1 2012-10-31 00:05:11 <runeks> yeah I know. I'm grasping straws. I can't figure out why the exact same function with seemingly the exact same input gives different output.
2 2012-10-31 00:07:27 <runeks> OK now it works...
3 2012-10-31 00:07:48 <jrmithdobbs> bad copy/paste? heh
4 2012-10-31 00:09:09 <runeks> I had reversed the byte order of the hash to be signed. so it would print out the same way it does when printing it from key.cpp. apparently I shouldn't do that.
5 2012-10-31 00:10:14 <kjj_> bitcoin is an ungodly tangle of endianness confusion
6 2012-10-31 00:11:44 <runeks> OK. now I see what was wrong. apparently, the hash, in bitcoin-qt, is stored in reverse byte order. in order to verify it in my Python code I had to reverse the byte order of *only* the hash (not the pubkey/signature)
7 2012-10-31 00:12:42 <runeks> so when I didn't reverse the byte order and my Python code verified it correctly, bitcoin-qt didn't (because it used the reverse byte order).
8 2012-10-31 00:12:44 <runeks> phew!
9 2012-10-31 00:12:46 <Diablo-D3> runeks: its stored in little endian
10 2012-10-31 00:12:52 <Diablo-D3> even though its fucking text
11 2012-10-31 00:12:57 <Diablo-D3> and no one can fix it now
12 2012-10-31 00:13:27 <kjj_> even worse, some of the hashes are of the ASCII hex strings
13 2012-10-31 00:13:46 <runeks> so it's inherent to the uint256 type?
14 2012-10-31 00:13:56 <Diablo-D3> runeks: nope
15 2012-10-31 00:14:06 <Diablo-D3> because I have no clue what it is on a big endian machine
16 2012-10-31 00:14:20 <kjj_> heh. Satoshi was a better cryptographer than he was a programmer
17 2012-10-31 00:14:30 <Diablo-D3> however, its invalid for a textual representation of a number to be anything but big endian
18 2012-10-31 00:14:52 <kjj_> some things are in network byte order, some things are in the memory byte order of the machine he wrote the first version on
19 2012-10-31 00:15:07 <Diablo-D3> kjj_: no
20 2012-10-31 00:15:14 <Diablo-D3> its quite entirely possible its in big endian on big endian
21 2012-10-31 00:15:43 <Diablo-D3> or you could end up with a broken endian
22 2012-10-31 00:15:56 <Diablo-D3> 256 bit numbers are not native on any cpu that bitcoin runs on
23 2012-10-31 00:16:12 <kjj_> yeah, some things in the protocol are Intel-endian
24 2012-10-31 00:16:27 <Diablo-D3> little
25 2012-10-31 00:16:31 <Diablo-D3> and like I said, I dont know
26 2012-10-31 00:16:32 <kjj_> I'd have to go back through my notes to remember what
27 2012-10-31 00:16:43 <Diablo-D3> but yeah, hashes are one of the things that are fucked up
28 2012-10-31 00:17:06 <jgarzik> very few things are in network byte order
29 2012-10-31 00:17:10 <runeks> Diablo-D3: it seems to be the case though. because when I print out the hash like this in Python: hash[::-1].encode('hex')
30 2012-10-31 00:17:12 <jgarzik> vast majority is little endian
31 2012-10-31 00:17:20 <Diablo-D3> jgarzik: thats the problem
32 2012-10-31 00:17:22 <Diablo-D3> its text
33 2012-10-31 00:17:25 <Diablo-D3> endian does not apply here
34 2012-10-31 00:17:29 <Diablo-D3> if its anything but big endian, it is wrong.
35 2012-10-31 00:17:31 <runeks> I get the same output as when I print it out like this in key.cpp: hash.GetHex().c_str()
36 2012-10-31 00:17:39 <runeks> so somewhere the byte order is being reversed
37 2012-10-31 00:18:03 <jgarzik> endian does apply, because we do less-than comparisons
38 2012-10-31 00:18:06 <jgarzik> not just byte compare
39 2012-10-31 00:18:10 <Diablo-D3> jgarzik: not in text you dont.
40 2012-10-31 00:18:14 <runeks> so in Python I have to reverse it and in bitcoin-qt I do not
41 2012-10-31 00:18:36 <jgarzik> Diablo-D3: in bitcoin we do, and this is #bitcoin-dev, so that's all that matters ;p
42 2012-10-31 00:18:47 <Diablo-D3> jgarzik: yes, and bitcoin does not comply with reality
43 2012-10-31 00:18:51 <Diablo-D3> which is a massive bug
44 2012-10-31 00:19:01 <Diablo-D3> and it cant be fixed because too much also broken software relies on that
45 2012-10-31 00:19:04 <Diablo-D3> so fuck satoshi
46 2012-10-31 00:19:13 <Diablo-D3> this is why I hate when academics write code
47 2012-10-31 00:19:28 <Diablo-D3> in theory their shit works, but goddamned is it so goddamned fucking useless
48 2012-10-31 00:20:41 <runeks> works in practice as well
49 2012-10-31 00:21:03 <jgarzik> runeks: JFYI: https://github.com/jgarzik/pynode/tree/master/bitcoin and http://yyz.us/bitcoin/poold.py
50 2012-10-31 00:21:22 <jgarzik> runeks: both of those codebases have examples of dealing with bitcoin data structures + python longs + hex numbers
51 2012-10-31 00:22:25 <runeks> I actually looked through your code. to find out how to use OpenSSL in Python.
52 2012-10-31 00:24:30 <runeks> I think I've learned what I need to wrt. dealing with numbers and hex and data by now. what threw me off was that uint256 was printed in reverse byte order, as opposed to all the other types I was printing
53 2012-10-31 00:32:15 <jgarzik> hum
54 2012-10-31 00:32:25 <jgarzik> sometimes code works, when you did not expect it to work.
55 2012-10-31 00:45:52 <DBordello> what would be the consequences of someone throwing a ton of hash power at the test net and then abandoning it?
56 2012-10-31 00:46:11 <kjj_> free coins for whoever did it
57 2012-10-31 00:47:02 <kjj_> testnet will accept a block resetting difficulty after a little while without any blocks
58 2012-10-31 00:47:36 <DBordello> ah, I didn't realize that
59 2012-10-31 00:53:25 <gmaxwell> DBordello: people abadoning it was why we added that rule
60 2012-10-31 00:53:43 <DBordello> gmaxwell, seems like a good one. How long does it take to reset?
61 2012-10-31 00:53:49 <gmaxwell> people would push the difficulty up to 500 then vanish and we'd go days without a block.
62 2012-10-31 00:54:15 <gmaxwell> DBordello: 20 minutes after the last block the network will accept one at diff 1.
63 2012-10-31 00:54:28 <DBordello> gmaxwell, great rule :)
64 2012-10-31 00:54:43 <gmaxwell> there is a bug in it that can cause the difficulty to be rest to one too..
65 2012-10-31 00:55:05 <gmaxwell> (if one of those 20 minute blocks is the retarget the difficulty is warped back to one)
66 2012-10-31 00:55:32 <DBordello> ah
67 2012-10-31 00:55:44 <DBordello> is it actively mined?
68 2012-10-31 00:55:50 <DBordello> or at least passively?
69 2012-10-31 00:57:05 <gmaxwell> it's continually mined, yes
70 2012-10-31 00:57:17 <gmaxwell> Jeff and I at least. If nothing else it gets those 20 minute blocks.
71 2012-10-31 01:48:38 <Icoin> bitcoind 0.7.1 @rpi https://bitcointalk.org/index.php?topic=93724.msg1308306#msg1308306
72 2012-10-31 01:48:51 <Icoin> ahh sorry bitcoin-qt
73 2012-10-31 01:57:28 <phantomcircuit> Icoin, im guessing that doesn't perform very well
74 2012-10-31 01:58:15 <Icoin> member:phantomcircuit: the performance is surprisingly good..
75 2012-10-31 01:58:28 <Icoin> of course the chain load isnt the fastest
76 2012-10-31 01:58:46 <Icoin> but once the chain is loaded it works
77 2012-10-31 02:00:44 <Icoin> bitcoin-qt 0.7.1 works stable on rpi
78 2012-10-31 02:01:39 <Icoin> 512 mb ram helps a lot
79 2012-10-31 02:06:44 <BlueMatt> wait...so why do you need to do a custom build...?
80 2012-10-31 02:09:51 <forsetifox> And why did you throw your build instructions in a thread about Devcoin Bounty Security? Heh.
81 2012-10-31 02:30:01 <jgarzik> picocoin appears to load and validate up to height 193000, in under 1 second
82 2012-10-31 02:30:13 <jgarzik> better than I expected
83 2012-10-31 02:30:46 <jgarzik> blockchain database correctly calculates hashBestChain etc. through forks
84 2012-10-31 02:31:10 <jgarzik> *headers only
85 2012-10-31 02:31:18 <Luke-Jr> ah
86 2012-10-31 02:31:21 <Luke-Jr> big difference :p
87 2012-10-31 02:31:30 <Luke-Jr> I was about to say "time to throw away Satoshi's code then?" ;)
88 2012-10-31 02:45:04 <amiller> jgarzik, that's cool.
89 2012-10-31 07:08:12 <ThomasV> why can a transaction be included in two blocks?
90 2012-10-31 07:08:21 <Icoin> bitcoin-qt and devcoin-qt @ rpi https://bitcointalk.org/index.php?topic=93724.msg1308306#msg1308306
91 2012-10-31 07:09:33 <sipa> ThomasV: can it?
92 2012-10-31 07:09:57 <ThomasV> oh, no, wait, it was an orphaned one, sorry
93 2012-10-31 08:11:56 <abrkn> if i send money to another address in my wallet, nothing is sent to the network?
94 2012-10-31 08:14:30 <sipa> sure it is
95 2012-10-31 08:14:53 <sipa> addresses are public, and ownership of coins by addresses is tracked by the network
96 2012-10-31 08:19:36 <abrkn> ah, it was just a little slow
97 2012-10-31 08:19:53 <abrkn> trying to set up a wallet balance tracking mechanism for my awesome game
98 2012-10-31 10:04:19 <abrkn> https://github.com/abrkn/blockchain <- finally made this pile of shit work!
99 2012-10-31 11:10:28 <drizztbsd> ACTION hates node.js trend.
100 2012-10-31 12:25:23 <drizztbsd> we should change bitcoin implementation when first asic board comes out :P
101 2012-10-31 12:28:35 <gmaxwell> drizztbsd: Thats a terrible idea. The introduction of asic mining hardware to the public is good news.
102 2012-10-31 12:29:18 <drizztbsd> why?
103 2012-10-31 12:29:46 <gmaxwell> Because only by doing that can the network be strong against an attacker who would have their own made.
104 2012-10-31 12:30:58 <gmaxwell> The whole idea of bitcoin is that attacking is expensive enough that you're better off cooperating, this doesn't work so well if the attacker has a huge price advantage. E.g. if the attacker uses the same technology as the honest users then he must spend more than them.
105 2012-10-31 12:31:05 <[Tycho]> !seen tcatm
106 2012-10-31 12:31:06 <gribble> tcatm was last seen in #bitcoin-dev 1 week, 6 days, 12 hours, 46 minutes, and 37 seconds ago: <tcatm> MC1984: 0.7.1 will load it automatically, for 0.7.0 -loadblock is needed
107 2012-10-31 12:31:22 <[Tycho]> Where is he ?
108 2012-10-31 12:31:41 <gmaxwell> drizztbsd: If the attacker uses technology 10x more efficient, then they need to spend much less.
109 2012-10-31 12:31:53 <drizztbsd> yep, 51% attach
110 2012-10-31 12:31:54 <drizztbsd> attac*
111 2012-10-31 12:31:54 <gmaxwell> [Tycho]: no idea.
112 2012-10-31 12:41:31 <[Tycho]> Got an e-mail from someone @icsi.berkeley.edu telling that they are investigate bitcoin-related botnets :)
113 2012-10-31 12:41:43 <[Tycho]> *investigating
114 2012-10-31 12:45:55 <Kakasper> Why is Bitcoin.org calling itself the official site of Bitcoin?
115 2012-10-31 12:46:53 <kjj_> Atlas, is that you?
116 2012-10-31 12:48:47 <Kakasper> Serious question.
117 2012-10-31 12:49:26 <kjj_> you know what's really funny? the word official only appears on that page in the negative
118 2012-10-31 12:50:01 <Kakasper> http://www.google.com/search?client=ubuntu&channel=fs&q=bitcoin&ie=utf-8&oe=utf-8
119 2012-10-31 12:50:15 <Kakasper> "Official site offering documentation, forums and the open source client software"
120 2012-10-31 12:50:25 <kjj_> what's your point?
121 2012-10-31 12:50:49 <kjj_> did you look at the page, or are you just bitching because google has it tagged as the official site?
122 2012-10-31 12:50:52 <Kakasper> It doesn't leave room for competition on Google if everyone is led to believe that is the central point.
123 2012-10-31 12:51:01 <kjj_> take it up with google then
124 2012-10-31 12:51:04 <Kakasper> No, you guys tagged it that way
125 2012-10-31 12:51:13 <kjj_> you guys?
126 2012-10-31 12:51:15 <Kakasper> It's in the pages meta
127 2012-10-31 12:51:29 <Kakasper> whoever here maintains the page
128 2012-10-31 12:52:10 <Icoin> hi can someone please add the rpi bitcoin-qt 0.7.1 biany add to the main bitcoin.org site ?
129 2012-10-31 12:52:26 <Icoin> binary*
130 2012-10-31 12:53:09 <Icoin> https://bitcointalk.org/index.php?topic=101559.msg1308072#msg1308072
131 2012-10-31 12:53:10 <kjj_> honestly, I have no idea who runs that site. the site itself doesn't make any claims, but even if it did, I wouldn't care
132 2012-10-31 12:53:56 <sipa> Kakasper: pull requests welcome; http://github.com/bitcoin/bitcoin.org
133 2012-10-31 12:54:07 <Kakasper> Thanks.
134 2012-10-31 12:54:29 <kjj_> well, that guy sure lived up to the Kaka part of his name. :(
135 2012-10-31 12:54:53 <kjj_> anyone ever use a hardware RNG other than on a newish Intel motherboard?
136 2012-10-31 12:55:48 <sipa> Icoin: why devcoin? patches needed -> please get them merged in mainline
137 2012-10-31 12:56:18 <Icoin> sipa: sorry wrong window https://bitcointalk.org/index.php?topic=101559.msg1308335#msg1308335
138 2012-10-31 12:56:56 <Icoin> sipa: devcoin is mergedmining capable since quite a while
139 2012-10-31 12:57:59 <sipa> Icoin: not sure what you mean, but we're not going to put an altcoin binary on the main page
140 2012-10-31 12:58:28 <Icoin> sipa: i talk about a bitcoin-qt binary not the devcoin-qt binary
141 2012-10-31 12:58:41 <sipa> Icoin: ok
142 2012-10-31 12:59:28 <Icoin> sipa: both are compiled but for you guys i just talk about bitcoin-qt for rpi https://bitcointalk.org/index.php?topic=101559.msg1308335#msg1308335
143 2012-10-31 13:00:47 <sipa> does it require any changes to the source code or makefiles? can its builds be verified through gitian?
144 2012-10-31 13:00:59 <Icoin> sipa: no changes
145 2012-10-31 13:01:16 <Icoin> sipa: its the original source compiled on a rpi
146 2012-10-31 13:01:54 <Icoin> sipa: you find a howto in the thread i posted earlyer
147 2012-10-31 13:02:22 <sipa> does it require an rpi to compile?
148 2012-10-31 13:02:33 <Icoin> sipa bitcoin-qt was compiled same as default without qr-code
149 2012-10-31 13:03:25 <sipa> a gitian build would be nice, but i don't think the userbase for rpis is large enough to burden releases
150 2012-10-31 13:04:01 <Icoin> since the compilation took more then two hours and there is just limited ram present 512 mb i have to wait to compile it with qr-code
151 2012-10-31 13:04:01 <sipa> (though i'll try it myself :p)
152 2012-10-31 13:07:07 <sipa> well, make it work in gitian (tedious probably, and will require some hacking, as you'll need a rbi sdcard image), and it becomes relatively easy
153 2012-10-31 13:07:28 <sipa> but i don't like promoting an image that can't be verified
154 2012-10-31 13:07:46 <Icoin> sipa: what do you need for verification ?
155 2012-10-31 13:08:24 <Icoin> sipa: thats why my tendency is for debian wheezy as it realy works flawless
156 2012-10-31 13:09:01 <Icoin> its slow but stable
157 2012-10-31 13:09:51 <sipa> Icoin: you know all bitcoind/-qt vinaries are produced by a system called gitian, which runs the compilation in a verifiable virtual machine, that produces a binary that is bit-for-bit identical for everyone?
158 2012-10-31 13:10:07 <Kakasper> How do you join the Bitcoin Github and make a pull request?
159 2012-10-31 13:10:17 <sipa> we don't release before enough developers signed the binary
160 2012-10-31 13:11:02 <Kakasper> or is the Bitcoin github an exclusive club?
161 2012-10-31 13:11:28 <Icoin> sipa: thanks
162 2012-10-31 13:13:17 <Kakasper> Hello?
163 2012-10-31 13:13:20 <kjj_> grr. anyone with a google wallet want to order something for me in exchange for bitcoins?
164 2012-10-31 13:20:14 <jakenickelson> How do you make a pull request on the Bitcoin Github?
165 2012-10-31 13:20:39 <kjj_> jakenickelson: https://bitcointalk.org/index.php?topic=4571.0
166 2012-10-31 13:30:34 <kjj_> ugh. why is it so hard to find a cheap, available hardware entropy source?
167 2012-10-31 13:31:00 <gmaxwell> kjj_: entropykey isn't terribly expensive
168 2012-10-31 13:31:16 <kjj_> yeah, but I'm sure as hell not going to use google wallet to buy one
169 2012-10-31 13:31:23 <kjj_> heh, no offense. :)
170 2012-10-31 13:33:06 <gmaxwell> kjj_: In the past I wondered if it might be a good idea to buy 10 and sell them for bitcoin, but I was skeptical that I'd get enough buyers to make it worthwhile.
171 2012-10-31 13:33:14 <kjj_> the alternatives seem to be a $200+ serial port dongle or a $80 PCI card that doesn't work in 64 bit machines
172 2012-10-31 13:33:56 <kjj_> gmaxwell: I wish you had. I was thinking the same thing just now
173 2012-10-31 13:35:17 <jasonbournethone> New pull request: https://github.com/bitcoin/bitcoin.org/pull/52
174 2012-10-31 13:35:20 <epscy> what about that website
175 2012-10-31 13:35:23 <epscy> random.org
176 2012-10-31 13:35:26 <epscy> is that no good?
177 2012-10-31 13:35:49 <kjj_> I don't trust random numbers that I don't generate myself
178 2012-10-31 13:37:17 <gmaxwell> epscy: I can't think of any conceivable use for random.org except getting yourself robbed.
179 2012-10-31 13:37:21 <epscy> can't you measure the entropy
180 2012-10-31 13:37:34 <epscy> like how do you trust a hardware RNG
181 2012-10-31 13:37:40 <epscy> gmaxwell: getting robbed?
182 2012-10-31 13:38:48 <gmaxwell> epscy: there have been several poorly designed bitcoin gambling sites that were just taking whatever random numbers that site gave them and generating their results... thus anyone in control of that site (or the network between you and it) could make themselves always a winner.
183 2012-10-31 13:39:03 <epscy> oh i see
184 2012-10-31 13:39:08 <kjj_> epscy: even if the entropy is good, it isn't secret
185 2012-10-31 13:39:28 <epscy> right
186 2012-10-31 13:39:36 <kjj_> with a hardware module, even if the entropy is poor, at least it is secret.
187 2012-10-31 13:39:42 <gmaxwell> epscy: you can't measure the entropy. If you know how the hardware works you can tell if it's probably broken, but if you don't know how it works you can't tell if it's high entopy or not.
188 2012-10-31 13:39:51 <epscy> is it conceivable that the hardware module could phone home
189 2012-10-31 13:40:05 <kjj_> how? telepathy?
190 2012-10-31 13:40:15 <epscy> well i was thinking over the network
191 2012-10-31 13:40:27 <gmaxwell> kjj_: well, not for sure. The device could just be a AES stream generator. :P But it's easy enough to have one taken apart. IIRC the entropy keys had been independantly audited.
192 2012-10-31 13:40:29 <epscy> but i guess your box won't have a network connection
193 2012-10-31 13:41:04 <kjj_> epscy: sure it will, but there isn't any magic in the kernel to route USB-serial or PCI over IP
194 2012-10-31 13:41:26 <epscy> kjj_: doesn't it need a driver?
195 2012-10-31 13:41:40 <epscy> or am showing my lack of kernel knowledge
196 2012-10-31 13:42:24 <kjj_> the USB ones show up as character devices, like a serial port or a keyboard
197 2012-10-31 13:42:47 <EskimoBob> .bitcoin size is 8G ... this shit is getting truly ridiculous. You guys are smart devs, is there any solution for this?
198 2012-10-31 13:42:50 <kjj_> the PCI ones show up as I/O ranges
199 2012-10-31 13:47:09 <Eliel> kjj_: I don't know how good entropy source this is, but it's present in practically every computer. I mean the built in soundcard. Record sound (with or without a microphone) and the least significant bit produced randomness that passed all the tests I had to throw at the randomness.
200 2012-10-31 13:47:41 <kjj_> yeah, I have software for that, but I'm not installing Alsa to use it
201 2012-10-31 13:48:03 <Eliel> I expect it won't need full alsa to work.
202 2012-10-31 13:48:11 <kjj_> it needs Alsa
203 2012-10-31 13:48:28 <sipa> EskimoBob: it should only be 4-5 GiB, and there may be a way for running a fully validating node that doesn't the entire history, but that may take some time still
204 2012-10-31 13:48:31 <Eliel> should not be too challenging to write software to only use the parts of alsa that are needed
205 2012-10-31 13:48:40 <Eliel> and leave the rest out
206 2012-10-31 13:48:59 <sipa> EskimoBob: also, if you can't afford such resources, maybe running a full node is not for you
207 2012-10-31 13:49:00 <epscy> what is wrong with alsa?
208 2012-10-31 13:49:16 <Eliel> kjj_: you could probably condense all that into a kernel module. should be small enough
209 2012-10-31 13:50:39 <EskimoBob> sipa: afford? lol, nothing to do with affording it. It's just absurd size and it keep growing and growing :)
210 2012-10-31 13:50:50 <Eliel> although, I'd probably xor the results with a good quality pseudo-random stream to guard against potential problems :)
211 2012-10-31 13:52:53 <sipa> EskimoBob: well running a full node implies you commit to serving history to those who need it, no way around that, as bitcoin requires people who do so.
212 2012-10-31 13:53:01 <gmaxwell> EskimoBob: It's not "absurd", it's pretty reasonable for a distributed server that supports and autonomously validates a whole currency. If it's not actually problematic for you, why are you complaining? It is problematic for some which is why we care to reduce it.
213 2012-10-31 13:54:28 <kjj_> Eliel: the motherboard's built in sound doesn't seem to be alsa compatible
214 2012-10-31 14:02:48 <kjj_> I'm very tempted to run down to radio shack to buy a couple of pots and wire them up as a voltage divider balanced at the threshold level of a serial pin
215 2012-10-31 14:05:10 <gmaxwell> kjj_: youre very likely to just be measuring 60hz humm or whatever other very periodic RF it picks up.
216 2012-10-31 14:05:32 <kjj_> agreed
217 2012-10-31 14:06:27 <kjj_> LOL. I can use Jgarzik's rngd to feed /dev/urandom back into /dev/random. now my kernel is reporting 3000+ bits of entropy available at all times rather than <200
218 2012-10-31 14:08:04 <gmaxwell> jgarzik's rngd has a certificational flaw.
219 2012-10-31 14:09:51 <kjj_> how so?
220 2012-10-31 14:10:24 <gmaxwell> (its random numbers are not uniform, it'll never produce random numbers that fail the internal fips 140 tests, e.g. long runs of zeros and such.
221 2012-10-31 14:10:27 <gmaxwell> )
222 2012-10-31 14:11:28 <kjj_> I thought that rngd just pulled data from a random source and fed it into the kernel random system
223 2012-10-31 14:12:48 <kjj_> so, the alternatives to the entropykey are all absurdly expensive
224 2012-10-31 14:14:48 <gmaxwell> kjj_: no, it runs the data through the fips 140 tests and then tosses it if it fails. IIRC it doesn't stop or sound an alarm.. just throws out data. Of course, I'm sure it's not a _pratical_ problem, but if you're worrying about using a hardware RNG...
225 2012-10-31 14:15:36 <kjj_> I emailed the entropykey guys to ask if they took any non-google payments
226 2012-10-31 14:16:10 <kjj_> if they'll take a check or direct credit card, I'll buy a 10 pack and resell them on the forums
227 2012-10-31 14:16:30 <kjj_> unless I can track down a google wallet user willing to buy one for me before they reply
228 2012-10-31 14:16:42 <EskimoBob> gmaxwell: I am using the default client and I do not see way around that huge file
229 2012-10-31 14:17:00 <EskimoBob> At least I did not have to download it all at once :)
230 2012-10-31 14:17:11 <sipa> EskimoBob: if you have 8 gigabytes, you did something wrong
231 2012-10-31 14:17:28 <sipa> EskimoBob: like download blk000*.dat files without the index, or used -loadblock on files in the datadir
232 2012-10-31 14:17:30 <EskimoBob> sipa: this is the total size of the catalog
233 2012-10-31 14:18:00 <gmaxwell> (in any entropy gatherer I implemented for non-oss stuff I also used the FIPS test, but if something fails I queue it and xor it with the next block that passes... and also throw nasty errors if the test fails too often)
234 2012-10-31 14:18:13 <sipa> EskimoBob: i believe you when you say that's how much it takes on your machine, but if that is the case, something is wrong
235 2012-10-31 14:18:24 <gmaxwell> EskimoBob: yes, and thats about 2x what it should be, so as sipa says??? you've likely done something to double it.
236 2012-10-31 14:18:50 <sipa> EskimoBob: but apart from that, for now the program needs the entire history
237 2012-10-31 14:18:57 <sipa> future versions will likely change that
238 2012-10-31 14:19:22 <EskimoBob> i hope so
239 2012-10-31 14:19:31 <kjj_> gmaxwell: that sounds like it would be a small patch to rngd
240 2012-10-31 14:19:50 <sipa> EskimoBob: you do realize that implies not everyone will be able to serve history to new nodes?
241 2012-10-31 14:20:07 <sipa> which is a bad thing, but maybe necessary
242 2012-10-31 14:21:10 <EskimoBob> why not prune that crap in regular bases and forward only balances and the full load is only useful if you want to mine
243 2012-10-31 14:21:23 <kjj_> well, for starters, there are no balances
244 2012-10-31 14:21:42 <EskimoBob> i do not know the correct terminology :)
245 2012-10-31 14:21:56 <EskimoBob> but I hope you get what I mean
246 2012-10-31 14:22:31 <sipa> EskimoBob: the reference client follows a zero-trust model, which means it will validate everything it receives from other nodes, and in order to do so, it needs to *see* the entire history - there is no way around that
247 2012-10-31 14:22:36 <gmaxwell> EskimoBobecause if you do that everywhere than miners could freely inflate the coin supply... which defeats the point and purpose of bitcoin.
248 2012-10-31 14:22:59 <sipa> EskimoBob: that doesn't mean it needs to remember the entire history
249 2012-10-31 14:23:28 <gmaxwell> The security model depends on a large number of distinct parties??? including ones without incentives like mining??? having independantly validated the history... they don't need to keep all the data, but they need to have seen it once.
250 2012-10-31 14:23:39 <sipa> but a new node on the network that doesn't trust anyone still needs to get access to this history "yeah yeah guys i know you tell me the balances are X, but i'd like you to prove it to me!"
251 2012-10-31 14:23:57 <EskimoBob> use some hash for that?
252 2012-10-31 14:24:12 <EskimoBob> not the full 100000 lines of transactions ?
253 2012-10-31 14:24:13 <sipa> where would the new node get the hash from?
254 2012-10-31 14:24:23 <sipa> remember, it does not trust anyone
255 2012-10-31 14:24:25 <EskimoBob> miners who have the full 4+ gb
256 2012-10-31 14:24:31 <gmaxwell> oh! of course. Magic! lets just use MAGIC! :P
257 2012-10-31 14:24:46 <EskimoBob> no need to get hysterical gmaxwell :)
258 2012-10-31 14:24:48 <gmaxwell> EskimoBob: then the miners just decide that they're going to award themselves 1000 BTC/block from now on.
259 2012-10-31 14:24:48 <sipa> i sure as hell hope miners aren't the only one who keep history (actually, they don't need it either)
260 2012-10-31 14:25:27 <EskimoBob> forget it. It was more like a question.
261 2012-10-31 14:26:14 <sipa> EskimoBob: that doesn't mean there are no solutions
262 2012-10-31 14:26:22 <sipa> EskimoBob: not every node needs to run zero-trust
263 2012-10-31 14:26:28 <gmaxwell> EskimoBob: We do ave some ideas that allow some more granularity in security levels... but fundimentally there needs to be a widespread base of independant non-mining parties that have validated the whole thing or the security properties are not upheld.
264 2012-10-31 14:27:00 <sipa> in fact, bitcoin was designed so nodes can run with far lower resources if they give up a very small amount of trust
265 2012-10-31 14:27:52 <sipa> this is what for example MultiBit does, and Electrum (will?) do
266 2012-10-31 14:28:11 <EskimoBob> but yes, you can split that file up so noone has the full 4.5 gigs but there are N "complete" files always available in the network... or something like that. mmmm torrent mmm. OK, thanx for the "answer". I'll be waiting for the solution ;)
267 2012-10-31 14:28:17 <gmaxwell> eh, I dunno that SPV is a 'very small' reduction, esp with people accepting single confirmed payments on them. But indeed.
268 2012-10-31 14:28:48 <sipa> gmaxwell: all depends on your attack model, i guess
269 2012-10-31 14:28:57 <EskimoBob> ACTION walks out the door
270 2012-10-31 14:29:20 <gmaxwell> sipa: I guess I'd agree more strongly were it not for hash proxy services.
271 2012-10-31 14:29:29 <sipa> right
272 2012-10-31 14:31:14 <amiller> there isn't yet any opportunistic mining client yet that purports to auction off your hash power to the highest bidder, is there?
273 2012-10-31 14:31:27 <sipa> amiller: gpumax?
274 2012-10-31 14:31:41 <amiller> oh yeah.
275 2012-10-31 14:31:46 <sipa> not sure if it's still running
276 2012-10-31 14:32:08 <sipa> gmaxwell: care to re-test -reindex ?
277 2012-10-31 14:33:06 <gmaxwell> gpumax is going away (or gone) but there are two (? more?) replacements already.
278 2012-10-31 14:33:55 <gmaxwell> sipa: sure.
279 2012-10-31 15:27:04 <jgarzik> ACTION posts about gangman style parody, just for BlueMatt
280 2012-10-31 15:27:09 <jgarzik> *another
281 2012-10-31 15:27:18 <jgarzik> gmaxwell: rngd corrections welcome :)
282 2012-10-31 15:28:43 <kjj_> jgarzik: just how unsafe is it to use /dev/urandom as the input to rngd?
283 2012-10-31 15:29:09 <gmaxwell> s/unsafe/pointless/
284 2012-10-31 15:29:37 <BlueMatt> jgarzik: thanks
285 2012-10-31 15:29:39 <jgarzik> kjj_: <shrug> you are sending PRNG data through FIPS testing, then into a cryptographically hashed mix, then into the entropy pool
286 2012-10-31 15:29:42 <gmaxwell> jgarzik: fix the kernel. a 4kbit random pool is too small, esp now that many kinds of systems only get the 100 bit per second timer input to it. :(
287 2012-10-31 15:29:43 <kjj_> heh. it filled my entropy pool up in a hurry when I tried it. I'm sure not using it to generate keys
288 2012-10-31 15:29:47 <sipa> just take 32 bytes of data from /dev/random, use it as an AES key, and use it in CTR mode
289 2012-10-31 15:30:12 <jgarzik> or as a random seed...
290 2012-10-31 15:30:30 <gmaxwell> kjj_: if you just want to keep your pool filled and don't want to patch the kernel, I suggest "haveged".
291 2012-10-31 15:31:44 <helo> ACTION wonders if gmaxwell is suggesting that kjj_ needs more education
292 2012-10-31 15:32:07 <kjj_> what I really want is a hardware generator, or some other high quality source. but that's not going to happen today, so I'm just playing around
293 2012-10-31 15:33:21 <gmaxwell> I actually had some services where ssh logins were infuriatingly slow, turnes out ssh was reading /dev/urandom and it was blocking because the system was habitually out of entropy. Not wanting to totally ditch urandom (for long term key generation), or patch all these debian packages.. I threw in havaged, solved it nicely.
294 2012-10-31 15:33:56 <gmaxwell> http://mf4.xiph.org/munin/xiph.org/mf4.xiph.org/entropy.html < you can see on this system where it was turned off.
295 2012-10-31 15:34:25 <kjj_> well, I do intend to generate a whole bunch of keys in the near future, and it would be nice if that was fast too
296 2012-10-31 15:35:44 <helo> "shhhhhhhhh" ;)
297 2012-10-31 15:35:48 <kjj_> I just tossed timer_entropyd on the box, and I'm running dieharder on it, just to see. timers seem unlikely to be high quality, in my view
298 2012-10-31 15:36:01 <jgarzik> rngd is entirely for hardware RNG -> kernel entropy pool
299 2012-10-31 15:36:24 <jgarzik> however, it should be expanded to include other sources: audio, video, /proc/* entropy sources
300 2012-10-31 15:36:37 <jgarzik> there are a raft of tiny entropy gathering daemons out there
301 2012-10-31 15:37:25 <jgarzik> most modern computers have TPM, which has an RNG
302 2012-10-31 15:37:57 <kjj_> I'm very annoyed that TPM isn't available as a generic PCI-express card
303 2012-10-31 15:39:41 <gmaxwell> What havaged does well that most daemons do not is that it does a very good job of keeping the tiny kernel pool from going empty.
304 2012-10-31 16:22:37 <aurigae> hi there, anybody awre of a working bitcoind for poolserverj
305 2012-10-31 16:22:39 <aurigae> ??
306 2012-10-31 16:42:23 <gmaxwell> aurigae: does it not work with the latest release?
307 2012-10-31 16:45:59 <aurigae> as i have read not
308 2012-10-31 16:46:39 <aurigae> i tried the luke-jr next eligius release as i run it with eloipool, didnt worked, then the 6.03 or some as i run wit hecoinpool , also didnt worked
309 2012-10-31 16:46:55 <aurigae> https://bitcointalk.org/index.php?PHPSESSID=49be0c3bf4ecb95f4cc32acf382d6e57&topic=33144.140
310 2012-10-31 16:47:02 <aurigae> then i found this post from flowz
311 2012-10-31 16:47:17 <aurigae> no idea how to get this one, and idkk if thats the best solution
312 2012-10-31 16:47:32 <aurigae> would love to have it working with the latest rlease
313 2012-10-31 16:47:49 <aurigae> no some ancient version xD
314 2012-10-31 17:02:32 <sipa> gmaxwell: it actually gets interrupted during the reindex?
315 2012-10-31 17:03:04 <gmaxwell> sipa: looks like it.
316 2012-10-31 17:05:30 <gmaxwell> ah .. interesting:
317 2012-10-31 17:05:31 <gmaxwell> 10/31/12 17:03:44 SetBestChain: new best=0000000000000f054b58 height=131235 work=18626622467183889844 tx=754307 date=06/16/11 11:19:08
318 2012-10-31 17:05:35 <gmaxwell> 10/31/12 17:03:44 ProcessBlock: ACCEPTED
319 2012-10-31 17:05:37 <gmaxwell> 10/31/12 17:03:44 SetBestChain: new best=0000000000000177ca06 height=131236 work=18630389015526577385 tx=754445 date=06/16/11 11:33:32
320 2012-10-31 17:05:41 <gmaxwell> 10/31/12 17:03:44 ProcessBlock: ACCEPTED
321 2012-10-31 17:05:43 <gmaxwell> 10/31/12 17:03:44 ProcessBlock: ORPHAN BLOCK, prev=0000000000000446bb1b
322 2012-10-31 17:05:46 <gmaxwell> 10/31/12 17:03:44 bool LoadExternalBlockFile(FILE*, CDiskBlockPos*)() : Deserialize or I/O error caught during load
323 2012-10-31 17:07:49 <gmaxwell> I don't understand what happening there, it looks like it's contiuning to reindex after that..
324 2012-10-31 17:07:53 <gmaxwell> and later it does it again:
325 2012-10-31 17:07:54 <gmaxwell> 10/31/12 17:07:27 SetBestChain: new best=00000000000009904cc5 height=181531 work=334154981753023227048 tx=3450528 date=05/25/12 13:03:48
326 2012-10-31 17:07:57 <gmaxwell> 10/31/12 17:07:27 ProcessBlock: ACCEPTED
327 2012-10-31 17:08:00 <gmaxwell> 10/31/12 17:07:27 ProcessBlock: ORPHAN BLOCK, prev=0000000000000238643f
328 2012-10-31 17:08:02 <gmaxwell> 10/31/12 17:07:27 ProcessBlock: ORPHAN BLOCK, prev=00000000000005bacafb
329 2012-10-31 17:08:05 <gmaxwell> 10/31/12 17:07:27 bool LoadExternalBlockFile(FILE*, CDiskBlockPos*)() : Deserialize or I/O error caught during load
330 2012-10-31 17:08:11 <aurigae> still looking for a valid bitcoind release for poolserverj
331 2012-10-31 17:08:29 <gmaxwell> and again, and again, the a huge wad of orphans.
332 2012-10-31 17:08:52 <gmaxwell> aurigae: I don't see anything there saying that it won't work with current versions. People are talking about old and silly patches which you shouldn't need.
333 2012-10-31 17:09:07 <kjj_> shit. does anyone know an android app for QR codes that DOESN'T need network access?
334 2012-10-31 17:10:14 <[Tycho]> QR Droid ?
335 2012-10-31 17:10:27 <aurigae> i pulled the latest release from git, ill check it asap
336 2012-10-31 17:10:31 <[Tycho]> May be NeoReader.
337 2012-10-31 17:10:35 <kjj_> requires full network access, even the "private" version
338 2012-10-31 17:10:41 <gmaxwell> aurigae: uh. no. Use bitcoin 0.7.1.
339 2012-10-31 17:10:52 <gmaxwell> kjj_: what is the sound of one hand clapping?
340 2012-10-31 17:10:54 <[Tycho]> Oh, you were talking about the permissions.
341 2012-10-31 17:11:27 <kjj_> gmaxwell: Bart Simpson solved that one years and years ago
342 2012-10-31 17:12:07 <kjj_> what I want is the QR version of my trusty USB barcode scanner. give me the data in the QR code. don't spew it to the internet
343 2012-10-31 17:14:51 <sipa> gmaxwell: the error is probably due to the block not ending at the end of the file, i probably need to put an extra try block around that to quiet the error... you should see one of those for every file
344 2012-10-31 17:14:52 <aurigae> k, ill try 7.1
345 2012-10-31 17:14:58 <aurigae> latest spammed me 404s
346 2012-10-31 17:15:05 <sipa> gmaxwell: the orphans are weird...
347 2012-10-31 17:15:18 <sipa> but they don't come from the network
348 2012-10-31 17:16:29 <sipa> what is weirder is that it continues after processing those orphans
349 2012-10-31 17:16:43 <gmaxwell> What it looks like is what happens is that it continues on a bit longer hits that huge wall of orphans and finally stops.
350 2012-10-31 17:17:21 <gmaxwell> then it starts pulling from the network, at about height=188742.
351 2012-10-31 17:17:41 <gmaxwell> So I was incorrect about it getting interrupted by the network (I was thrown by those inline orphans)
352 2012-10-31 17:18:20 <sipa> but how the hell you can have orphans in your blockchain at all is very strange
353 2012-10-31 17:19:11 <gmaxwell> yea, dunno this node has had previously aborted reindexes but otherwise nothing strange.
354 2012-10-31 17:19:17 <gmaxwell> I dunno how it could continue after hitting one.
355 2012-10-31 17:20:02 <sipa> ah, i need to add something to make reindex continuing after stop/start
356 2012-10-31 17:20:42 <sipa> but an interrupted reindex would explain orphans
357 2012-10-31 17:21:12 <gmaxwell> I'm not quite clear on why it would. it's not rewriting the new blocks.. so an interupted reindex would be like a fresh start, no?
358 2012-10-31 17:21:29 <sipa> it would start downloading blocks
359 2012-10-31 17:21:39 <sipa> those would end up in the existing blocm files
360 2012-10-31 17:21:52 <sipa> after the reindexed part
361 2012-10-31 17:22:03 <sipa> thusly iverwriting perfectly good blocks
362 2012-10-31 17:22:08 <sipa> overwriting
363 2012-10-31 17:23:05 <sipa> if they're a bit off, the pre-allocation would at some part write zeroes over blocks that are not yrt downloaded or reindex
364 2012-10-31 17:23:42 <Luke-Jr> aurigae: I'm not sure PSJ is up to date with Bitcoin, though I did see its author planning to update it in the last week
365 2012-10-31 17:24:06 <Luke-Jr> aurigae: what problems did you have with Eloipool?
366 2012-10-31 17:24:22 <sipa> restart reindex after that, and it would reindex a) the old reindexed part b) the recently diwnloaded blocks, then see a bunch of zero bytes and then c) a bunch of old data, considered orphans because the connecting blocks were overwritten with zeroes
367 2012-10-31 17:40:01 <aurigae> hi luke
368 2012-10-31 17:40:46 <aurigae> with eloipool there is no option to get a frontend up
369 2012-10-31 17:41:18 <aurigae> with bitcoind 7.1 following output on poolj
370 2012-10-31 17:41:22 <aurigae> gD pw: 908a7r8ajjjjas7t9A68dsf87f5df76rSSSd86dc realm: jsonrpc allKeys: [127.0.0.1:8332/, localhost:8332, localhost:8332/, 127.0.0.1:8332]
371 2012-10-31 17:41:23 <aurigae> [18:39:08.215] /
372 2012-10-31 17:43:56 <aurigae> @ luke, if eloi would work with wizkid frontend i would be ???ore than glad to use it
373 2012-10-31 17:45:58 <helo> aurigae: for future reference, #eligius is made for these kinds of questions ;)
374 2012-10-31 17:46:52 <aurigae> this is about poolserverj and a working bitcoind
375 2012-10-31 17:47:03 <aurigae> luke just asked why i dont use eloipool
376 2012-10-31 17:47:36 <aurigae> still need a wroking bitcoind for poolserverj
377 2012-10-31 17:50:59 <D34TH> aurigae, try git head
378 2012-10-31 17:52:08 <aurigae> git heat? how do i do that?, an link or command?
379 2012-10-31 17:52:37 <D34TH> git clone git://github.com/bitcoin/bitcoin.git
380 2012-10-31 17:52:40 <D34TH> cd bitcoin
381 2012-10-31 17:52:44 <D34TH> actually
382 2012-10-31 17:52:48 <D34TH> cd bitcoin/src
383 2012-10-31 17:52:59 <D34TH> make -f makefile.unix USE_UPNP=
384 2012-10-31 17:56:45 <Luke-Jr> aurigae: Eloipool should work with any frontend PSJ does???
385 2012-10-31 18:08:25 <aurigae> the suggested bitcoind also prints 404s, is there nobody operating a poolserverj pool?
386 2012-10-31 18:09:29 <aurigae> @ luke-jr, i have a buddy who understands the database stuff better, he will look into it. For me ,as soo n as manually schema creation is required i have a problem
387 2012-10-31 18:14:06 <D34TH> aurigae, what do you mean prints 404s
388 2012-10-31 18:14:19 <D34TH> what method are you calling
389 2012-10-31 18:14:36 <D34TH> and are you sure its pointed correctly
390 2012-10-31 18:16:26 <aurigae> the error quoted above, the 404s come with the latest 7. releases, 6 releases got some warning that the configuration file is maybe misconfigured. Il try some ancient version but thats not what i want
391 2012-10-31 18:16:55 <aurigae> however, i have a buddy looking into eloipool and a possible frontend solution again
392 2012-10-31 18:28:48 <jgarzik> anybody with big endian Linux, willing to run a quick test?
393 2012-10-31 18:28:51 <jgarzik> git clone git://github.com/jgarzik/picocoin.git && cd picocoin && ./autogen.sh && ./configure && make -s all check
394 2012-10-31 18:29:09 <jgarzik> ACTION is curious if all tests pass, on big endian
395 2012-10-31 18:29:10 <Luke-Jr> aurigae: I don't think anyone runs PSJ, since PSJ doesn't make good blocks even anymore
396 2012-10-31 18:33:34 <D34TH> what is picocoin
397 2012-10-31 18:33:35 <D34TH> DL
398 2012-10-31 18:36:00 <D34TH> oh
399 2012-10-31 18:36:02 <D34TH> thats nice
400 2012-10-31 18:36:39 <BlueMatt> its jgarzik's second alt client in 6 months
401 2012-10-31 18:38:12 <gmaxwell> jgarzik: testing
402 2012-10-31 18:42:18 <gmaxwell> All 8 tests passed
403 2012-10-31 18:42:34 <jgarzik> woot
404 2012-10-31 18:42:38 <gmaxwell> Note: autoconf didn't mandate the jansson dep but it didn't build without it.
405 2012-10-31 18:42:56 <gmaxwell> also unsigned int salt[] = { 4185398345, 2729682459 };
406 2012-10-31 18:43:00 <gmaxwell> could use a U
407 2012-10-31 18:43:07 <gmaxwell> cause thats not an unsigned constant.
408 2012-10-31 18:45:34 <jgarzik> gmaxwell: updated
409 2012-10-31 18:47:52 <sipa> ah the joy of an own project where you can push "upstream" at will :)
410 2012-10-31 18:48:16 <jgarzik> ;p
411 2012-10-31 18:48:38 <Luke-Jr> sipa: I prefer having others review the code :P
412 2012-10-31 18:48:49 <sipa> Luke-Jr: details
413 2012-10-31 18:48:50 <sipa> !
414 2012-10-31 18:48:53 <gmaxwell> There is a time and place for everything. :P
415 2012-10-31 18:49:27 <gmaxwell> jgarzik: missed another one on line 125.
416 2012-10-31 18:49:30 <sipa> now, let's see if -reindex continues nicely after a restart...
417 2012-10-31 18:49:39 <jgarzik> I prefer to drive at 100 m.p.h., and let brick walls and blind curves surprise me. Life is more exciting that way.
418 2012-10-31 18:51:06 <jgarzik> Of course, that only works with an open source, peer reviewed car.
419 2012-10-31 18:51:12 <jgarzik> (and route)
420 2012-10-31 18:51:19 <sipa> which nobody is driving yet
421 2012-10-31 18:51:32 <jgarzik> especially that
422 2012-10-31 18:51:53 <jgarzik> Life is great when you have no users to piss off.
423 2012-10-31 19:17:40 <sipa> oh damn... i was wondering how the -reindex implementation could have slowed down IBD that much
424 2012-10-31 19:17:48 <sipa> turns out my laptop was unplugged
425 2012-10-31 19:23:30 <gmaxwell> ;;bc,blocks
426 2012-10-31 19:23:31 <gribble> 201949
427 2012-10-31 19:23:42 <gmaxwell> uhh
428 2012-10-31 19:24:14 <sipa> seems legit
429 2012-10-31 19:24:32 <gmaxwell> https://blockexplorer.com/q/getblockcount < yea .. but the front page of blockexplorer.com is right
430 2012-10-31 19:24:47 <gmaxwell> pretty fishy
431 2012-10-31 19:27:43 <gmaxwell> uhhh
432 2012-10-31 19:27:44 <gmaxwell> 10/31/12 20:27:16 Reindexing block file blk00000.dat...
433 2012-10-31 19:28:02 <gmaxwell> yea, current patch not so good. :P
434 2012-10-31 19:28:37 <gmaxwell> hah
435 2012-10-31 19:28:38 <gmaxwell> "blocks" : -1,
436 2012-10-31 19:28:48 <gmaxwell> on the plus side, it's producing those errors _really fast_.
437 2012-10-31 19:31:09 <jgarzik> ;p
438 2012-10-31 19:35:29 <D34TH> wait a minute
439 2012-10-31 19:35:35 <D34TH> i have to redownload the chain?
440 2012-10-31 19:35:36 <D34TH> wat
441 2012-10-31 19:36:20 <gmaxwell> D34TH: What are you asking about?
442 2012-10-31 19:36:59 <D34TH> just upgraded to 127 from 109 (bluematt's jenkins) making me redownload chain to new format
443 2012-10-31 19:37:25 <gmaxwell> 127, 109 what are these numbers?
444 2012-10-31 19:37:49 <gmaxwell> oh git patche since figures? those are mostly useless.
445 2012-10-31 19:38:19 <D34TH> basically from oct 14 to oct 31
446 2012-10-31 19:38:28 <D34TH> need to redownload chain
447 2012-10-31 19:39:07 <gmaxwell> uh yea, that would be across the ultraprune merge. Autoupgrade hasn't been implemented yet.
448 2012-10-31 19:39:20 <D34TH> i just did loadblocks
449 2012-10-31 19:39:24 <D34TH> no biggie
450 2012-10-31 19:39:42 <D34TH> i found out by accident that you can chain -loadblock for each file and it loads them all
451 2012-10-31 19:39:48 <gmaxwell> Don't enter the hardhat area if you can't tolerate a little dust. :P https://bitcointalk.org/index.php?topic=119525.0
452 2012-10-31 19:39:49 <sipa> gmaxwell: uhh, is that with current patch?
453 2012-10-31 19:39:55 <gmaxwell> sipa: yep.
454 2012-10-31 19:40:19 <gmaxwell> sipa: it finally finished resyncing after the last try, I switched to the new patch and it instantly failed.
455 2012-10-31 19:40:38 <sipa> doh, worked fine here
456 2012-10-31 19:40:42 <sipa> will test more
457 2012-10-31 19:44:14 <D34TH> gmaxwell, http://i.qkme.me/3rl53a.jpg
458 2012-10-31 19:46:51 <sipa> gmaxwell: was that with or without -reindex?
459 2012-10-31 19:47:15 <gmaxwell> with.
460 2012-10-31 19:47:34 <gmaxwell> chain is trashed too, pulling a clean copy without the patch now.
461 2012-10-31 19:47:34 <sipa> D34TH: you should know better than to pick up a duck in a dungeon
462 2012-10-31 19:47:54 <D34TH> sipa: it was +10 INT
463 2012-10-31 19:48:02 <D34TH> couldn't resist
464 2012-10-31 19:49:13 <sipa> gmaxwell: there may be a problem when the genesis block is not the first block in blk00000.dat... no reason why it wouldn't be, though
465 2012-10-31 20:03:54 <gmaxwell> I'm beginning to suspect that linus uses google+ specifically because it gives up a huge audience of idiots so that he can feel maximally smart while making fun of them.
466 2012-10-31 20:04:05 <gmaxwell> er, didn't mean to send that to this channel. :P
467 2012-10-31 20:04:23 <Diablo-D3> gmaxwell: um
468 2012-10-31 20:04:29 <Diablo-D3> huh.
469 2012-10-31 20:04:44 <Diablo-D3> gmaxwell: would it be wrong if I got a g+ account expressly for that purpose?
470 2012-10-31 20:05:56 <gmaxwell> [context being] https://plus.google.com/+LinusTorvalds/posts/ByVPmsSeSEG
471 2012-10-31 20:06:08 <Diablo-D3> let me guess, the monitor one?\\
472 2012-10-31 20:06:09 <DBordello> Any way to determine when a transaction was signed?
473 2012-10-31 20:06:18 <Diablo-D3> HA HA knew it
474 2012-10-31 20:06:27 <Diablo-D3> gmaxwell: honestly, I like his thinking on that
475 2012-10-31 20:06:30 <Diablo-D3> gmaxwell: but hes got the wrong res
476 2012-10-31 20:06:51 <Diablo-D3> should be 3840xwhatever
477 2012-10-31 20:07:00 <gmaxwell> Diablo-D3: oh sure, I agree. but if you go into the comments there is the inevitable trickle of morons arguing with him.
478 2012-10-31 20:07:16 <Diablo-D3> yeah, I noticed that earlier
479 2012-10-31 20:07:18 <Diablo-D3> its kinda weird =/
480 2012-10-31 20:07:39 <Diablo-D3> I need to bug him to do another linux desktops suck dick rant
481 2012-10-31 20:07:53 <Diablo-D3> like, for shits and giggles, I set my laptop dpi to what it actually is (103)
482 2012-10-31 20:07:56 <Diablo-D3> fonts get bigger, sure
483 2012-10-31 20:08:06 <Diablo-D3> everything else doesnt.
484 2012-10-31 20:09:17 <Diablo-D3> I almost want to try to out-OSX OSX
485 2012-10-31 20:09:24 <Diablo-D3> and create my own desktop environment
486 2012-10-31 20:09:36 <Diablo-D3> but goddamn thats going to take a loooooot of work
487 2012-10-31 20:10:27 <Diablo-D3> gmaxwell: oh, and btw, I think Im finally just going to create my own language
488 2012-10-31 20:10:36 <Diablo-D3> its going to be a language in which everything is done as wrong as possible
489 2012-10-31 20:11:30 <Diablo-D3> gmaxwell: but good news
490 2012-10-31 20:11:35 <Diablo-D3> it will run on massively parallel clusters.
491 2012-10-31 20:18:27 <kjj_> I thought someone already did a parallel Intercal
492 2012-10-31 20:20:06 <MC1984> anyone looked at the parallela thing for bitcoin?
493 2012-10-31 20:20:20 <D34TH> i wonder if parallela could support OpenCL
494 2012-10-31 20:20:20 <MC1984> 64 risc cores or something
495 2012-10-31 20:20:24 <kjj_> yup. there's even a whole thread on it
496 2012-10-31 20:20:33 <kjj_> doesn't look useful for mining
497 2012-10-31 20:20:37 <MC1984> link?
498 2012-10-31 20:20:50 <D34TH> kjj_ OpenCL can do more things than mine
499 2012-10-31 20:21:41 <kjj_> https://bitcointalk.org/index.php?topic=115021.0;all
500 2012-10-31 20:22:56 <kjj_> it can probably do ECDSA
501 2012-10-31 20:39:00 <helo> ACTION can do ECDSA too
502 2012-10-31 20:45:06 <MC1984> thx
503 2012-10-31 20:47:27 <MC1984> didnt it turn out that litecoin was a scam seeing as the way they implemented scrypt meant that you can GPU mine it after all and someone had probably been doing so in secret
504 2012-10-31 20:48:05 <TD> woo
505 2012-10-31 20:48:11 <helo> i think it's still maintaining an exchange value, so it's ~alive
506 2012-10-31 20:48:11 <TD> google hard down
507 2012-10-31 20:48:27 <[Tycho]> at least gmail is up
508 2012-10-31 20:48:30 <kjj_> I'm not sure that rises to the level of scam
509 2012-10-31 20:48:44 <helo> gmail is down for everyone i know
510 2012-10-31 20:48:58 <helo> i shouldn't say that... "down for everyone who i've asked"
511 2012-10-31 20:49:00 <kjj_> It still requires a lot more memory than SHA256 does, but not as much as it *could*
512 2012-10-31 20:49:07 <kjj_> or even should
513 2012-10-31 20:49:28 <[Tycho]> Just tried browsing some folders on gmail, seems normal.
514 2012-10-31 20:49:31 <MC1984> thats what i mean
515 2012-10-31 20:50:06 <MC1984> it was fudged so that someone could gpu mine it in secret and no one found out for over a year right
516 2012-10-31 20:50:32 <[Tycho]> Isn't LTC already GPU-mined ?
517 2012-10-31 20:50:36 <TD> it went down briefly. seems back now
518 2012-10-31 20:50:40 <helo> yeah
519 2012-10-31 20:50:40 <MC1984> so much for open source
520 2012-10-31 20:50:56 <forsetifox_> Heh.
521 2012-10-31 20:51:04 <kjj_> I don't think it was a secret. I just don't think that anyone that understood scrypt parameters cared enough to look at the source
522 2012-10-31 20:51:27 <helo> MC1984: nobody knows that someone was gpu mining it during that time
523 2012-10-31 20:51:41 <helo> well, it's at least not widely known
524 2012-10-31 20:51:56 <MC1984> i thought it turned out someone probably was?
525 2012-10-31 20:52:07 <kjj_> and it isn't like GPUs don't have access to memory. cranking it up would just reduce the number of threads possible on a given card
526 2012-10-31 20:52:37 <forsetifox_> So what. If someone has the knowledge of code/ bitcoins/ electronics go for it. Who cares if one person mined it faster than the others.
527 2012-10-31 20:53:00 <MC1984> its a shame litecoin might have had some legs
528 2012-10-31 20:53:04 <kjj_> SHA runs entirely in registers and cache. the hit going to memory is enormous, even for small-ish amounts of memory
529 2012-10-31 20:53:24 <MC1984> might have even resisted ASIC given large enough memory requirements
530 2012-10-31 20:54:13 <helo> if we were still on CPU mining, botnets would be a bigger threat, i think
531 2012-10-31 20:54:51 <helo> 10% of hashrate is due to one CPU-mining botnet currently
532 2012-10-31 20:54:51 <kjj_> helo: bingo.
533 2012-10-31 20:55:11 <helo> so that could be easily >50% if we hadn't moved up an order of magnitude
534 2012-10-31 20:56:25 <helo> kind of a terrifyingly narrow successful path bitcoin seems to be treading
535 2012-10-31 20:56:41 <MC1984> indeed
536 2012-10-31 20:57:12 <MC1984> if it fails i highly doubt it could ever be restarted
537 2012-10-31 20:58:01 <TD> helo: how do you know that?
538 2012-10-31 20:59:39 <helo> TD: the 10% figure from http://threatpost.com/en_us/blogs/zeroaccess-botnet-cashing-click-fraud-and-bitcoin-mining-103012
539 2012-10-31 21:00:15 <helo> "Sophos estimated that ZeroAccess could be earning as much as $2.7 million annually via Bitcoin mining."
540 2012-10-31 21:00:33 <helo> at the current exchange rate, that's ~10%
541 2012-10-31 21:00:48 <TD_> ah
542 2012-10-31 21:00:51 <TD_> but how did they estimate that
543 2012-10-31 21:01:07 <helo> i got 2.4Mhash/s per node
544 2012-10-31 21:01:21 <helo> based on half of the 2,000,000 infected nodes they say mine
545 2012-10-31 21:01:31 <helo> seems pretty reasonable for cpu mining
546 2012-10-31 21:17:53 <sipa> gmaxwell: does reindex work on a cleanly downloaded chain?
547 2012-10-31 21:18:41 <ibno> where can I find a list over which transaction types are considered standard?
548 2012-10-31 21:19:11 <gmaxwell> sipa: going to try that next.
549 2012-10-31 21:32:16 <jgarzik> ibno: source code
550 2012-10-31 21:32:58 <jgarzik> ibno: Solver()'s mTemplates
551 2012-10-31 21:33:45 <ibno> jgarzik: thanks, I found what I was looking for
552 2012-10-31 22:49:46 <gmaxwell> sipa: seems to work fine for me on a clean chain, unfortunately I was stupid and didn't save the database it caught fire on.
553 2012-10-31 22:50:56 <MC1984> anyone notice the bitcoin scene is sort of stagnating
554 2012-10-31 22:51:11 <MC1984> nothings happening
555 2012-10-31 22:51:31 <MC1984> prices are down, no interesting new startups
556 2012-10-31 22:51:41 <MC1984> maybe the critical mass of scamming is taking its toll
557 2012-10-31 22:52:13 <conman> scams drive economy too
558 2012-10-31 22:52:15 <MC1984> thats ok, just keep plugging on the technical side until people are ready to play nice
559 2012-10-31 22:53:21 <MC1984> maybe its best that the scene cool down for a bit, so that it doesnt run away before the technical basis is ready for it
560 2012-10-31 22:53:50 <conman> we have enough on the mining side to contend with shortly...
561 2012-10-31 22:54:09 <phantomcircuit> MC1984, the scammers have taken over far more than is obvious
562 2012-10-31 22:54:13 <phantomcircuit> they're insidious
563 2012-10-31 22:54:38 <MC1984> yea
564 2012-10-31 22:55:02 <sipa> gmaxwell: i assume a buggy version of reindex overwrote the genesis block in your chain, and placed a new genesis block elsewhere, and eventually reconnected
565 2012-10-31 22:55:15 <gmaxwell> hah
566 2012-10-31 22:55:50 <gmaxwell> reindex should be a little more robust against that kind of corruption, though I'm not sure abou testing it.
567 2012-10-31 22:57:15 <MC1984> people STILL need to earn how not to be taken in though thats the main problem
568 2012-10-31 22:57:22 <MC1984> scammers gonna scam
569 2012-10-31 22:57:37 <sipa> well, what i assume would have happened if you continued the reindex, was that it would process all block files, find nothing connectable, and start downloadinging, overwriting your files
570 2012-10-31 22:57:54 <sipa> perhaps it should delete them, but i feel less easy about deleting than about overwriting
571 2012-10-31 22:58:02 <MC1984> its a harsh tutelege
572 2012-10-31 23:00:42 <phantomcircuit> MC1984, im not talking about the small time scammers in -otc :/
573 2012-10-31 23:00:47 <phantomcircuit> (and no i wont elaborate)
574 2012-10-31 23:01:29 <MC1984> yea yea pirate etc
575 2012-10-31 23:01:41 <phantomcircuit> not even talkgn about pirate
576 2012-10-31 23:02:10 <MC1984> well thts as far as my knowledge goes
577 2012-10-31 23:02:28 <MC1984> bet theres some bitcoin startup investment scamming going on
578 2012-10-31 23:03:18 <gmaxwell> sipa: I was able to break it again.
579 2012-10-31 23:03:44 <sipa> uh-oh
580 2012-10-31 23:03:49 <sipa> how?
581 2012-10-31 23:04:50 <gmaxwell> sequence: Sync a clean node normally. Reindex. Abort. Start without reindex and let it download a couple blocks. Stop. Reindex again. Blows up again at 0.
582 2012-10-31 23:05:22 <gmaxwell> Interestingly I aborted again (while it was spewing those errors) and restarted again with reindex and it appears to be reindexing correctly now.
583 2012-10-31 23:06:20 <sipa> what does 'abort' mean?
584 2012-10-31 23:08:26 <gmaxwell> kill -TERM
585 2012-10-31 23:08:36 <gmaxwell> (and it closed itself cleanly)
586 2012-10-31 23:08:46 <sipa> during reindex, or after?
587 2012-10-31 23:09:47 <gmaxwell> during.
588 2012-10-31 23:09:52 <gmaxwell> sorry, that wasn't clear.
589 2012-10-31 23:09:57 <gmaxwell> I aborted at about height 200k.
590 2012-10-31 23:10:53 <gmaxwell> then rainbitcoin without reindex and let it get a dozen blocks or so, aborted, and started again with reindex and it blew up. Aborted. Started again with reindex. And it's up to 191k now.
591 2012-10-31 23:14:28 <sipa> gmaxwell: is download using -connect=127.0.0.1 or so?
592 2012-10-31 23:15:56 <gmaxwell> yes.
593 2012-10-31 23:16:27 <sipa> maybe IBD started before the reindex initiated
594 2012-10-31 23:17:25 <gmaxwell> ah, if so that I should be able to keep retrying.
595 2012-10-31 23:17:49 <gmaxwell> (and trigger it on and off)
596 2012-10-31 23:20:29 <sipa> heh
597 2012-10-31 23:20:39 <gmaxwell> restart .. restart ... restart.. boom... boom... ... hm. is it going to work at all anymore?
598 2012-10-31 23:20:39 <sipa> oh, wrong branch
599 2012-10-31 23:20:45 <gmaxwell> it doesn't look like that:
600 2012-10-31 23:20:47 <gmaxwell> 11/01/12 00:19:22 nFileVersion = 79900
601 2012-10-31 23:20:49 <gmaxwell> 11/01/12 00:19:22 ERROR: CheckBlock() : size limits failed
602 2012-10-31 23:20:52 <gmaxwell> 11/01/12 00:19:22 ERROR: ProcessBlock() : CheckBlock FAILED
603 2012-10-31 23:21:29 <gmaxwell> oh tere it goes .. so yea out of six starts two of them fail.
604 2012-10-31 23:22:54 <sipa> update pushed
605 2012-10-31 23:24:44 <sipa> assuming that was the problem, this should fix it
606 2012-10-31 23:28:47 <Joric> what's // NOLINT ?
607 2012-10-31 23:29:07 <vazakl> disables lint checking
608 2012-10-31 23:30:56 <Joric> ah got it, it's for the code analyzer
609 2012-10-31 23:31:05 <gmaxwell> sipa: ah, first run with new patch failed.
610 2012-10-31 23:31:14 <sipa> grr!
611 2012-10-31 23:31:49 <gmaxwell> as did the second, lemme see if it will ever work again. :P
612 2012-10-31 23:32:06 <gmaxwell> Third time worked.
613 2012-10-31 23:32:18 <Joric> somebody should perform some static analysis on bitcoin source code
614 2012-10-31 23:32:24 <sipa> gmaxwell: anything special on the command line?
615 2012-10-31 23:32:26 <sipa> or config?
616 2012-10-31 23:33:54 <gmaxwell> the logs look indentical in the pass and fail cases except for the whole passing and failing part.
617 2012-10-31 23:33:59 <gmaxwell> Joric: people do.
618 2012-10-31 23:35:58 <sipa> gmaxwell: try under valgrind, maybe? i don't like this indeterminism
619 2012-10-31 23:36:05 <gmaxwell> sipa: -detachdb -daemon=1 -reindex plus http://pastebin.com/h8FsfmEs
620 2012-10-31 23:36:59 <gmaxwell> yea already ahead of you...
621 2012-10-31 23:37:00 <gmaxwell> slooow
622 2012-10-31 23:37:33 <gmaxwell> oh there we go
623 2012-10-31 23:37:34 <gmaxwell> ==30545== Conditional jump or move depends on uninitialised value(s)
624 2012-10-31 23:37:37 <gmaxwell> ==30545== by 0x43F8F8: ThreadImport(void*) (init.cpp:353)
625 2012-10-31 23:40:42 <sipa> same here; compiling in debug mode
626 2012-10-31 23:55:40 <sipa> gmaxwell: got it
627 2012-10-31 23:55:54 <sipa> nType and nVersion not initialized in CBufferedFile