NaruFGT> I'm actually really interested in finding such a proof
677 2012-10-23 10:29:24 <sipa> you can create a transaction with txout script OP_FALSE
678 2012-10-23 10:29:28 <NaruFGT> I won't be convinced it's impossible until I see a proof that every btc address has an associated private key ;)
679 2012-10-23 10:29:30 <sipa> that's guaranteed to be lost
680 2012-10-23 10:29:41 <NaruFGT> oh?
681 2012-10-23 10:30:07 <sipa> it's not a standard script, so it may not be accepted by any miner or easily relayed through the network
682 2012-10-23 10:30:11 <sipa> but it's certainly valid
683 2012-10-23 10:31:14 <NaruFGT> and it's verifyable that the script was performed?
684 2012-10-23 10:31:31 <sipa> if you could perform it, it would be spent
685 2012-10-23 10:31:55 <sipa> you can prove that no input exists that makes OP_FALSE return true - trivially
686 2012-10-23 10:32:49 <NaruFGT> I'm not familiar with how the inputs and outputs work with the scripts o: I'll have to do more reading
687 2012-10-23 10:33:23 <sipa> outputs contain a script; inputs contain data that makes the output they spend evaluate to true
688 2012-10-23 10:34:11 <sipa> when you do a send to btc address X, it will create a script that says "take as input a signature S and a public key P, and check that S is a valid signature from P, and that the hash of P is X"
689 2012-10-23 10:36:59 <NaruFGT> So deleting coins is possible
690 2012-10-23 10:37:03 <NaruFGT> that's really kindof neat
691 2012-10-23 10:37:44 <NaruFGT> although can brute force find (even with nearly 0 probability) another script that will evaluate true?
692 2012-10-23 10:38:07 <sipa> if the script is OP_FALSE, it will unconditionally always return false
693 2012-10-23 10:38:17 <sipa> there's no way to make OP_FALSE not return false
694 2012-10-23 10:38:30 <sipa> there's not cryptography involved at all
695 2012-10-23 10:38:42 <NaruFGT> that's hash collision
696 2012-10-23 10:39:06 <sipa> ??? there's no hash involved
697 2012-10-23 10:39:32 <sipa> OP_FALSE is defined as "return false"
698 2012-10-23 10:39:36 <NaruFGT> I thought scripts...?
699 2012-10-23 10:40:07 <sipa> if the script is "return false", tell me which input to that script will make it return true
700 2012-10-23 10:40:45 <NaruFGT> What are the script hashes then?
701 2012-10-23 10:40:58 <NaruFGT> I'm feeling more and more naiive
702 2012-10-23 10:41:00 <sipa> are you talking about P2SH?
703 2012-10-23 10:41:27 <zveda> address is a hash of the public key
704 2012-10-23 10:41:52 <sipa> an address is a shorthand for a script
705 2012-10-23 10:41:59 <NaruFGT> the hashes that begin with 5
706 2012-10-23 10:42:19 <sipa> i think you need to understand more before we dive into that :)
707 2012-10-23 10:42:43 <NaruFGT> yeah I agree >_<
708 2012-10-23 10:43:02 <zveda> I remember Mike Hearn talking about parallelizing the blockchain checking
709 2012-10-23 10:43:06 <zveda> are people still talking about that?
710 2012-10-23 10:43:34 <sipa> i already implemented that, but there were some bugs, so it's delayed for now
711 2012-10-23 10:43:49 <zveda> oo
712 2012-10-23 10:43:52 <NaruFGT> making a non-protocol service for blockchain checking would be a fine feature.. for the non-official clients.
713 2012-10-23 10:44:30 <NaruFGT> lol I'm going to shut up since I don't know what I'm talking about
714 2012-10-23 10:44:39 <zveda> so in your implementation.. how does it work. I can have many machines that each check part of the blockchain ?
715 2012-10-23 10:45:56 <zveda> would be nice if I could prove that I checked a bunch of transactions..
716 2012-10-23 11:00:00 <kuldeepdhaka> help: im a php developer, i want to create a exchange for indian currency, im new to bitcoin
717 2012-10-23 11:03:39 <_dr> can someone enlighten me about this magic number: MAX_BLOCK_SIGOPS = MAX_BLOCK_SIZE/50?
718 2012-10-23 12:09:58 <gavinandresen> _dr : that says a block cannot contain more than 1000000/50 == 20,000 ECDSA signature-checking operations
719 2012-10-23 12:32:30 <gmaxwell> NaruFGT: You might find this rant I wrote about the 'majority' beliefs informative https://bitcointalk.org/index.php?topic=61922.msg723476#msg723476 certantly if 'all' the users agreed to run bitcoin2 instead of bitcoin then that would be that; but _within_ the system the rules are not defined by a majority and certantly not a simple majority.
720 2012-10-23 12:33:54 <NaruFGT> Oh?
721 2012-10-23 12:34:12 <helo> what's the "accepted" translation of hashes/sec to flops?
722 2012-10-23 12:34:57 <helo> (for penis length comparisons between the top500 list and the bitcoin network)
723 2012-10-23 12:36:05 <upb> the factor could be 0, 1 or infinite
724 2012-10-23 12:36:50 <helo> i guess down the line if someone is going to do an attack, it is going to be with all asic
725 2012-10-23 12:36:51 <BlueMatt> there are a few threads, like https://bitcointalk.org/index.php?topic=50720.0;all and https://bitcointalk.org/index.php?topic=52303.0
726 2012-10-23 12:37:05 <helo> so we'd have to figure out how many hashes/sec/transistor a good asic will get
727 2012-10-23 12:37:06 <NaruFGT> "Now it would have been Satoshi's dream to make the entire system work completely like this but sadly Einstein came along and screwed everything up: relativity says that the temporal ordering of events is different for every observer and depends on your mutual locations in spacetime. A decentralized system does not exist in just one place and thus there can be no single constant decentralized
728 2012-10-23 12:37:14 <helo> and then how many transistors are in the top500 list
729 2012-10-23 12:37:37 <NaruFGT> I would think the distance to time here, and the latency between transactions would make this irrelevant
730 2012-10-23 12:38:08 <helo> NaruFGT: the principle still holds... there can be no correct order arrived at
731 2012-10-23 12:38:21 <helo> NaruFGT: no universally verifiable order
732 2012-10-23 12:38:22 <gmaxwell> NaruFGT: after I wrote that someone directed me to a lovely paper by Leslie Lamport in the 1970s on that subject... I'm kinda bummed that I wasn't the first person to make that physics/distributed systems connection.
733 2012-10-23 12:38:28 <kjj_> why won't people let the FLOP thing go?
734 2012-10-23 12:38:38 <helo> kjj_: i just did
735 2012-10-23 12:39:15 <gmaxwell> NaruFGT: no, because no matter what the latency is, an attacker can arrange their timing so they're right on the boundary for some participants and not others. In pratical systems the vulgarities of real networks create a similar effect without the actual physics coming into play.
736 2012-10-23 12:40:31 <NaruFGT> lol well until the network exists in vastly different general reletivity contexts' it's trivial right?
737 2012-10-23 12:40:53 <BlueMatt> kjj_: because its a semi-valid comparison, at least when you phrase it as "if we were actually using flops, assuming most of our power is coming from a standard amd gpu, we would have X flops"
738 2012-10-23 12:41:39 <kjj_> BlueMatt: except that hashes and FLOPS are apples and chairs.
739 2012-10-23 12:43:09 <BlueMatt> kjj_: yes, but if you assume that most hashes are from a standard amd gpu you can state that the amount of flop power which could otherwise be used, but is instead pointed at bitcoin is X (assuming no intop/flop multitasking on those gpus, which happens some, but not much)
740 2012-10-23 12:43:54 <gmaxwell> NaruFGT: huh, no??? it doesn't require relativistic speeds. This works as just a consequence of conventional observers as midly seperated locations. Actual networks just exacerbate it by having non-uniform speeds between users. Believe me, we wouldn't bother with this messy consensus algorithim if we knew a better way to do order that wasn't pratically insecure.
741 2012-10-23 12:44:13 <BlueMatt> kjj_: and unless you have a better way to compare the total energy draw of bitcoin to the energy draw (and other metrics) of eg f@h...
742 2012-10-23 12:44:45 <helo> NaruFGT: special relativity concepts come into play whenever there is not an infinitesimal latency between participants
743 2012-10-23 12:44:46 <gmaxwell> kjj_: but when most of the hashpower is asics then that will be even more nuts. Perhaps better to report MIPS. :P
744 2012-10-23 12:44:54 <kjj_> BlueMatt: I just don't compare them.
745 2012-10-23 12:45:11 <BlueMatt> kjj_: but...but...epeen...
746 2012-10-23 12:45:38 <gavinandresen> Yeah, I don't think comparing hashpower to general-purpose computing power is the right thing to do, either.
747 2012-10-23 12:45:43 <kjj_> I suppose you could tie it all back to entropy, but still, different processors, different processes
748 2012-10-23 12:45:46 <gmaxwell> NaruFGT: I just invoked the relativity argument to show that the issue is fundimental, we can't just fix it with faster computers/networks.
749 2012-10-23 12:45:47 <BlueMatt> yea, when asics become a reasonably fraction of hash power, it all goes out the window
750 2012-10-23 12:46:29 <gavinandresen> It might be good PR, though. Then again, it might not, because it makes people think about all of that "wasted" electricity.
751 2012-10-23 12:46:36 <gmaxwell> If you want brag point out our sum difficulty.
752 2012-10-23 12:46:49 <BlueMatt> gavinandresen: yea, Ive seen the "wasted" power argument a lot...
753 2012-10-23 12:47:00 <gmaxwell> But then again, pointing out that we've done more computations that .. well. anything has the problem gavinandresen just mentioned. :P
754 2012-10-23 12:47:12 <gmaxwell> BlueMatt: Its pretty easily dismantled.
755 2012-10-23 12:48:17 <BlueMatt> gmaxwell: hmm? aside from (invalidly) arguing that intop != flop, it is valid to state that bitcoin uses a large amount of power that would (likely) not have been used for anything given no bitcoin-like system in (popular) existence
756 2012-10-23 12:48:36 <gmaxwell> (Point out the enormous security costs conventional currencies have??? especially ones that don't have inherent counterfeiting resistance bitcoin has; mostly people complain about energy when they're thinking of mining as a competition for coins rather than a source of security)
757 2012-10-23 12:48:38 <BlueMatt> (obviously intop != flop, but for the sake of power cost, its similar(ish))
758 2012-10-23 12:49:29 <BlueMatt> gmaxwell: sure, but given that no bitcoin-like system exists, you are still using electricity that would otherwise have gone unused (or uncreated, or...whatever you call it)
759 2012-10-23 12:50:29 <gmaxwell> And perhaps the power usage is higher or lower than it should be??? but once you're comparing it to prisons for counterfeiters, currency police, armored cars, money counters, bank vaults.. it's not so obviously inefficient anymore. It's true, it's more energy, but if everyone used it would we be better off energy wise than the alternatives it displaces? The answer is not obviously no at least.
760 2012-10-23 12:51:00 <BlueMatt> ok, thats fair
761 2012-10-23 12:51:26 <BlueMatt> so now we should argue the "green" (well gov't cost-savings) aspects of bitcoin as a national/international currency replacement
762 2012-10-23 12:52:25 <BlueMatt> :)
763 2012-10-23 12:55:17 <gmaxwell> Well??? thats a bit political for my taste.. at the same time currencies aren't the only thing the bitcoin replaces though. E.g. for people that use it can replaces checks. Or escrow systems for large payments. It can potentially replace deed registration systems (e.g. smart property ideas). I guess it's just an argument appealing to the categorical imperative??? if everyone uses it it doesn't become bad for the enviroment (in fact it should
764 2012-10-23 12:57:09 <NaruFGT> wow I never thought of the non-trivial latency's effect on consensus
765 2012-10-23 12:59:22 <BlueMatt> gmaxwell: yep
766 2012-10-23 13:06:27 <helo> so if there is something in nature that requires distributed consensus, part of the bitcoin proto may be out there already :)
767 2012-10-23 13:09:40 <gmaxwell> helo: sadly not, the weird thing is that bitcoin isn't fully generalizable. If you decouple the economic incentives it may lose its security.
768 2012-10-23 13:10:30 <NaruFGT> Incentives are abstract, as long as there is a strong drive it will work
769 2012-10-23 13:15:11 <helo> so each time a new ordering of events is established, (some) participants must get some energy
770 2012-10-23 13:17:51 <helo> yeah, i can't really see that in nature
771 2012-10-23 13:18:43 <gmaxwell> helo: right, bitcoin requires establishing ordering to have a _cost_, so we don't have to worry about assigning identities to people, we just take the ordering that has the greatest total expenended cost. Without the cost an attacker could start 10,000 clones and use them to gain control of the ordering.
772 2012-10-23 13:19:29 <gmaxwell> It's viable in bitcoin because??? as a currency??? we can compensate people for that cost, via txn fees.
773 2012-10-23 13:20:15 <gmaxwell> But e.g. if you wanted to use the bitcoin algorithim to decenteralize wikipedia (to reach a consensus on which version of pages was the current one) it's less obvious to me that it would be secure.
774 2012-10-23 13:20:49 <BlueMatt> merged mining chains are close to a generalized bitcoin (though they obviously depend on bitcoin itself's security)
775 2012-10-23 13:23:09 <gmaxwell> Right. Indeed. Though they still have incentive problems. Why bother merge mining something at all? and if few people do it, then a small number that do can still overpower. Though since they copy times from bitcoin (namecoin doesn't but something else could), I suppose some kinds of attacks could be closed even with small numbers of miners.
776 2012-10-23 13:27:34 <BlueMatt> yea, you need /some/ miners, but in many systems finding "goodwill" miners isnt hard, you just may not get many of them
777 2012-10-23 13:29:05 <gmaxwell> BlueMatt: It's _really_ easy to mergemine random stuff with p2pool, but generally hard otherwise.
778 2012-10-23 13:30:11 <BlueMatt> gmaxwell: yea, and (hopefully) p2pool grows and systems like it (or p2pool itself) become the majority of pools
779 2012-10-23 13:30:21 <BlueMatt> at least distributed pools where you can easily merged-mine your own stuff
780 2012-10-23 13:37:44 <xblitz> anybody familiar with the new CCoinsView ?? I had a fork with some modifications to transactions informations display .. and of course it doesnt work anymore.. how would i do this now: http://pastebin.com/BzHDchwQ
781 2012-10-23 13:41:33 <gmaxwell> xblitz: no longer possible to do reliably in ultraprune.
782 2012-10-23 13:42:04 <gmaxwell> Unless I misunderstand what you're doing there? I assume this was to show the prior inputs on random transactions?
783 2012-10-23 13:42:45 <gmaxwell> Sipa had talked about adding an optional index for historical transactions, but it's not there yet... so you can only look up unspent transactions right now.
784 2012-10-23 13:42:45 <xblitz> yup pretty much
785 2012-10-23 13:42:57 <xblitz> crap...
786 2012-10-23 13:44:19 <BlueMatt> Ive got some optional code to do that in bitcoinj...come to the dark side (we have cookies!)
787 2012-10-23 13:44:42 <xblitz> cookies!
788 2012-10-23 13:45:40 <BlueMatt> also, its in java and isnt really production-ready (at least the verification part)...do you like debugging?
789 2012-10-23 13:48:32 <xblitz> humm.. well Im just kind of having fun on the main client and putting features that i like..
790 2012-10-23 13:49:16 <gmaxwell> xblitz: sadly thats a feature that is inherently really expensive to support.
791 2012-10-23 13:50:06 <xblitz> but what about full nodes? will they have to have a different client?
792 2012-10-23 13:50:07 <gmaxwell> (it's pruning incompatible, requires additional indexes, etc)
793 2012-10-23 13:50:17 <gmaxwell> xblitz: We're still a full node.
794 2012-10-23 13:50:21 <BlueMatt> xblitz: its still a full node
795 2012-10-23 13:50:23 <BlueMatt> aww
796 2012-10-23 13:51:00 <xblitz> oh .. it still keeps all the blockdata but only indexes the "remaining" coins in memory?
797 2012-10-23 13:51:11 <BlueMatt> s/in memory/on disk/
798 2012-10-23 13:51:12 <BlueMatt> yes
799 2012-10-23 13:51:32 <gmaxwell> A fully pruned node is still a full node in the bitcoin systems parlance, though thats confusing... but we're not a pruned node either for that matter. The reason it currently doesn't work is that we don't have an index for fast random access.
800 2012-10-23 13:51:44 <BlueMatt> there is (theoretically) no set of messages you can send to an old node that an ultraprune node will not respond the same to (over the p2p network)
801 2012-10-23 13:51:53 <gmaxwell> If you wanted to do a linear rescan of the blockchain you could look up that data.. but you'd wait several minnutes for that. :P
802 2012-10-23 13:52:15 <xblitz> okay i see!
803 2012-10-23 13:53:21 <gmaxwell> We'll probably end up with a model where there are {SPV, Full, Archive} which do no validation except longest chain, applies all the rules and has reviewed the history and has full security, can bootstrap new nodes or do historical data queries; respectively..
804 2012-10-23 13:54:02 <gmaxwell> And then probably later {SPV, SPV-Full, Full, Archive} with SPV-Full being a node that has full security going forward but only SPV security for the past.
805 2012-10-23 13:54:02 <xblitz> yeah I guess it makes sense..
806 2012-10-23 13:55:22 <BlueMatt> and Full-Pruned, but SPV-Full is close to that but with less verification of old data
807 2012-10-23 13:55:31 <BlueMatt> or do you mean Full is also pruned?
808 2012-10-23 13:55:43 <gmaxwell> I mean full is pruned; only archive is not pruned.
809 2012-10-23 13:55:47 <BlueMatt> ah, ok
810 2012-10-23 13:56:05 <BlueMatt> (though archive doesnt necessarily have to do more than store the data, you can be spv-archive)
811 2012-10-23 13:56:17 <xblitz> okay so about my code at http://pastebin.com/BzHDchwQ .. If I would still like to get the info for unspent coins.. how would i go about?
812 2012-10-23 13:56:36 <gmaxwell> ugh. Then you're potentially a DOS vector when your serve bad data. But I guess it could be spv validated.
813 2012-10-23 13:58:01 <xblitz> hummm wait.. i think it doesnt make sens
814 2012-10-23 13:58:02 <xblitz> sense
815 2012-10-23 13:58:20 <xblitz> forgetr my question.. i'll reread the code
816 2012-10-23 13:58:57 <BlueMatt> gmaxwell: meh, you could be spv-nonvalidated-archive without much issue...as long as you are only serving up to some checkpoint or a point in the chain fairly far back...
817 2012-10-23 14:01:31 <gavinandresen> more likely you're serving FROM some checkpoint, I think
818 2012-10-23 14:01:54 <BlueMatt> well that is obv a potential dos issue
819 2012-10-23 14:02:25 <sipa> BlueMatt: any further problems/progress with building git head on buildtester?
820 2012-10-23 14:02:43 <BlueMatt> sipa: not afaik (aside from the leveldb thing last night, sorry I havent looked into it much yet)
821 2012-10-23 14:02:50 <sipa> ok, no prob
822 2012-10-23 14:03:05 <sipa> it works on gitian, but that's all i know
823 2012-10-23 14:03:09 <gmaxwell> BlueMatt: Can you spare me a minute to figure out how to get FullBlockTestGenerator compiled? I don't see it in the maven stuff. I'm mostly java infrastructure ignorant.
824 2012-10-23 14:03:35 <BlueMatt> sipa: ok, it may very well be a config/jenkins/pull-tester specific thing then, Ill look further
825 2012-10-23 14:04:30 <BlueMatt> gmaxwell: Im also fairly java infrastructure ignorant...all I know is I did mvn eclipse:eclipse, opened the project in eclipse, opened the file and hit run...
826 2012-10-23 14:04:50 <BlueMatt> though I suppose you could javac the specific java file and java to run it?
827 2012-10-23 14:04:57 <BlueMatt> with enough classpath variables, that is
828 2012-10-23 14:05:22 <gmaxwell> BlueMatt: hah. yea, okay I can do that. :P
829 2012-10-23 14:07:31 <molecular> sipa, ok, my bitcoin-qt (ultraprune) finally is done syncing... should I swap in the old wallet.dat?
830 2012-10-23 14:09:12 <molecular> sipa, I have the wallet.dat backed up from before I updated to ultraprune. probably try that one?
831 2012-10-23 14:09:35 <sipa> molecular: that should work
832 2012-10-23 14:10:10 <sipa> molecular: how long did it take?
833 2012-10-23 14:11:58 <sipa> molecular: note that several bugs have been found and fixed in git head in the mean time
834 2012-10-23 14:12:34 <molecular> something like 30 hours
835 2012-10-23 14:13:13 <molecular> can I back up the currently synced blockstore by doing "cp -r blocks blktree /tmp/backup" ?
836 2012-10-23 14:13:31 <molecular> ^sipa
837 2012-10-23 14:13:46 <sipa> molecular: yes!
838 2012-10-23 14:13:48 <molecular> sipa, I'll do what you want, should I "git pull" first?