1 2012-07-05 00:05:52 <jrmithdobbs> sipa: i've confirmed just setting nsize to sizeof()-1 works with tor .2.2.37, .2.3.18-rc, and openssh -D, done partial syncs over all 3
  2 2012-07-05 00:07:50 <BlueMatt> gmaxwell: been looking at these results some more, a huge portion of the speed increases appear to be the inlining (and lto) of quite a bit of stdc++, the inlinging (and lto) of quite a bit of boost, and the inlining (and lto) of quite a bit of our serialization
  3 2012-07-05 00:08:21 <BlueMatt> gmaxwell: potential things that may provide optimization: special-casing some of the serialization code for often-serialized objects like txes and maybe blocks
  4 2012-07-05 00:08:39 <gmaxwell> But you were LTO on gcc too, weren't you?
  5 2012-07-05 00:08:57 <jrmithdobbs> BlueMatt: so does bitcoin compile clean with clang now?
  6 2012-07-05 00:09:14 <jrmithdobbs> BlueMatt: is that what prompted all the warning/etc fixes? ;p
  7 2012-07-05 00:09:31 <BlueMatt> gmaxwell: yes
  8 2012-07-05 00:09:40 <gmaxwell> BlueMatt: what gcc version?
  9 2012-07-05 00:09:57 <BlueMatt> jrmithdobbs: close, there is still one in json_spirit that is pretty non-clear
 10 2012-07-05 00:09:59 <jrmithdobbs> gmaxwell: doesn't matter that's par for gcc code on clang vs gcc
 11 2012-07-05 00:10:00 <BlueMatt> gmaxwell: 4.7
 12 2012-07-05 00:10:08 <gmaxwell> interesting.
 13 2012-07-05 00:10:09 <BlueMatt> and clang 3.0
 14 2012-07-05 00:10:12 <jrmithdobbs> gmaxwell: err c++ not gcc code
 15 2012-07-05 00:11:11 <jrmithdobbs> BlueMatt: good work.
 16 2012-07-05 00:11:15 <jrmithdobbs> that's awesome
 17 2012-07-05 00:11:44 <BlueMatt> I just cleaned up the last two, the real work was elsewhere
 18 2012-07-05 00:11:54 <BlueMatt> I believe jgarzk did a lot of it
 19 2012-07-05 00:13:18 <BlueMatt> clang does quite a wonder on BOOST_FOREACH...
 20 2012-07-05 00:14:35 <BlueMatt> also BOOST_FOREACH/BOOST_REVERSE_FOREACH is absolutely aweful performance-wise
 21 2012-07-05 00:14:50 <BlueMatt> like...wtf is it doing?
 22 2012-07-05 00:21:59 <jrmithdobbs> i don't get boost at all tbqh
 23 2012-07-05 00:22:09 <luke-jr> BlueMatt: probably better to fix boost than reinvent it
 24 2012-07-05 00:22:27 <BlueMatt> jrmithdobbs: it has some very useful libraries, esp for cross-platform compatibility
 25 2012-07-05 00:22:31 <gmaxwell> it does a lot of boilerplate stuff. Not _well_... perhaps ...but if you want to drive the minutia of your machine you use C not C++.
 26 2012-07-05 00:23:12 <BlueMatt> luke-jr: yep, but telling the boost guys to sped up _FOREACH isnt gonna work, methinks
 27 2012-07-05 00:23:32 <BlueMatt> gmaxwell: yep...
 28 2012-07-05 00:24:01 <luke-jr> BlueMatt: sendign them a patch might
 29 2012-07-05 00:24:19 <BlueMatt> Id think to do that you'd end up just removing a ton of features, and they wont like that, was my point
 30 2012-07-05 00:25:03 <jgarzik> sending boost people a patch is useful to the world, but likely not to bitcoin, which likely cannot mandate boost version X, where X==very recent.
 31 2012-07-05 00:25:09 <BlueMatt> but, really: clang BOOST_REVERSE_FOREACH 78 hits on that line: 21m cycles, gcc: 78 hits, 132m on that line and >1b in boost::foreach_detail
 32 2012-07-05 00:25:39 <BlueMatt> like...wtf? and wtf is clang doing so well?
 33 2012-07-05 00:39:48 <BlueMatt> anyway, Im gonna dig deeper tomorrow, these results are definitely interesting, gnight all
 34 2012-07-05 00:42:38 <jrmithdobbs> wow that is impressive even for clang
 35 2012-07-05 00:51:52 <Keverw> I feel like a noob but I've installed Bitcoin-QT on my Ubuntu system. Any ideas how I edit the config? I went to the finder thing on it and went to '/usr/bin/bitcoin-qt' but it ended up starting it
 36 2012-07-05 00:54:25 <gmaxwell> to edit the config you make a file in your home directory .bitcoin/bitcoin.conf
 37 2012-07-05 00:54:27 <Keverw> Is that similar to the .app files on Mac? Like I don't see a way to right click it and view it's contents.
 38 2012-07-05 00:54:34 <Keverw> oh& okay.
 39 2012-07-05 00:54:48 <Keverw> It will read that insead of the one it comes with?
 40 2012-07-05 00:54:50 <gmaxwell> It's just a program. Nowhere but on mac are settings inside the program object. :)
 41 2012-07-05 00:55:14 <gmaxwell> Yes... well there isn't one it comes with.
 42 2012-07-05 00:55:16 <Keverw> so where I says 'test user", that's my home?
 43 2012-07-05 00:55:45 <Keverw> oh&. .bitcoin already exists and asks if I want to over write it.
 44 2012-07-05 00:55:46 <Keverw> 
 45 2012-07-05 00:55:54 <gmaxwell> /home/testuser/.bitcoin/bitcoin.conf or the like also ~/  is a shortcut for /home/yourusername/
 46 2012-07-05 00:55:54 <Keverw> Looks like it's a hidden folder. Hmm. Let me google on how to show those
 47 2012-07-05 00:56:03 <gmaxwell> .bitcoin does because you've run it, yes.
 48 2012-07-05 00:57:03 <Keverw> can I delete everything from the folder then?
 49 2012-07-05 00:58:28 <gmaxwell> you can, assuming you don't have a wallet you care about.
 50 2012-07-05 00:58:41 <gmaxwell> But why? there shouldn't be a bitcoin.conf file there yet.
 51 2012-07-05 00:58:48 <Keverw> nope. A fresh install in a VM. There isn't
 52 2012-07-05 00:59:18 <Keverw> just some .dat files and stuff. but I deleted everything and created a 'bitcoin.conf' I copied from the wiki
 53 2012-07-05 00:59:50 <gmaxwell> ah.. er. you probably don't want to do that. The example on the wiki flips a lot of random switches to just show you that they're there.
 54 2012-07-05 00:59:57 <Keverw> and grr& I hate auto and love autocorrect on the Mac. I mean on iOS it's okay but like when I say bitcoin.conf it likes to add a space& Well now it's not since I edited it last time it did that. Guess it learns.
 55 2012-07-05 01:00:10 <Keverw> oh& Most of it is commented ouyt.
 56 2012-07-05 01:00:15 <Keverw> I just uncommented the test net line
 57 2012-07-05 01:00:24 <gmaxwell> okay good. comment out everything you don't know you need. :)
 58 2012-07-05 01:00:45 <Keverw> Okay. It looks good after skimming over it.
 59 2012-07-05 01:02:24 <Keverw> test net has like 57,000 compared to 187,500ish. Just compared the blocks with my production client. So I can confirm I'm on test net!
 60 2012-07-05 01:03:18 <Keverw> muAmesPFUaiR9KfXXiSMZdrjtFSYcaNqFk is the address.
 61 2012-07-05 01:03:52 <Keverw> Then once it downloads, I'm gonna search on sliming down Ubuntu some more, and then make 2 copies. one for a second user, and a developer wallet!
 62 2012-07-05 01:04:26 <Keverw> but I was chatting with some people on Ubuntu and they gave me a link to a mini version of Ubuntu which runs way faster and has less stuff by default!
 63 2012-07-05 01:21:30 <jgarzik> "BOOST_FOREACH is designed for ease-of-use and efficiency. It does no dynamic allocations, makes no virtual function calls or calls through function pointers, and makes no calls that are not transparent to the compiler's optimizer. This results in near-optimal code generation"
 64 2012-07-05 01:21:32 <jgarzik> :)
 65 2012-07-05 01:31:13 <D34TH> keverw
 66 2012-07-05 01:31:40 <D34TH> ubuntu-minimal
 67 2012-07-05 01:36:09 <Keverw> yep. That's it.
 68 2012-07-05 01:36:22 <Keverw> 2.6GB it says when I run df -h, gonna Google to see if I can make it any smaller
 69 2012-07-05 01:36:55 <gmaxwell> 2.6gb? that sounds like you have a copy of non-testnet there.
 70 2012-07-05 01:37:08 <Keverw> no& the whole OS
 71 2012-07-05 01:37:23 <gmaxwell> ahh okay
 72 2012-07-05 01:37:33 <gmaxwell> jgarzik: They did not say how near.
 73 2012-07-05 01:38:06 <Keverw> 19 items totaling 144.9
 74 2012-07-05 01:38:12 <Keverw> it says for the .bitcoin folder.
 75 2012-07-05 02:22:59 <[Tycho]> Wow, someone with 700+ Gh/s in my pool...
 76 2012-07-05 02:23:07 <[Tycho]> GPUmax is surprising sometimes :)
 77 2012-07-05 02:40:27 <Keverw> Does TestNet run a little slower then the normal network?
 78 2012-07-05 02:42:33 <freewil> yes there is usually a much larger gap between blocks
 79 2012-07-05 02:42:52 <Keverw> oh dang&.
 80 2012-07-05 02:43:25 <Keverw> Is there a faster way or am I stuck waiting?
 81 2012-07-05 02:43:33 <freewil> it's really easy to mine on it if you have a miner
 82 2012-07-05 02:43:53 <Keverw> I have a Nvida GeForce 9300M GS
 83 2012-07-05 02:43:55 <freewil> you can also startup your own testnet
 84 2012-07-05 02:44:01 <Keverw> Would that be faster?
 85 2012-07-05 02:44:10 <freewil> i would recommend that
 86 2012-07-05 02:44:24 <Keverw> To mine or install my own testnet?
 87 2012-07-05 02:44:28 <freewil> own testnet
 88 2012-07-05 02:44:40 <freewil> you basically just start two nodes and thats your own private testnet
 89 2012-07-05 02:44:54 <Keverw> Oh& Can my friends join it?
 90 2012-07-05 02:45:03 <Keverw> They are testing the app with me
 91 2012-07-05 02:45:13 <freewil> mmm... probably just be easier to use the real testnet then
 92 2012-07-05 02:45:29 <freewil> but you could just open your ports and they could join
 93 2012-07-05 02:45:37 <freewil> if you wanted to do your own
 94 2012-07-05 02:45:45 <Keverw> I rather just use the normal testnet
 95 2012-07-05 02:45:59 <Keverw> if I mine, would it help speed it up?
 96 2012-07-05 02:46:06 <freewil> for sure
 97 2012-07-05 02:46:23 <Keverw> Oh& so how do I mine?
 98 2012-07-05 02:46:48 <Keverw> Edit the config file on my Ubutnu VM and then have a miner on my PC connect to it?
 99 2012-07-05 02:47:12 <freewil> yeah
100 2012-07-05 02:47:23 <Keverw> solo or pool?
101 2012-07-05 02:47:28 <freewil> solo
102 2012-07-05 02:48:17 <Keverw> oh& the difficulty is at 1 still& cool
103 2012-07-05 02:48:22 <Keverw> It stays at 1?
104 2012-07-05 02:48:27 <freewil> yeah i think so
105 2012-07-05 02:48:39 <Keverw> oh& so my weak GPU would have a party.
106 2012-07-05 02:48:48 <freewil> yeah thats the point
107 2012-07-05 02:48:54 <freewil> i think testnet is hardcoded to 1
108 2012-07-05 02:49:28 <Keverw> Should I use CPU or GPU?
109 2012-07-05 02:49:51 <freewil> i would just use the gpu
110 2012-07-05 02:50:01 <Keverw> okay. Need to Google how to set it up
111 2012-07-05 02:50:24 <freewil> yeah you should be able to figure it out
112 2012-07-05 02:50:30 <Keverw> What do I do? Set up RPC and contact the miner to it?
113 2012-07-05 02:50:40 <freewil> you basically just need to start up bitcoin on testnet
114 2012-07-05 02:50:59 <freewil> then have your miner connect to the rpc host port and specify the correct rpc username and password
115 2012-07-05 02:52:23 <Keverw> okay. set up RPC on it. :) gonna start it
116 2012-07-05 02:52:29 <Keverw> Had it set to allow 10.0.1.* to it
117 2012-07-05 02:53:08 <Keverw> Now what GPU miner should I use? A normal bit coin one?
118 2012-07-05 02:55:13 <Keverw> I guess I'll try Phoenix 2
119 2012-07-05 02:55:19 <D34TH> cgminer is always good
120 2012-07-05 02:55:45 <Keverw> does it really matter? uh& Sorry. Being such a noob...
121 2012-07-05 02:56:07 <D34TH> imo if your just testnetting not really
122 2012-07-05 02:56:19 <D34TH> its basically features
123 2012-07-05 02:57:41 <Keverw> oh okay. I need to edit my VM's so they get there own IP.
124 2012-07-05 02:58:02 <Keverw> but I'll try to mine some :)
125 2012-07-05 03:02:48 <Keverw> my password is password and my user is user :)
126 2012-07-05 03:03:36 <Keverw> Oh& fatal error: no kernal
127 2012-07-05 03:05:26 <D34TH> sounds pretty fatal
128 2012-07-05 03:05:28 <Keverw> gonna try the gui miner on the form.
129 2012-07-05 03:05:56 <Keverw> yay!
130 2012-07-05 03:06:03 <Keverw> Getting 577 khash
131 2012-07-05 03:06:22 <Keverw> 0 accepted so far...
132 2012-07-05 03:06:37 <Keverw> but doubt much transfers are being made right now
133 2012-07-05 03:10:22 <D34TH> 577 khs on a bitcoin testnet?
134 2012-07-05 03:10:32 <Keverw> yeah
135 2012-07-05 03:10:42 <D34TH> are you using a geforce mx 2?
136 2012-07-05 03:11:03 <Keverw> Nvidia GeForce 9300M GS 512MB i a laptop
137 2012-07-05 03:11:30 <D34TH> never mind
138 2012-07-05 03:11:47 <Keverw> so that's normal?
139 2012-07-05 03:12:58 <D34TH> a regular 9300 gs gets ~1.7 mh/s
140 2012-07-05 03:13:11 <D34TH> so mobile
141 2012-07-05 03:13:14 <D34TH> i suppose
142 2012-07-05 03:14:01 <Keverw> something from like 20 mins ago only has 1 conformation...
143 2012-07-05 03:14:21 <Keverw> but you can set the software to only check for 1, then make it 6 in production.
144 2012-07-05 03:14:56 <galambo> wow my processer gets like 4mh/s
145 2012-07-05 03:15:10 <D34TH> my proc gets ~3 mh/s
146 2012-07-05 03:15:20 <D34TH> my gpu gets ~250 mh/s
147 2012-07-05 03:16:10 <galambo> well you got me there mine only gets 50 :)
148 2012-07-05 03:16:25 <galambo> oh this is dev channel whiips
149 2012-07-05 03:16:37 <luke-jr> my proc gets 12 MH/s
150 2012-07-05 03:16:55 <galambo> one processor?
151 2012-07-05 03:17:00 <luke-jr> yes
152 2012-07-05 03:17:08 <luke-jr> my BitForce Single gets 830 MH/s
153 2012-07-05 03:17:15 <D34TH> D:
154 2012-07-05 03:17:21 <D34TH> i still want one
155 2012-07-05 03:17:24 <luke-jr> XD
156 2012-07-05 03:17:33 <D34TH> i just cant afford it
157 2012-07-05 03:17:43 <Keverw> I want the one that's gonna be 149
158 2012-07-05 03:17:47 <Keverw> in oct
159 2012-07-05 03:17:55 <luke-jr> that's 3.5 GH/s
160 2012-07-05 03:18:07 <Keverw> yeah for about 200 bucks(with shipping)
161 2012-07-05 03:18:43 <galambo> so because bitcoin d doesnt have long pooling when i mine with an external program it doesnt even realize that i just got a block
162 2012-07-05 03:19:06 <galambo> so basically if theres no one on the network i have to mine 2 blocks to get one?
163 2012-07-05 03:19:07 <luke-jr> galambo: you didn't.
164 2012-07-05 03:19:41 <luke-jr> oh, you mean received
165 2012-07-05 03:19:55 <luke-jr> that doesn't follow though
166 2012-07-05 03:20:16 <galambo> im getting blocks rejected in cgminer and im the only one mining
167 2012-07-05 03:20:26 <Keverw> on TestNet?
168 2012-07-05 03:21:02 <luke-jr> galambo: cgminer requires longpolling
169 2012-07-05 03:21:14 <luke-jr> galambo: try BFGMiner
170 2012-07-05 03:21:15 <Keverw> I'm on TestNet also mining also. So far 0 accepted and - stale
171 2012-07-05 03:21:21 <Keverw> Don't see a rejected area
172 2012-07-05 03:21:39 <galambo> its running i was just wondering what the rejected blocks were from
173 2012-07-05 03:22:04 <galambo> this is a test net for an experiment (test net in a box?)
174 2012-07-05 03:22:38 <da2ce7> is there anyway to tell the bitcoin client to re-send a transaction?
175 2012-07-05 03:32:07 <da2ce7> I have a transaction that shows up on my bitcoin client, except dosn't show up on block exploer or blockchain.info
176 2012-07-05 03:32:27 <da2ce7> is there any way to check if it conflicting, or hasn't been sent?
177 2012-07-05 03:32:43 <gmaxwell> da2ce7: it'll resend periodically (after a random span triggered by each new block without it)
178 2012-07-05 03:33:05 <da2ce7> well 6 blocks have been found without it.
179 2012-07-05 03:33:57 <Keverw> My mining does not seem to be speeding up Testnet&. :(
180 2012-07-05 03:34:18 <jgarzik> da2ce7: this is a good list to check, http://bitcoincharts.com/bitcoin/txlist/
181 2012-07-05 03:35:10 <da2ce7> jgarzik: nope not there.
182 2012-07-05 03:38:27 <Keverw> oh& Sneding within my own network is instant it seems!
183 2012-07-05 03:38:39 <luke-jr> da2ce7: I can check Eligius
184 2012-07-05 03:38:50 <da2ce7> ac5bdc79a87837e4ccb05928a5d4c9ba9dec7c6a2853c82b070d920e0bf31f93
185 2012-07-05 03:42:10 <da2ce7> I've reloaded the bitcoin client with the --rescan option.
186 2012-07-05 03:42:31 <da2ce7> the unconfirmed tx wasn't removed, so dose that mean that it dosn't conflict with anything?
187 2012-07-05 03:45:25 <gmaxwell> da2ce7: Nope, it can happily have a conflicting txn in the wallet.
188 2012-07-05 03:45:42 <da2ce7> why? is there a purpose for that?
189 2012-07-05 03:46:14 <gmaxwell> da2ce7: chain might reorg and leave it non-conflicting ... also, you don't want evidence of a bad transaction to just vanish.
190 2012-07-05 03:46:26 <gmaxwell> Somewhere on the todo list for someone is showing negative confirmations for those.
191 2012-07-05 03:47:17 <da2ce7> maybe the gui should put a big red 'x' on the tx if it has a conflicting block.  and a red '?' if there is a conflicting tx (unconfimed).
192 2012-07-05 03:47:39 <gmaxwell> da2ce7: you'll usually never hear of a conflicting txn.
193 2012-07-05 03:47:46 <gmaxwell> (unless its confirmed)
194 2012-07-05 03:48:29 <da2ce7> so what process do I go about to check for a conflicting tx?
195 2012-07-05 03:48:42 <da2ce7> I made more than 1 tx offline, then connected my client later.
196 2012-07-05 03:48:48 <da2ce7> so it is _possible_
197 2012-07-05 03:53:07 <Keverw> What does the BC bot do? Does it take any commands?
198 2012-07-05 03:57:52 <X-Scale> Is it mathematically possible for a bitcoin to be mined by two different users ?
199 2012-07-05 03:59:38 <luke-jr> X-Scale: not unless two different users manage to make the same address
200 2012-07-05 03:59:53 <luke-jr> which is much much less likely than the universe imploding randomly
201 2012-07-05 04:00:13 <jine> First of all, you don't mine "a bitcoin" - you mine a block. Secondly, it happens "all" the time, hence the orphans from time to time.
202 2012-07-05 04:00:22 <jine> Thirdly... what luke-jr said.
203 2012-07-05 04:00:31 <da2ce7> looks like it re-sent the tx after 9 blocks.
204 2012-07-05 04:00:32 <da2ce7> ac5bdc79a87837e4ccb05928a5d4c9ba9dec7c6a2853c82b070d920e0bf31f93
205 2012-07-05 04:05:16 <gmaxwell> why do you said 'looks like it re-sent the tx after 9 blocks' instead of .. say 8 blocks? .. or 1 block?
206 2012-07-05 04:05:20 <Karmaon__> bitcoinexpress boasting about ddosing bitcointalk
207 2012-07-05 04:05:24 <Karmaon__> in #litecoin apparently
208 2012-07-05 04:05:44 <gmaxwell> Why would he do that? kind of an empty achievement.
209 2012-07-05 04:05:49 <Karmaon__> <BitcoinEXpress> wanna see me DDoS Bitcointalk LOL
210 2012-07-05 04:05:54 <gmaxwell> "LOOK, I PEED ON THE WALL!"
211 2012-07-05 04:06:02 <Karmaon__> is btctalk protected by prolexic?
212 2012-07-05 04:06:10 <Karmaon__> not that it is uptime critical, or antyhing
213 2012-07-05 04:07:11 <Karmaon__> <BitcoinEXpress> damn they got some heavy duty, gimme about 60 seconds to drop 200 more hitters in the mix
214 2012-07-05 04:07:41 <Keverw> Yeah, I'm in the chat also...
215 2012-07-05 04:08:01 <Keverw> Who is this dude?
216 2012-07-05 04:08:09 <Karmaon__> last time he admited to be using ec2 for his shenanigans
217 2012-07-05 04:08:16 <jine> 08:08:09 < Karmaon__> last time he admited to be using ec2 for his shenanigans
218 2012-07-05 04:08:20 <jine> hahaha.
219 2012-07-05 04:08:29 <jine> Sorry for the paste, I'm a bit tired :)
220 2012-07-05 04:08:46 <Keverw> Oh... He told me use was using a bot net
221 2012-07-05 04:08:47 <Karmaon__> Keverw, some alt currency troll.
222 2012-07-05 04:09:07 <Keverw> Sounds like a jerk...
223 2012-07-05 04:09:09 <Karmaon__> Keverw, he sure wasn't a few months ago.
224 2012-07-05 04:09:10 <jine> To we have any IP on this troll?
225 2012-07-05 04:09:13 <jine> /whois :)
226 2012-07-05 04:09:33 <Keverw> Doubt I should say that. Afaird he would attack me
227 2012-07-05 04:09:34 <Karmaon__> I got hit by him a few months back on linode, got nulled route for 24 hours.
228 2012-07-05 04:09:36 <jine> And no, bitcointalk.org is not behind prolexic. It's hosted on softlayer directly.
229 2012-07-05 04:10:22 <Karmaon__> heh, they must be able to handle 1gbit+
230 2012-07-05 04:10:25 <Karmaon__> and its down...
231 2012-07-05 04:11:02 <Karmaon__> I spoke too soon, web server probably taking its sweet time.
232 2012-07-05 04:13:14 <Karmaon__> jine, he is very public about his ddos activities. Doesn't care about anonymity either.
233 2012-07-05 04:19:21 <luke-jr> [06:05:54] <gmaxwell> "LOOK, I PEED ON THE WALL!" <-- lol
234 2012-07-05 04:20:29 <Karmaon__> Heres the hilarious chat log: http://pastebin.com/dNRymrae
235 2012-07-05 04:22:49 <MagicalTux> [15:10:22] <Karmaon__> heh, they must be able to handle 1gbit+ <- not that I want anyone to actually try it
236 2012-07-05 04:22:57 <gmaxwell> Karmaon__: not polite to post logs of channels without announced public logs.. then again, that looks like total play-school time there.
237 2012-07-05 04:24:12 <Karmaon__> haven't you read my responsible disclosure section of my personal privacy statement?! lol
238 2012-07-05 04:25:10 <MagicalTux> Karmaon__: just checked bitcointalk's network graphs, no variation at all
239 2012-07-05 04:25:53 <Joric> i just sold at 6.75 at mtgox / bought at 6.53 at cryptoxchange, took 5 minutes, it was a way too easy O_o
240 2012-07-05 04:26:26 <Joric> pure profit is about 1.5% of 3.3% difference
241 2012-07-05 04:26:43 <jine> MagicalTux: He lost his connection... :(
242 2012-07-05 04:27:07 <Karmaon> MagicalTux, some pages took a second or two more to load, thats about exciting as it gets.
243 2012-07-05 04:27:19 <jine> He unforturnatly was able to close the webchat on freenode before he pinged out. A shame!
244 2012-07-05 04:28:16 <Joric> call bitcoin police!
245 2012-07-05 04:30:48 <midnightmagic> sigh. more summer douches returning home to roost and mark up the pavement.
246 2012-07-05 04:31:35 <midnightmagic> it was so nice and quiet over the winter.
247 2012-07-05 05:08:34 <gribble> New news from bitcoinrss: gmaxwell opened issue 1558 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/issues/1558>
248 2012-07-05 05:25:14 <Joric> did anyone see 'Indie Game: The movie'? http://www.indiegamethemovie.com
249 2012-07-05 05:25:18 <Joric> funny this soap opera with Fez lasted 5 years while some guy made his own Fez, alone, in 6 weeks http://www.kongregate.com/games/Neutronized/sky-island
250 2012-07-05 05:32:33 <dooglus> I saw it recently
251 2012-07-05 05:33:41 <dooglus> the braid guy was funny, complaining how nobody 'got it' - interspersed with some rapper totally not getting it :)
252 2012-07-05 06:03:02 <tgs3> bitcoind is "slow as fuck" when getting to end of chain
253 2012-07-05 06:03:14 <Joric> huge blocks?
254 2012-07-05 06:03:14 <tgs3> why is that error not being fixed?
255 2012-07-05 06:03:39 <Joric> someone should 'fix' that satoshidice thing
256 2012-07-05 06:03:56 <tgs3> 0.1 blocks per second in nonsence. huge blocks or not
257 2012-07-05 06:41:05 <sipa> BlueMatt: BOOST_FOREACH is certainly intended to result in code that gets optimized away entirely
258 2012-07-05 06:41:57 <sipa> BlueMatt: i admit i never verified that; if it doesn't seem to be the case, i'll gladly switch to stl iterators and normal for loops
259 2012-07-05 06:42:24 <sipa> BlueMatt: regarding 0.7, i think gavin still has a few high-priority todo's left
260 2012-07-05 07:34:08 <tgs3> ok so I found a bug in bitcoin
261 2012-07-05 07:34:41 <tgs3> bitcoind getinfo  fucks up even very strong computers - causing 100% system load with disk i/o for half minute even
262 2012-07-05 07:35:13 <sipa> is that while syncing up?
263 2012-07-05 07:39:13 <tgs3> yes
264 2012-07-05 07:39:20 <tgs3> it takes few DAYS to sync
265 2012-07-05 07:40:57 <sipa> what hardware?
266 2012-07-05 07:41:08 <tgs3> 4x2 GHz 8 GB ram
267 2012-07-05 07:41:13 <sipa> in particular cpu, ram and disk?
268 2012-07-05 07:41:23 <tgs3> and kvm 1x2 GHz 1 GB ram server
269 2012-07-05 07:41:26 <tgs3> non-ssd
270 2012-07-05 07:41:44 <tgs3> only computer that is not being messed up by bitcoind is ssd based one
271 2012-07-05 07:42:20 <tgs3> althou that one is anyway synced
272 2012-07-05 07:42:29 <sipa> storage is not network based?
273 2012-07-05 07:42:35 <tgs3> normal storage
274 2012-07-05 07:42:50 <sipa> which bitcoin version?
275 2012-07-05 07:43:09 <tgs3> latest, 0.6.3
276 2012-07-05 07:44:09 <sipa> anyway, getinfo taking a long time while syncing up is a known problem. it's not doing anything, but it has to wait for exclusive access to some data structures, which are already used by the block connecting logic
277 2012-07-05 07:44:23 <tgs3> thoug strace -p of bitcoi9nd nd  processs shows only sleeping/gettieme. huh?
278 2012-07-05 07:44:44 <kinlo> sipa: isn't it possible to used cached values for getinfo?
279 2012-07-05 07:44:57 <kinlo> sipa: also, is getmemorypool blocked by any of those data structures?
280 2012-07-05 07:45:24 <tgs3> kinlo: more important then just getinfo, bitcoind fucks up entire computer (now all programs lag) and itself (it even can't catch up with blockchain)
281 2012-07-05 07:45:37 <sipa> kinlo: yes, all RPC calls preemptively require those locks
282 2012-07-05 07:45:38 <tgs3> so again I ask
283 2012-07-05 07:45:43 <kinlo> tgs3: I know...
284 2012-07-05 07:45:49 <tgs3> did someone got on rapge and added dozen of fucking fsyncs()
285 2012-07-05 07:46:04 <tgs3> *rampage
286 2012-07-05 07:46:17 <kinlo> sipa: would be nice if some are from cache, bitcoind needs to be much more responsive :/
287 2012-07-05 07:46:20 <sipa> tgs3: bdb does a lot of things to make sure db updates are flushed to disk
288 2012-07-05 07:46:25 <tgs3> in example I can not use this irssis program here bearly, it pauses for 5 seconds while I type lines
289 2012-07-05 07:46:50 <sipa> kinlo: solution is moving to a sane encapsulation and locking
290 2012-07-05 07:46:55 <tgs3> is there yet a bounty to poverize bdb developers responsible?
291 2012-07-05 07:47:16 <sipa> tgs3: we plan to move away from bdb maybe
292 2012-07-05 07:47:39 <tgs3> or, is there a flag -i-dont-five-a-flying-f-about-sync-only-sync-my--wallet-and-corrupt-watever-else-just-dont-turn-my-computer-into-486-please
293 2012-07-05 07:48:19 <tgs3> actually working on 486 was much much smoother then on 4x2 GHz 8GB ram with bitcoind
294 2012-07-05 07:48:42 <kinlo> tgs3: you should use electrum or something
295 2012-07-05 07:48:43 <tgs3> also you managed to superprass freenet in the category of destroying computers with high IO. that is a remarkable achievement
296 2012-07-05 07:49:00 <sipa> tgs3: not without giving up database consistency
297 2012-07-05 07:49:10 <tgs3> I do not give a flying f about db consistency
298 2012-07-05 07:49:11 <kinlo> endusers shouldn't run bitcoin imho
299 2012-07-05 07:49:22 <tgs3> and also so do all users
300 2012-07-05 07:49:45 <tgs3> kin
301 2012-07-05 07:49:46 <kinlo> tgs3: did you ever look at electrum?
302 2012-07-05 07:50:07 <tgs3> no
303 2012-07-05 07:50:26 <Diablo-D3> https://bitcointalk.org/index.php?topic=91710.0
304 2012-07-05 07:50:41 <kinlo> why would we have failed?
305 2012-07-05 07:50:46 <sipa> tgs3: you do, or rather the network does; if significant numbers of people would run bitcoin with inconsistent databases, we'd risk getting many blockchain forks
306 2012-07-05 07:50:57 <tgs3> because if almost no users use bitcoind on their own just use wallet servers,
307 2012-07-05 07:51:04 <tgs3> then basically we have a banking industry
308 2012-07-05 07:51:22 <sipa> also, i sync the blockchain in around 1-w hours
309 2012-07-05 07:51:22 <tgs3> in terms of owning own money  and being self sufficient, its a complete fiasco
310 2012-07-05 07:51:41 <kinlo> tgs3: not really, every user can verify the system, plus regardless of the fact that we would use trusted servers, the coins would still be on the end-computer, not on the central server
311 2012-07-05 07:51:43 <sipa> 1-2 hours, on my laptop with an encrypted disk
312 2012-07-05 07:52:00 <kinlo> sipa: full blockchain?
313 2012-07-05 07:52:04 <tgs3> kinlo: well more or less maybe, but
314 2012-07-05 07:52:13 <tgs3> but it JUST WORKED till last month
315 2012-07-05 07:52:16 <sipa> tgs3: so i wonder what is causing the huge delays for you
316 2012-07-05 07:52:16 <tgs3> what happened?
317 2012-07-05 07:52:25 <kinlo> dice?
318 2012-07-05 07:52:27 <tgs3> revert what ever shit was introduced to guarantee DB consistency
319 2012-07-05 07:52:37 <kinlo> revert satoshidice? :)
320 2012-07-05 07:52:43 <tgs3> what dice? block are now too big?
321 2012-07-05 07:52:55 <sipa> consistency guarantees were only more and more relaxed to improve performance
322 2012-07-05 07:53:01 <tgs3> isnt there some mechanism to forbid too much spam or limit block sizes?
323 2012-07-05 07:53:20 <kinlo> tgs3: http://blockchain.info/charts/n-transactions
324 2012-07-05 07:53:26 <sipa> but bdb does not have a "i dont care about durability, i just want transactions to be atomic"
325 2012-07-05 07:53:45 <tgs3> ok then why handling huge blocks ok then bitcpoind is not written to handle big transactions yet?
326 2012-07-05 07:54:01 <kinlo> tgs3: there are plenty of mechanisms
327 2012-07-05 07:54:17 <sipa> again, i wonder what is causing such load on your system
328 2012-07-05 07:54:23 <tgs3> bitcoind
329 2012-07-05 07:54:25 <kinlo> tgs3: the block fee's have also been a lot higher last few months: miners earn more
330 2012-07-05 07:54:48 <kinlo> sipa: I must say once bitcoind is open on my system and syncing, it is very slow
331 2012-07-05 07:55:00 <sipa> kinlo: in an experimental branch of mine i synced to 172k blocks in 9 minutes
332 2012-07-05 07:55:05 <sipa> yesterday
333 2012-07-05 07:55:15 <kinlo> sipa: full blocks?
334 2012-07-05 07:55:18 <tgs3> Diablo Mining Company will never purchase Butterfly Labs hardware.  <--- oh drama
335 2012-07-05 07:55:31 <sipa> kinlo: how do you mean?
336 2012-07-05 07:56:04 <kinlo> sipa: syncing block 170000 to 180000 takes more time then syncing block 70000 to 80000
337 2012-07-05 07:56:12 <kinlo> sipa: just due to the number of transactions
338 2012-07-05 07:56:22 <tgs3> sipa it will take 30 hours to snynd last 00200 blocks (2000)
339 2012-07-05 07:56:27 <tgs3> Diablo Mining Company will be the largest Bitcoin startup in the world,
340 2012-07-05 07:56:38 <tgs3> ha. :)
341 2012-07-05 07:56:53 <sipa> kinlo: of course, and due to the fact that signatures are only verified after the last checkpoint
342 2012-07-05 07:57:13 <kinlo> tgs3: uh, 30 hours is a bit exagerated, I agree a full sync takes hours, but any decent pc should manage a full download in a few hours, not a few days
343 2012-07-05 07:57:30 <sipa> kinlo: that's why i say 172k, to give you an idea; what comes afterwards will certainly be slower
344 2012-07-05 07:57:31 <kinlo> given a fast enough peer transmitting the blocks
345 2012-07-05 07:57:37 <tgs3> sipa: but why it uses so muhch disk io
346 2012-07-05 07:58:02 <sipa> kinlo: but 0.6.3 takes over half an hour to sync 172k blocks for me on the same system
347 2012-07-05 07:58:57 <sipa> tgs3: because it needs a working set of over 2 gigabytes of data, and all updates are done using logged and atomic transactions
348 2012-07-05 07:59:05 <tgs3> why  it needs disk IO to verify signatures
349 2012-07-05 07:59:21 <sipa> verifying signatures is not the problem
350 2012-07-05 07:59:42 <tgs3> it is receinvg simply som data from net to save to hdd... bittorent clients figured this out 10 years ago
351 2012-07-05 08:00:17 <tgs3> you must be doing something wrong there
352 2012-07-05 08:00:30 <sipa> your cpu can verify 3000 signatures per second, and there are around 5 million signatures in the entire chain
353 2012-07-05 08:01:08 <sipa> tgs3: it is keeping a very complex data structure up to date
354 2012-07-05 08:01:09 <tgs3> sipa I would be ok with that, but why it eats all availble hdd io
355 2012-07-05 08:01:32 <tgs3> ok here is an idea: fuck consistency
356 2012-07-05 08:01:35 <sipa> you can run it on a tmpfs if you don't care about durability and have ram
357 2012-07-05 08:01:53 <tgs3> instead, prepare all data in memory, and once per 10 minutes write all verified blocks / complex datastructure in one go
358 2012-07-05 08:02:04 <sipa> tgs3: you want fuck durability, not consistency
359 2012-07-05 08:02:14 <tgs3> that too
360 2012-07-05 08:02:19 <sipa> that is exactly what you asked for now
361 2012-07-05 08:02:42 <tgs3> but we can have both if we prepare the new portion of data (blocks, verified transfers etc?) in memory and then save all in one go once ready?
362 2012-07-05 08:03:17 <tgs3> or, I hope we can
363 2012-07-05 08:03:37 <sipa> you'd lose durability, as a crash during those 10 minutes would mean losing everything done during those minutes
364 2012-07-05 08:03:48 <sipa> and that would not at all be a problem
365 2012-07-05 08:04:02 <sipa> but bdb does not allow that without giving up consistency
366 2012-07-05 08:04:51 <sipa> anyway, we're very much working on improving that
367 2012-07-05 08:04:52 <tgs3> then bdb really sucks
368 2012-07-05 08:05:16 <sipa> it is very good software, but not very appropriate for our needs
369 2012-07-05 08:05:56 <tgs3> too good database lib causing high disk IO
370 2012-07-05 08:06:00 <tgs3> is exactly what killed Freenet-Freetalk projects
371 2012-07-05 08:06:30 <tgs3> so I hopefully we avoid that :)
372 2012-07-05 08:07:21 <sipa> that said, i'm working on a branch that has a reworked validation engine, and only needs around a hundred MB of working set (right now, as opposed to two GB)
373 2012-07-05 08:07:59 <sipa> and TD was experimenting with switching the DB to LevelDB
374 2012-07-05 08:09:08 <sipa> and BlueMatt was working on improving the locking system, so more things can be done simultaneously
375 2012-07-05 08:09:18 <tgs3> is there any bitcoind like software that supports wallets switching and key import/export?
376 2012-07-05 08:09:47 <sipa> bitcoind has key import/export
377 2012-07-05 08:10:02 <sipa> armory certainly has multiwallet suppott
378 2012-07-05 08:10:11 <sipa> electrum i suppose too
379 2012-07-05 08:10:16 <sipa> multibit unsurr
380 2012-07-05 08:10:19 <sipa> unsure
381 2012-07-05 08:11:18 <sipa> still, i want to know why it hammers your disk so hard
382 2012-07-05 08:11:53 <tgs3> me too
383 2012-07-05 08:12:01 <tgs3> strace shows only sleep/gettime
384 2012-07-05 08:12:37 <sipa> well, those are not the problem
385 2012-07-05 08:12:39 <sipa> :)
386 2012-07-05 08:12:44 <tgs3> nokidding
387 2012-07-05 08:13:29 <sipa> i know
388 2012-07-05 08:13:43 <sipa> but it's not only those calls, just mostly
389 2012-07-05 08:14:53 <sipa> and bdb uses memory mapped files to store things, so you won't see disk writes as separate syscalls
390 2012-07-05 08:15:02 <tgs3> neat
391 2012-07-05 08:15:22 <tgs3> how to see it then
392 2012-07-05 08:48:55 <tgs3> if I want to have another node but NOT redownload blockchain, can I copy all files and then remove wallet.dat for the new node? or what else?
393 2012-07-05 08:49:22 <tgs3> or just start a new node, stop it, copy only thoes blk*.dat files?
394 2012-07-05 08:52:24 <ersi> I think you need to do a 'database detach' (there's an option in bitcoin-qt as well as -detach in just bitcoind) first before moving to another location
395 2012-07-05 09:01:44 <Keverw> my comp freezes, but the FBI watches these IRC channels? Neat!
396 2012-07-05 09:02:12 <Keverw> heard the ban sound in my IRC client for the first time and looked.
397 2012-07-05 09:03:49 <Keverw> rg: is BitVPS your site? I've been wanting a second VPS. Right now I have one with Linode, bur If I'm going to use the Bitcoin API, I would prefer to have my Bitcoins on a separate server than my personal site.
398 2012-07-05 09:06:43 <Keverw> rg: No XEN servers?
399 2012-07-05 09:08:14 <galambo> 
400 2012-07-05 09:08:36 <Keverw> rg: https://client.bitvps.com/billing/cart.php?a=confproduct&i=0 looks nice. I think I might pick you guys for my second VPS when I'm readying for one.
401 2012-07-05 09:15:29 <Keverw> Ops. I was replaying on the wrong channel&.  I feel stupid. Sleepy
402 2012-07-05 09:16:26 <Keverw> What's the best way to store Bitcoin in a MySQL database?
403 2012-07-05 09:16:47 <sipa> why do you want that?
404 2012-07-05 09:16:55 <Keverw> like store the amount.
405 2012-07-05 09:17:06 <sipa> of?
406 2012-07-05 09:17:11 <Keverw> Bitcoin and Bitcents
407 2012-07-05 09:17:37 <Keverw> would it be the same as US dollars?
408 2012-07-05 09:17:42 <Keverw> just higher?
409 2012-07-05 09:18:46 <sipa> you mean how to store amounts in mysql?
410 2012-07-05 09:18:50 <Keverw> yes.
411 2012-07-05 09:19:03 <sipa> ah, i thought you were asking about a client that used mysql as backend
412 2012-07-05 09:19:13 <Keverw> nope.
413 2012-07-05 09:19:14 <sipa> doesn't mysql have a fixed-precision type?
414 2012-07-05 09:19:32 <Keverw> Not sure& I know this one site says to store it as a decimal
415 2012-07-05 09:19:35 <Keverw> for USD
416 2012-07-05 09:19:47 <Keverw> 10,2 but I guess for BTC you need to adjust them
417 2012-07-05 09:20:36 <sipa> Decimal seems to be the name mysql uses for fixed-precision values :)
418 2012-07-05 09:20:42 <tgs3> ersi: database? no its just blockchain
419 2012-07-05 09:21:26 <Keverw> do if statements even compare values if they use decimals?
420 2012-07-05 09:21:26 <tgs3> just operate on integer
421 2012-07-05 09:21:31 <tgs3> btc * 10^6 afair
422 2012-07-05 09:21:32 <sipa> Keverw: yes
423 2012-07-05 09:21:34 <Keverw> like if 3.40 > 2.50
424 2012-07-05 09:21:48 <Keverw> >= would be better I guess, like in Node.js or PHP?
425 2012-07-05 09:21:49 <tgs3> use integeres
426 2012-07-05 09:22:05 <tgs3> store number of satoshis
427 2012-07-05 09:22:12 <Keverw> like if someone with 3.40 wants to buy an item for 2.50
428 2012-07-05 09:22:18 <sipa> tgs3: that's harder, and in lower-level languages the only right way
429 2012-07-05 09:22:32 <tgs3> sipa: how is using integers harder
430 2012-07-05 09:22:34 <sipa> but if your languages provides fixed-precision numbers, they're perfect for the job
431 2012-07-05 09:22:39 <ersi> tgs3: I don't know for sure. I said, I *THINK* one has to do that before moving the data. And in case you didn't know, the blockchain is a hueg database
432 2012-07-05 09:22:48 <ersi> Don't take my word for it though
433 2012-07-05 09:26:03 <Keverw> So I need to google if fixed-precision numbers are supported in PHP, Javascript?
434 2012-07-05 09:27:02 <sipa> oh, if you need to process values (apart from just passing them through) in PHP or Javascript, definitely use integer amounts in satoshis
435 2012-07-05 09:27:27 <Keverw> satoshis?
436 2012-07-05 09:27:42 <sipa> 1 satoshi = 0.00000001 BTC
437 2012-07-05 09:27:57 <Keverw> o
438 2012-07-05 09:28:13 <sipa> ersi: the blockchain is not entirely a database, actually; the blocks/transactions are (for now) just stored in a flat file
439 2012-07-05 09:29:24 <ersi> sipa: What exactly does 'detach the database' mean, by the way? I'm kind of curious about that
440 2012-07-05 09:29:38 <ersi> Guess one has to define what 'the database' means then ^_^
441 2012-07-05 09:30:05 <sipa> ersi: BDB creates many files in the database directory, and the actual db files are tied to those (they can refer to log files in that directory, and such)
442 2012-07-05 09:30:21 <sipa> detaching the DB file means making it standalone
443 2012-07-05 09:30:40 <Keverw> so 9 places is the max for bitcents?
444 2012-07-05 09:30:48 <sipa> Keverw: 8
445 2012-07-05 09:30:53 <Keverw> o
446 2012-07-05 09:31:13 <sipa> (1 BTC can be subdivides 100 million times)
447 2012-07-05 09:31:24 <Keverw> 0.00000001 looks like 9& .or maybe I can't count
448 2012-07-05 09:31:33 <Keverw> oh& I see what I did
449 2012-07-05 09:31:42 <Keverw> so so far I know we have *,8 for the value
450 2012-07-05 09:31:55 <Keverw> 21 million cap
451 2012-07-05 09:32:21 <ersi> sipa: So, it'd be advisable to do if you'd like to move the blockchain data from one computer to another?
452 2012-07-05 09:32:27 <Keverw> so that's 21,000,000
453 2012-07-05 09:32:30 <sipa> ersi: yes
454 2012-07-05 09:32:40 <ersi> Neat. Thanks for the answers sipa
455 2012-07-05 09:32:41 <sipa> ersi: if will most certainly fail to load otherwise
456 2012-07-05 09:32:47 <ersi> Yeah, I've tried that :)
457 2012-07-05 09:33:14 <Keverw> so 8?
458 2012-07-05 09:33:18 <Keverw> 8,8?
459 2012-07-05 09:33:23 <Keverw> or 9,8
460 2012-07-05 09:34:10 <sipa> Keverw: in mysql-speak, you want DECIMAL(16,8)
461 2012-07-05 09:34:33 <sipa> (first number is the total amount of significant digits)
462 2012-07-05 09:34:59 <Keverw> oh. okay Thanks!
463 2012-07-05 09:35:39 <sipa> but as i said, if you need to do anything but just showing/input/passthrough the amounts (even just adding them together) in PHP or JS, store them as integer multiples of 1 satoshi
464 2012-07-05 09:36:53 <Keverw> I guess&. completed.
465 2012-07-05 09:36:57 <Keverw> compliated*
466 2012-07-05 09:37:02 <Keverw> I hate Autocorrect....
467 2012-07-05 09:37:07 <sipa> compliated?
468 2012-07-05 09:37:27 <Keverw> compicated.
469 2012-07-05 09:37:31 <sipa> complicated? compiled? compliant?
470 2012-07-05 09:37:38 <Keverw> I'm a horrible speller& just confusing. like hard&. I guess
471 2012-07-05 09:41:06 <sipa> not sure what you mean, still
472 2012-07-05 09:41:10 <Keverw> I wonder how Paypal does it& Do they store everything in cents?
473 2012-07-05 09:41:34 <sipa> probably
474 2012-07-05 09:42:26 <JFK911> i think paypal stores everything in floats
475 2012-07-05 09:42:44 <Keverw> Yeah& Why would DECIMAL(16,8) have problems?
476 2012-07-05 09:43:06 <Keverw> Just don't want to over complicate things.
477 2012-07-05 09:43:39 <Keverw> and can't wait till MT Lion is on the  app store. For us horrible spellers we can just press the function key and say it
478 2012-07-05 09:43:53 <Keverw> system wide speech input.
479 2012-07-05 12:36:10 <MysteryBanshee> gmaxwell: :/
480 2012-07-05 12:37:10 <copumpkin> MysteryBanshee: please stop PMing me with threats. I don't have the power to fix your mess, and if I did, I wouldn't choose to.
481 2012-07-05 12:37:31 <MysteryBanshee> copumpkin: pm
482 2012-07-05 12:37:45 <copumpkin> please stop PMing me at all
483 2012-07-05 12:37:48 <copumpkin> I don't care
484 2012-07-05 12:38:05 <tgs3> copumpkin: Y U NO care
485 2012-07-05 12:38:33 <MysteryBanshee> ok fair enough, if you dont care :)
486 2012-07-05 12:39:05 <tgs3> MysteryBanshee: try calling his mobile Im sure it will help
487 2012-07-05 12:39:25 <copumpkin> MysteryBanshee: keep being a creep. See how far it gets you
488 2012-07-05 12:39:33 <MysteryBanshee> why am I a creep copumpkin?
489 2012-07-05 12:40:05 <tgs3> MysteryBanshee: can you give your bitcoin address
490 2012-07-05 12:40:06 <copumpkin> uh, threatening random people with mysterious unknown punishments for things you brought upon yourself and that they had little say in?
491 2012-07-05 12:40:20 <copumpkin> anyway, this isn't the place
492 2012-07-05 12:40:22 <copumpkin> #bitcoin-otc-foyer
493 2012-07-05 12:40:28 <MysteryBanshee> copumpkin: hah, yeh, its my fault I get called a scammer 24/7 and peopel think its ok to impersonate me
494 2012-07-05 12:41:24 <sipa> what did he do?
495 2012-07-05 12:42:47 <Keverw> Yeah& I was wondering the same but the ban sound in Colloguy is nice.
496 2012-07-05 12:44:11 <gmaxwell> In here, nothing yet. In OTC he spent his first weak there buying positive ratings then switched to threating to scam people if $random_demand wasn't met.
497 2012-07-05 12:44:32 <sipa> huh, ok
498 2012-07-05 12:44:52 <gmaxwell> s/weak/week/ gah, dunno how I make errors like that.
499 2012-07-05 12:46:36 <tgs3> gmaxwell: Im not thinking its 100% ok to ban someone here on -dev based on his actions related to bitcoin trading in trading channel
500 2012-07-05 12:46:53 <MrTiggr> omfg the banhammer swings slow in here .. banshees do not dev
501 2012-07-05 12:48:04 <Keverw> tgs3: yeah, but he was starting to bring it in to this room also.
502 2012-07-05 12:48:08 <MrTiggr> the only reason we haz poltergiests in here is coz it got butthurt in every other channel there might be an OP   .. swing away gmaxwell i say
503 2012-07-05 12:49:00 <ersi> tgs3: If the person is question only brings drama and no -dev on topic discussions, I say it's fair game
504 2012-07-05 12:49:07 <ersi> s/is/in/
505 2012-07-05 12:58:30 <tgs3> ok maybe it is. persoanlly I would wait longer tho :)\n3932167
506 2012-07-05 13:22:57 <sipa> gmaxwell: 172k blocks in less than 5 minutes!
507 2012-07-05 13:23:07 <tgs3> sipa: how?
508 2012-07-05 13:23:31 <gmaxwell> Ultraprune"
509 2012-07-05 13:23:43 <sipa> (ok, i cheated, processing 50 blocks at a time during IBD)
510 2012-07-05 13:24:14 <BlueMatt> its not cheating if there is a branch that does that in a mergeable way ;)
511 2012-07-05 13:24:35 <gmaxwell> well, it doesn't reflect non-checkpointed performance. Still the numbers without that were good too.
512 2012-07-05 13:25:28 <BlueMatt> non-checkpointed the biggest difference is sig checking, and that can be threaded (again, there is a merge-able branch that does that too)
513 2012-07-05 13:26:30 <sipa> TD suggested using a queue for signatures to be checked, and preliminarily connect blocks before checking sigs
514 2012-07-05 13:26:44 <gmaxwell> dist-sigcheck
515 2012-07-05 13:26:50 <sipa> oh yeah
516 2012-07-05 13:27:12 <gmaxwell> that would make it a _lot_ easier to integrate gpu / FPGA sig checking code.
517 2012-07-05 13:27:29 <BlueMatt> https://github.com/TheBlueMatt/bitcoin/commit/17b39780fb69f56a65455317e481154520206b01
518 2012-07-05 13:27:59 <BlueMatt> in terms of doing block connect before doing sig checking...meh
519 2012-07-05 13:27:59 <sipa> :o
520 2012-07-05 13:28:16 <sipa> how can you thread connectinputs?
521 2012-07-05 13:28:47 <BlueMatt> you have to loack it a bit: https://github.com/TheBlueMatt/bitcoin/commit/e1c5efb81f1c994c4e7925c816e29352d7e7ac42
522 2012-07-05 13:28:50 <gmaxwell> In general block connect before doing sig checking would also support a structure that would also work for proofs-of-treachery, e.g. where you stochastically validate and listen for other nodes to send you alerts of baddness they found.
523 2012-07-05 13:28:53 <BlueMatt> s/loack/lock/
524 2012-07-05 13:29:46 <BlueMatt> never got around to actually pre-loading inputs (otherwise the locking kills in the first commit) but thats not hard
525 2012-07-05 13:31:02 <BlueMatt> ooo...9 line change that decreases cpu cycles spent in bitcoin-qt during initial load from 3.3bill to 1.7bill
526 2012-07-05 13:31:13 <sipa> BlueMatt: which?
527 2012-07-05 13:31:39 <BlueMatt> the only important line is this in Serialize in IMPLEMENT_SERIALIZE:
528 2012-07-05 13:32:02 <sipa> which?
529 2012-07-05 13:32:15 <BlueMatt> ie pre-allocate CDataStream when serializing
530 2012-07-05 13:32:39 <sipa> 184k blocks now
531 2012-07-05 13:32:43 <sipa> (from network)
532 2012-07-05 13:34:19 <gmaxwell> BlueMatt: so the savings is all just excess heap allocation load?
533 2012-07-05 13:35:06 <gmaxwell> I noticed in my callgrind results that there were a number of places where thrashing the heap allocator was a major cpu sink, at least as far as callgrind was concerned.
534 2012-07-05 13:35:13 <BlueMatt> gmaxwell: yep
535 2012-07-05 13:35:36 <gmaxwell> BlueMatt: are you're measurements just callgrind or gcov  or are you using something less invasive like oprofile?
536 2012-07-05 13:35:48 <BlueMatt> callgrind
537 2012-07-05 13:36:01 <sipa> damn, 50 block connects take around 20s now
538 2012-07-05 13:36:14 <gmaxwell> I ask because simple instruction counters tend to overstate the cost of stuff which has a lot of very well predicted branches (like the heap allocator)... so it may be a bit less than your measurements suggest.
539 2012-07-05 13:36:19 <BlueMatt> though that was -checkblocks=100 to keep from having to spend an hour per benchmark...
540 2012-07-05 13:36:26 <gmaxwell> sipa: need parallel ECDSA. ;)
541 2012-07-05 13:36:28 <BlueMatt> gmaxwell: it probably is
542 2012-07-05 13:36:44 <sipa> gmaxwell: still below last checkpoint
543 2012-07-05 13:36:51 <gmaxwell> oh. hm!
544 2012-07-05 13:36:58 <sipa> gmaxwell: oh, no!
545 2012-07-05 13:37:01 <sipa> just passed it
546 2012-07-05 13:37:12 <BlueMatt> sipa: Im really interested to see what ultraprune looks like after threaded block commit+threaded connectinputs+all the GetHash() removal
547 2012-07-05 13:37:37 <sipa> BlueMatt: i must warn you that the block connection logic was changed a lot
548 2012-07-05 13:37:48 <sipa> (though it's a lot cleaner now, imho)
549 2012-07-05 13:38:01 <BlueMatt> yea, I have a bad feeling that the whole cblockstore stuff is gonna die again after ultraprune...
550 2012-07-05 13:38:47 <sipa> there's a abstract CCoinsView that represents a set of coins/connect blocks, with several implementations (one backed by DB, one by mempool, and one that adds a cache on top of it)
551 2012-07-05 13:39:08 <sipa> and ConnectBlock is just CBlock::ConnectBlock(CBlockIndex *pindex, CCoinsView &view)
552 2012-07-05 13:39:15 <BlueMatt> fun...
553 2012-07-05 13:39:18 <sipa> instead of 15 parameters for what kind of mode it is used in
554 2012-07-05 13:40:21 <gavinandresen> so... how are we going to test to make sure there's not a horrible edge-case-network-forking bug lurking?
555 2012-07-05 13:40:38 <sipa> gavinandresen: thorough testing?
556 2012-07-05 13:41:05 <gavinandresen> you writing lots of edge-case blockchains we can throw at it?
557 2012-07-05 13:41:11 <sipa> haha
558 2012-07-05 13:41:25 <sipa> one thing i think about: the coins db and undo files are entirely deterministic
559 2012-07-05 13:41:45 <sipa> reorg'ing to an older state should give you exactly the same state again
560 2012-07-05 13:43:15 <gmaxwell> It may make sense to set a goal of getting testcases that give 100% branch coverage of operative branches.  (The hard part will be sorting out true operative branches vs unreachable error handling added by boost and friends)
561 2012-07-05 13:44:44 <gmaxwell> And probably mutations to test the tests. (e.g. go and invert or nullify each test in the connection logic one at a time, looping on recompiling and running the tests to make sure that every change kills it) I'll work on that when its ready.
562 2012-07-05 13:44:47 <BlueMatt> 100% branch coverage of branches in bitcoin code itself (is there a tool to test that?) is a easier goal that gets much of the benifit (could be seen as a first goal towards getting more overall coverage)
563 2012-07-05 13:45:12 <gmaxwell> BlueMatt: GCOV will do so, but you'll get branches leaking in from macros.
564 2012-07-05 13:45:48 <BlueMatt> mmm,yea
565 2012-07-05 13:45:50 <gmaxwell> the LCOV reports look like this: https://mf4.xiph.org/jenkins/job/opus-coverage/ws/coverage/celt/index.html
566 2012-07-05 13:46:23 <sipa> the compact serializers are a bit complex (and probably overkill), but the worst ones already have unit tests
567 2012-07-05 13:46:48 <gavinandresen> The case I really care about is some complicated edge-case block-chain re-org that is rejected by the new code but accepted by the old code (or vice-versa)
568 2012-07-05 13:47:02 <gavinandresen> ... because that would split the network
569 2012-07-05 13:47:15 <sipa> since BIP30, reorgs are fulle deterministic
570 2012-07-05 13:47:20 <gmaxwell> (by branchs it means 'modified condition decision' basically)
571 2012-07-05 13:47:37 <sipa> so you could even define a function that calculates a hash (merkle root?) of all coins in the database
572 2012-07-05 13:47:42 <sipa> implement that for old and new code
573 2012-07-05 13:47:54 <sipa> and verify that they are exactly identical
574 2012-07-05 13:48:26 <sipa> and then fire your worst at it (testnet?)
575 2012-07-05 13:48:58 <gmaxwell> gavinandresen: right. If we're able to generate a block file that hits every outcome in every branch of the validation, then run on both and show that the txout set is the same at every step... that would be fair evidence. Though not conclusive proof.
576 2012-07-05 13:49:23 <sipa> ok, fully synced
577 2012-07-05 13:49:35 <sipa> -rw-------  1 pw pw  38289408 2012-07-05 17:47 chain.dat
578 2012-07-05 13:50:09 <gavinandresen> gmaxwell: 100% testing coverage on both old and new code would make me sleep much better
579 2012-07-05 13:50:40 <gmaxwell> We could also release such changes by forking the software. e.g. release it as bitcoin-qt-turbo encouraged for home users, discouraged for high value services for a while before merging it back. Sadly an attacker would wait in order to maximize their impact, but at least running in parallel for a while would increase confidence.
580 2012-07-05 13:50:54 <Keverw> should varchar 34 be good for storing an address in mysql?
581 2012-07-05 13:51:04 <sipa> SetBestChain is now just one (slightly longer) function now, instead of having parts split off for reorganisation and normal connections
582 2012-07-05 13:51:18 <sipa> and there is only a special case for the genesis block
583 2012-07-05 13:51:29 <gavinandresen> sipa: nice!
584 2012-07-05 13:51:34 <sipa> all other reorgs are just one code path, with hardly any conditionals
585 2012-07-05 13:51:35 <gmaxwell> sipa: 162MiB ... not so shabby.
586 2012-07-05 13:52:10 <gmaxwell> sipa: do you know how much data is actually in them?
587 2012-07-05 13:52:37 <sipa> gmaxwell: coins.dat contains around 67 MB of data
588 2012-07-05 13:52:55 <sipa> gmaxwell: what does "not so shabby" mean?
589 2012-07-05 13:53:05 <gmaxwell> sipa: That looks good.
590 2012-07-05 13:53:24 <sipa> chain.dat can be made at least 4 times smaller, if we'd store block indices a bit more efficiently
591 2012-07-05 13:53:30 <luke-jr> gavinandresen: fwiw, txn_prio contains some unit tests for block creation
592 2012-07-05 13:54:02 <sipa> now... let me try syncing again without the 50-blocks-at-a-time
593 2012-07-05 13:54:08 <sipa> since i didn't get any reorgs yet
594 2012-07-05 13:58:14 <sipa> ok, pushed to ultraprune, if someone would like to benchmark
595 2012-07-05 14:11:13 <gribble> New news from bitcoinrss: sje397 opened pull request 1559 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/1559>
596 2012-07-05 14:15:53 <BlueMatt> gavinandresen: what are your current plans for 0.7?
597 2012-07-05 14:17:40 <gavinandresen> BlueMatt: dunno, haven't looked at all the pulls.  I want to finish the blockheigh-in-coinbase change.
598 2012-07-05 14:18:30 <gavinandresen> The highest priority not-yet-done item on my wish list is fee rework
599 2012-07-05 14:19:00 <gavinandresen> speaking of which... I updated the gist for that yesterday:  https://gist.github.com/2961409
600 2012-07-05 14:19:51 <BlueMatt> fee rework is 0.7?
601 2012-07-05 14:20:54 <gavinandresen> Maybe part of fee rework-- miners need to change their transaction-including logic before clients can start to rely on it
602 2012-07-05 14:21:24 <jgarzik> gavinandresen: raw tx api for 0.7, yes?
603 2012-07-05 14:21:40 <gavinandresen> So doing something fairly straightforward like sorting transactions first by fee-per-kb might be a 0.7 thing
604 2012-07-05 14:21:55 <gavinandresen> jgarzik: yes, raw tx api is ready to pull I think
605 2012-07-05 14:22:06 <BlueMatt> s/transactions/tx groups/
606 2012-07-05 14:22:29 <gmaxwell> gavinandresen: On fee rework, I hate to move the goalpost... though I do think we need to contemplate how ultraprune fits in with it. The updated fee schedule should incentivize reducing the txout set. Otherwise future improvements to make clients reduce the txout set size at the expense of a somewhat larger transaction may be resisted because they increase fees.
607 2012-07-05 14:22:32 <gavinandresen> maybe transaction groups, but that's more dangerous
608 2012-07-05 14:23:01 <BlueMatt> dangerous?
609 2012-07-05 14:23:12 <jgarzik> we can go ahead and start putting height into coinbase now, right?
610 2012-07-05 14:23:33 <gavinandresen> yes, dangerous, if there's a hidden "I can create a set of dependent transactions that make you perform O(n^2) work"
611 2012-07-05 14:23:34 <jgarzik> i.e. absent any block shunning/rejecting code
612 2012-07-05 14:23:41 <gmaxwell> jgarzik: not if we're going to use the height as a signal for a rule change, it would have to go hand in hand with the rule.
613 2012-07-05 14:23:51 <BlueMatt> gavinandresen: ah, performance dangerous, yea...that is an issue
614 2012-07-05 14:24:04 <jgarzik> oh yeah -- what about RPC locking?  table with 'unlocked' member, or lock push-down?
615 2012-07-05 14:24:29 <gavinandresen> consensus was lock push-down, but that'll mean I'll have some work to do in the raw transactions API
616 2012-07-05 14:24:32 <BlueMatt> push-down: makes it way easier to do shared locking
617 2012-07-05 14:24:47 <gavinandresen> BlueMatt: no, it doesn't.
618 2012-07-05 14:24:59 <BlueMatt> well, less work at least
619 2012-07-05 14:25:03 <gavinandresen> BlueMatt: if you want a shared lock, you just tell the table "don't lock for me, I'll do all my locking myself"
620 2012-07-05 14:25:34 <BlueMatt> and when so many rpc functions are shared lock, you end up with only a few trues in the table, which...well whatever
621 2012-07-05 14:26:28 <gavinandresen> BlueMatt: right, I lost that battle, I concede defeat.  ANd will tell you all "I told you so" when there's a subtle locking-related bug because a new RPC call forgets to obtain the right set of locks....
622 2012-07-05 14:27:17 <BlueMatt> fair enough
623 2012-07-05 14:29:22 <gmaxwell> sipa: the seperate blk and unds probably waste a fair bit of space.. on many (most?) file systems a file needs to consume at least one block.
624 2012-07-05 14:32:57 <gmaxwell> 07/05/12 16:28:02 SetBestChain: new best=00000000839a8e6886ab  height=1 work=8590065666
625 2012-07-05 14:33:00 <gmaxwell> 07/05/12 16:28:33 SetBestChain: new best=000000000003ba27aa20  height=100000 work=451709610344319134
626 2012-07-05 14:33:03 <gmaxwell> 07/05/12 16:31:57 SetBestChain: new best=0000000000000136be7d  height=164956 work=225916907792042738117
627 2012-07-05 14:33:06 <gmaxwell> sipa: and then it hits a reorg and stops.
628 2012-07-05 14:35:50 <gmaxwell> In any case, 165k blocks in 4 minutes is good.
629 2012-07-05 14:58:17 <ersi> gmaxwell: Success with locking adjustments?
630 2012-07-05 14:58:59 <gmaxwell> no, you wish. We don't appear to be bound by locking on initial sync. Thats ultraprune.
631 2012-07-05 14:59:26 <ersi> Aww, still cool though. Four fucking minutes.. that's nothing
632 2012-07-05 14:59:43 <BlueMatt> use a tmpfs, 4 minutes is easy
633 2012-07-05 15:02:09 <jgarzik> rpcrawtransaction.cpp
634 2012-07-05 15:02:16 <jgarzik> even a German could not invent a longer filename
635 2012-07-05 15:02:53 <gmaxwell> Where germans fail apply java developers!
636 2012-07-05 15:03:30 <jgarzik> RPCTypeCheck(params, list_of(int_type)(int_type));
637 2012-07-05 15:03:45 <jgarzik> an unusual number of parens...  probably a weird C++ template or macro thingy
638 2012-07-05 15:07:58 <jgarzik> gavinandresen: bleh
639 2012-07-05 15:08:07 <jgarzik> gavinandresen: you added a bunch of new compile warnings: http://pastebin.com/8bidY7Ez
640 2012-07-05 15:11:28 <ersi> jgarzik: You could add FactoryModel afterwards
641 2012-07-05 15:12:48 <gavinandresen> jgarzik: ack, I'll fix after I eat some lunch.  -Wall doesn't seem to be catching them in my dev environment for some reason
642 2012-07-05 15:13:47 <jgarzik> gavinandresen: probably a -Wno-sign-compare hidden in osx makefile somewhere
643 2012-07-05 15:14:22 <jgarzik> sipa: this is an example of a problem I mentioned a couple months ago...  excessive wallet flushing: http://pastebin.com/nwHNdRcM
644 2012-07-05 15:14:49 <jgarzik> just started git HEAD client, and downloaded a couple hundred blocks
645 2012-07-05 15:14:53 <jgarzik> several wallet flushes per minute
646 2012-07-05 15:15:09 <jgarzik> no wallet tx's in any of those blocks
647 2012-07-05 15:28:30 <gavinandresen> jgarzik: fixed the warnings, although I'm getting one in unrelated code: main.cpp:1920: warning: comparison between signed and unsigned integer expressions
648 2012-07-05 15:29:30 <jgarzik> gavinandresen: does the warning go away if you suffix 'L' onto the 0x7F000000 constant?
649 2012-07-05 15:29:59 <gavinandresen> jgarzik: no
650 2012-07-05 15:31:00 <jgarzik> gavinandresen: this is osx?  ftell() returns long on that platform, yes?
651 2012-07-05 15:31:32 <gavinandresen> according to 'man ftell', yes it returns long
652 2012-07-05 15:32:26 <jgarzik> owel, have to cast the entire RHS I guess
653 2012-07-05 15:35:44 <gavinandresen> if (ftell(file) < static_cast<long>(0x7F000000 - MAX_SIZE))    .... fixes it.  MAX_SIZE is unsigned.
654 2012-07-05 15:36:33 <jgarzik> yep
655 2012-07-05 15:37:03 <gavinandresen> ... maybe    if (ftell(file)+MAX_SIZE < 0x7F...) would be a better fix, though
656 2012-07-05 15:47:36 <jgarzik> gavinandresen: well, think about overflowing a long on 32-bit
657 2012-07-05 15:50:35 <gavinandresen> jgarzik: I'm not enough of a C/C++ language lawyer... is the result of    int + unsigned int    signed or unsigned ?
658 2012-07-05 15:55:04 <gmaxwell> unsigned.
659 2012-07-05 15:56:42 <gavinandresen> gmaxwell: thanks.  so ftell() returning signed long plus an unsigned MAX_SIZE cannot overflow, yes?
660 2012-07-05 15:58:38 <gmaxwell> So long as ftell()+MAX_SIZE can't be over ULONG_MAX.
661 2012-07-05 15:58:54 <gavinandresen> yeah, can't be, const unsigned int MAX_SIZE = 0x02000000;
662 2012-07-05 15:58:57 <drizztbsd> ftello can use 64bit :P
663 2012-07-05 16:03:27 <jgarzik> gavinandresen: well, ULONG_MAX is quite small on 32-bit :)
664 2012-07-05 16:03:54 <jgarzik> gavinandresen: ftell() + MAX_SIZE can conceivably overflow
665 2012-07-05 16:04:15 <jgarzik> you'd have to construct a block file Of Unusual Size, I imagine
666 2012-07-05 16:05:42 <jgarzik> probably why Satoshi wrote it in the safer manner (ftell < (x - y))
667 2012-07-05 16:05:58 <jgarzik> ftell itself can return ULONG_MAX
668 2012-07-05 16:06:20 <gavinandresen> ftell returns unsigned longs on some systems?
669 2012-07-05 16:06:54 <gavinandresen> (this is why I'm not qualified to fix that warning, by the way....)
670 2012-07-05 16:07:15 <jgarzik> ahh, sorry.  my mistake.  yes, ftell can return up to LONG_MAX.
671 2012-07-05 16:09:48 <jgarzik> programmers are apparently hardwired to write "if (address + offset > boundary)" which is not so safe
672 2012-07-05 16:11:12 <gmaxwell> Good thing there are nice guides for this stuff.
673 2012-07-05 16:11:28 <gmaxwell> e.g. https://www.securecoding.cert.org/confluence/display/seccode/INT02-C.+Understand+integer+conversion+rules  (though be warned, my fuzzy memory said I didn't like something about their advice)
674 2012-07-05 16:12:58 <gavinandresen> gmaxwell: excellent, thanks.
675 2012-07-05 16:21:07 <luke-jr> gmaxwell: some of the rules seem non-intuitive to me, so I tend to always be explicit when I'm even slightly unsure
676 2012-07-05 16:22:09 <jgarzik> I wrote a C compiler and still don't know the rules by heart :)
677 2012-07-05 16:22:30 <gmaxwell> luke-jr: it really really helps if you've programmed in C on a 16 bit machine. The promotion to unsigned makes a ton more sense then. Otherwise you have constant problems with loop counters overflowing.
678 2012-07-05 16:31:56 <forrestv> arbitrary precision math ftw
679 2012-07-05 16:32:28 <luke-jr> forrestv: if only that didn't come with implied lossyness :/
680 2012-07-05 16:34:11 <forrestv> eh? no. not floating point, but like perl's bignum or python's longs
681 2012-07-05 16:41:46 <luke-jr> right, .. or bitcoin's CBigNum <.<
682 2012-07-05 16:49:06 <BlueMatt> jgarzik: can we get lock push-down rebased, add the one missing cs_vNodes and merged? I wanna build on it but it doesnt apply on master atm...
683 2012-07-05 16:51:50 <BlueMatt> yay! 60 open pulls...
684 2012-07-05 17:08:44 <gribble> New news from bitcoinrss: TheBlueMatt opened issue 1560 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/issues/1560>
685 2012-07-05 17:11:15 <luke-jr> got a crash, looks like related to CBlockStore
686 2012-07-05 17:11:17 <luke-jr> BlueMatt:
687 2012-07-05 17:11:33 <luke-jr> maybe not though, looking into it more
688 2012-07-05 17:11:39 <luke-jr> mapBlockIndex seems to have a NULL in it
689 2012-07-05 17:13:45 <BlueMatt> I did see a bug like that in cblockstore, though iirc i fixed it...
690 2012-07-05 17:13:58 <luke-jr> ah
691 2012-07-05 17:14:04 <luke-jr> my code may be old
692 2012-07-05 17:14:16 <luke-jr> happen to know the fix offhand?
693 2012-07-05 17:14:27 <BlueMatt> no, I really cant remember wtf was going on there
694 2012-07-05 17:15:38 <luke-jr> XD
695 2012-07-05 17:16:23 <luke-jr> well, I think I see a bug in autoprune
696 2012-07-05 17:16:31 <luke-jr> which isn't the problem I hit, since I don't have it enabled
697 2012-07-05 17:17:03 <luke-jr> CTxDB::PruneBlockIndex doesn't check that hashPruneFrom is in mapBlockIndex before using [] to get it
698 2012-07-05 17:17:23 <luke-jr> and main.cpp calls it with that set to the last checkpoint, even if we don't have the block yet
699 2012-07-05 17:17:26 <BlueMatt> actually that might have been it
700 2012-07-05 17:17:39 <BlueMatt> no, it may have been a [] in net
701 2012-07-05 17:17:55 <BlueMatt> oh, wait, heres the problem
702 2012-07-05 17:18:29 <BlueMatt> https://github.com/TheBlueMatt/bitcoin/commit/235a0277cfebf4dd65e40070c13f82ac5b6eb6ff
703 2012-07-05 17:18:31 <BlueMatt> that was it
704 2012-07-05 17:18:37 <BlueMatt> "This fixes a potential segfault if -blockbuffersize is overly big."
705 2012-07-05 17:19:00 <BlueMatt> (only happened if your -blockbuffersize was > than the number of blocks getting returned by a remote node
706 2012-07-05 17:20:12 <BlueMatt> luke-jr: huh? PruneBlockIndex's hashPruneFrom and hashPruneTo are always in mapBlockIndex
707 2012-07-05 17:20:29 <luke-jr> BlueMatt: main.cpp seems to call it with the last checkpoint in hashPruneFrom
708 2012-07-05 17:20:51 <BlueMatt> no it calls it with the previous checkpoint...or should
709 2012-07-05 17:21:23 <BlueMatt> it calls it with pPrevCheckpoint
710 2012-07-05 17:21:45 <BlueMatt> which is the most recent checkpoint in mapBlockIndex before we commit that block
711 2012-07-05 17:22:07 <BlueMatt> (to mapBlockIndex, that is)
712 2012-07-05 17:22:33 <luke-jr> pPrevCheckpoint = Checkpoints::GetLastCheckpoint();
713 2012-07-05 17:22:39 <BlueMatt> where?
714 2012-07-05 17:22:42 <luke-jr> Checkpoints::GetLastCheckpoint returns the last checkpoint period
715 2012-07-05 17:22:53 <luke-jr> main.cpp ; remember though, I have older code
716 2012-07-05 17:22:58 <BlueMatt> which prune, there are two versions, the prune pull and the one in parallelprune
717 2012-07-05 17:23:04 <BlueMatt> (aka remove tons of GetHash()s)
718 2012-07-05 17:23:18 <luke-jr> parallelprune IIRC
719 2012-07-05 17:23:34 <BlueMatt> look at the src for GetLaslCheckpoint
720 2012-07-05 17:23:45 <luke-jr> I did
721 2012-07-05 17:23:47 <luke-jr> return checkpoints.rbegin()->second;
722 2012-07-05 17:23:48 <BlueMatt> it caches checkpoints as they come in using a listener for CommitBlock callbacks
723 2012-07-05 17:24:15 <luke-jr> not the one I have :x
724 2012-07-05 17:24:52 <BlueMatt> heh wtf did I do here?
725 2012-07-05 17:25:06 <luke-jr> >_<
726 2012-07-05 17:25:15 <BlueMatt> heh well thats very broken
727 2012-07-05 17:25:30 <BlueMatt> anywhoo, doesnt matter, Im gonna let that prune stuff die until sipa's ultraprune stuff is done
728 2012-07-05 17:25:45 <BlueMatt> its not gonna get merged, so itl just die
729 2012-07-05 17:26:21 <BlueMatt> thats really odd though, I know Ive benchmarked that branch so many times and had no problems pruning...
730 2012-07-05 17:26:27 <BlueMatt> whatever
731 2012-07-05 17:30:55 <luke-jr> is there a way to print a uint256 in gdb? :/
732 2012-07-05 17:32:02 <BlueMatt> ToString().c_str()
733 2012-07-05 17:32:53 <BlueMatt> (at this point Im thinking chub/cblockstore/all the groundwork for spv mode/threading/etc is gonna end up dying too, but...whatever)
734 2012-07-05 17:33:07 <luke-jr> BlueMatt: gdb doesn't like that
735 2012-07-05 17:33:20 <BlueMatt> really? I know Ive done that before
736 2012-07-05 17:33:23 <luke-jr> (gdb) printf "%s\n", block.hashPrevBlock.ToString().print()
737 2012-07-05 17:33:24 <luke-jr> Attempt to take address of value not located in memory.
738 2012-07-05 17:33:28 <luke-jr> err
739 2012-07-05 17:33:32 <luke-jr> it does that with c_str too ;)
740 2012-07-05 17:33:41 <luke-jr> or anything
741 2012-07-05 17:33:44 <luke-jr> even non-existent methods
742 2012-07-05 17:33:55 <BlueMatt> just print block.hashPrevBlock.ToString().c_str()
743 2012-07-05 17:34:00 <BlueMatt> I know Ive done that
744 2012-07-05 17:34:16 <BlueMatt> oh, block isnt in memory
745 2012-07-05 17:34:21 <BlueMatt> well, have fun
746 2012-07-05 17:34:28 <luke-jr> rather, ToString returns a temporary - I think
747 2012-07-05 17:34:47 <BlueMatt> can you even print block.hashPrevBlock?
748 2012-07-05 17:35:45 <luke-jr> yes
749 2012-07-05 17:35:50 <luke-jr> surprisingly
750 2012-07-05 17:36:00 <BlueMatt> old version of gdb or something?
751 2012-07-05 17:36:10 <BlueMatt> I know for sure Ive done a print .ToString.c_str()
752 2012-07-05 17:36:27 <BlueMatt> can you print .ToString()
753 2012-07-05 17:39:09 <gribble> New news from bitcoinrss: TheBlueMatt opened issue 1561 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/issues/1561>
754 2012-07-05 17:40:10 <luke-jr> hum, debug.log hit 2 GB and just stopped working
755 2012-07-05 17:40:20 <luke-jr> BlueMatt: no
756 2012-07-05 17:40:30 <luke-jr> gdb 7.3.1
757 2012-07-05 17:40:35 <BlueMatt> god, we have a lot of structures that dont get locked that should
758 2012-07-05 17:41:03 <luke-jr> $65 = {<base_uint<256u>> = {pn = {414994827, 820876553, 44801826, 2014180509, 3375580397, 1423085369, 868, 0}}, <No data fields>}
759 2012-07-05 17:41:07 <luke-jr> that's good enough
760 2012-07-05 17:41:27 <luke-jr> 0000000000000364 is the one I don't have but is in mapBlockIndex
761 2012-07-05 17:41:30 <luke-jr> (as NULL)
762 2012-07-05 17:41:58 <BlueMatt> do you have that commit I just linked in your git log?
763 2012-07-05 17:43:12 <Azureus777> Hello all!
764 2012-07-05 17:43:41 <luke-jr> BlueMatt: no, it doesn't cherry-pick cleanly either
765 2012-07-05 17:43:50 <luke-jr> first mention:   got inventory: block 0000000000000364909e  have
766 2012-07-05 17:44:05 <BlueMatt> luke-jr: well, get that commit in there and see if you still see it
767 2012-07-05 17:44:17 <BlueMatt> because that commit was written specifically to address the NULL in mapBlockIndex
768 2012-07-05 17:45:28 <luke-jr> 385f730f31ab7218d8c0fcc76f5c7a2b35783c6a (sipa/unstuck, origin-pull/1315/head)
769 2012-07-05 17:45:32 <luke-jr> that one looks suspicious to me
770 2012-07-05 17:46:09 <BlueMatt> oh, yea most of the chub stuff is out of date
771 2012-07-05 17:46:16 <BlueMatt> if I were you, Id drop it from next-test
772 2012-07-05 17:46:35 <jgarzik_> Let's test hospital wifi with a bitcoin node...
773 2012-07-05 17:46:45 <BlueMatt> ouch? why you in hospital?
774 2012-07-05 17:46:58 <BlueMatt> also...you need some tethering
775 2012-07-05 17:47:10 <jgarzik_> Boring preggo tests for the wifey
776 2012-07-05 17:47:22 <BlueMatt> ah
777 2012-07-05 17:48:44 <Azureus777> hello all, i need a little help with alternative bitcoin block chain, can anyone help me?
778 2012-07-05 17:49:25 <BlueMatt> but...seriously get you some tethering, dunno about Raleigh, but in chapel hill (a launch market) sprint wimax 4g is actually really fast (probably because Im the only one using it but...)
779 2012-07-05 17:49:46 <BlueMatt> Azureus777: ask a question, dont ask to ask
780 2012-07-05 17:49:53 <Azureus777> :)
781 2012-07-05 17:50:17 <jgarzik_> Bluematt: tethering costs money w Sprint, I think
782 2012-07-05 17:50:25 <Hunner> Verizon 4G LTE iPad offers tethering. $10 per 2GB per month if you add it to an existing plan
783 2012-07-05 17:50:28 <Azureus777> i am trying to create my own bitcoin like cryptocurrency, i am newbie, so please help me, i cant find source code where i can make changes for my new currency
784 2012-07-05 17:50:28 <BlueMatt> jgarzik_: not if you root your phone...
785 2012-07-05 17:50:34 <jgarzik_> Heh
786 2012-07-05 17:50:40 <BlueMatt> Hunner: I use waaayyy more than 2g/month
787 2012-07-05 17:50:41 <luke-jr> BlueMatt: contract law?
788 2012-07-05 17:50:59 <Hunner> BlueMatt: Yeah, but it's just multiples of 2GB, not limited
789 2012-07-05 17:51:17 <luke-jr> Azureus777: just don't.
790 2012-07-05 17:51:25 <Hunner> (And tethering when you need it, not as a primary internet connection)
791 2012-07-05 17:51:29 <BlueMatt> Hunner: also, I dont have an existing plan with verizon, and buying one would mean more cost than just using my single sprint phone with unlimited 4g...
792 2012-07-05 17:51:37 <Azureus777> but i really need it, i cant sleep at nights:))
793 2012-07-05 17:51:40 <Hunner> BlueMatt: Ah yep. That would do it :)
794 2012-07-05 17:51:54 <Hunner> Unlimited... right
795 2012-07-05 17:51:54 <jgarzik_> Azur: premine a bunch of coins! All the cool kids do it
796 2012-07-05 17:52:17 <luke-jr> I have unlimited T-Mobile.
797 2012-07-05 17:52:30 <BlueMatt> luke-jr: meh, fuck 'em, I use a vpn all the time when Im tethered (to my vps) so they cant tell anything more than I like to use a lot of bw on a single ssl connection on port 443...
798 2012-07-05 17:52:52 <BlueMatt> luke-jr: they dont sell that anymore though, do they?
799 2012-07-05 17:53:09 <luke-jr> BlueMatt: they kicked me off that plan; I have to pay $30/mo now
800 2012-07-05 17:53:12 <Azureus777> i am not a cool kid, i need some advices, i cant find anything on internet, so i decide to ask specialists here
801 2012-07-05 17:53:14 <luke-jr> (old plan was $10 every 3 months)
802 2012-07-05 17:53:23 <jgarzik_> Poof
803 2012-07-05 17:53:30 <BlueMatt> luke-jr: wtf? 30$/mo with unlimited data???
804 2012-07-05 17:53:34 <luke-jr> BlueMatt: yes
805 2012-07-05 17:53:41 <BlueMatt> how the hell did you get that deal?
806 2012-07-05 17:53:49 <BlueMatt> been with them for 20 years or something?
807 2012-07-05 17:54:16 <luke-jr> BlueMatt: no, it's a normal plan "exclusive" to Walmart and T-Mobile.com
808 2012-07-05 17:54:21 <luke-jr> http://prepaid-phones.t-mobile.com/prepaid-plans
809 2012-07-05 17:54:49 <BlueMatt> "First 5GB at up to 4G speeds." darn, that excludes me
810 2012-07-05 17:55:05 <luke-jr> I never get to 1 GB afaik
811 2012-07-05 17:55:18 <BlueMatt> I probably average like 5
812 2012-07-05 17:55:30 <luke-jr> meh, cellular is slow anyway
813 2012-07-05 17:55:43 <luke-jr> I probably wouldn't notice a difference after 4 GB
814 2012-07-05 17:55:44 <luke-jr> 5*
815 2012-07-05 17:55:59 <luke-jr> BlueMatt: also has native IPv6
816 2012-07-05 17:56:09 <BlueMatt> ooo, that I like, but so does my vpn so...meh
817 2012-07-05 17:56:14 <luke-jr> lol
818 2012-07-05 17:56:35 <BlueMatt> hey, gonna tether you have to use a vpn, at least on verizon and who was the other one? tmo?
819 2012-07-05 17:56:56 <luke-jr> I just put my SIM in my PC
820 2012-07-05 17:57:13 <luke-jr> if I need a bigger PC, I use iptables
821 2012-07-05 17:57:15 <BlueMatt> yea...you still have to use a vpn
822 2012-07-05 17:57:19 <luke-jr> &
823 2012-07-05 17:57:23 <luke-jr> why?
824 2012-07-05 17:57:44 <BlueMatt> for quite a while they were using dpi to catch unauthorized tethers and canceling accounts
825 2012-07-05 17:57:45 <Azureus777> any source codes i could use?
826 2012-07-05 17:57:55 <BlueMatt> probably http headers, Id guess
827 2012-07-05 17:58:12 <BlueMatt> I know verizon was doing it, and there was one other, but Im not sure who the other was
828 2012-07-05 17:58:26 <luke-jr> BlueMatt: so how do you tell the difference between Chromium on the laptop I have the SIM in, and Chromium on the laptop I've iptables'd?
829 2012-07-05 17:58:55 <luke-jr> Azureus777: it's bad enough that people make scams out of Bitcoin in the first place, don't expect someone to hold your hand to do ti
830 2012-07-05 17:58:57 <BlueMatt> what? the difference they care about is if you are using a mobile browser or a desktop browser
831 2012-07-05 17:59:19 <luke-jr> BlueMatt: I use a desktop browser on the laptop the SIM is in
832 2012-07-05 17:59:41 <BlueMatt> luke-jr: and you pay for access to their network on a mobile phone, not for access on a desktop
833 2012-07-05 17:59:50 <luke-jr> BlueMatt: there is no difference
834 2012-07-05 17:59:54 <BlueMatt> whether the sim is in it or not
835 2012-07-05 17:59:58 <BlueMatt> there certainly is to them
836 2012-07-05 18:00:05 <BlueMatt> technically, there isnt, but they dont care
837 2012-07-05 18:00:46 <luke-jr> their terms only talk about tethering
838 2012-07-05 18:00:55 <luke-jr> not using from desktop apps
839 2012-07-05 18:01:04 <luke-jr> and they know I use my SIM in a computer
840 2012-07-05 18:01:12 <BlueMatt> then take them to court, they will use the "we can terminate you for any reason for whatever we want" clause and screw you anyway
841 2012-07-05 18:01:21 <luke-jr> shrug
842 2012-07-05 18:01:22 <Azureus777> i dont have any intentions to do scam, i am honest guy, i just like this stuff, so i need some help
843 2012-07-05 18:01:22 <luke-jr> they haven't
844 2012-07-05 18:01:30 <BlueMatt> meh, it might not have been tmo
845 2012-07-05 18:01:32 <luke-jr> and if they did, I'd just sign up a new account?
846 2012-07-05 18:01:33 <BlueMatt> may have been att then
847 2012-07-05 18:01:42 <luke-jr> Azureus777: making a scamcoin = scam
848 2012-07-05 18:02:12 <BlueMatt> Azureus777: if you have a specific reason why you want to make a chain-based cryptocurrency you may get help, "because I want to" isnt really one
849 2012-07-05 18:02:48 <BlueMatt> luke-jr: if they cancel your account, they wont let you just sign up for a new one the next day, they do have your name...
850 2012-07-05 18:02:53 <Azureus777> i have to explain somehow what my intention so i decide to write "i want to", who cares about that?
851 2012-07-05 18:02:54 <luke-jr> BlueMatt: no, they don't.
852 2012-07-05 18:03:06 <luke-jr> BlueMatt: I keep my TMo anonymous
853 2012-07-05 18:03:26 <BlueMatt> heh, nice
854 2012-07-05 18:04:40 <Azureus777> i need some help, thats why i come here, i came here a months ago with the same question, but except dissapointment i got nothing, i hope that this time i will be lucky and someone will explain to me the principles of how to do this, from what to start, which library to use, which part of code need to be edited to get my result, i really hope someone can help me today.
855 2012-07-05 18:05:55 <luke-jr> Azureus777: you're not supposed to do it, so the code isn't designed to make it easy, and nobody is likely to help you unless they're also a scammer
856 2012-07-05 18:06:06 <gmaxwell> Azureus777: I don't think many people want to see a new crypto currency created by someone who can't figure out how to do it on their own.
857 2012-07-05 18:06:24 <gmaxwell> It can be an interesting learning expirence, sure, but you won't learn if someone just tells you how.
858 2012-07-05 18:06:57 <gmaxwell> I would help you if I could answer your questions off the top of my head but I'm sure I can't... and I'm not interested in spending time researching this for other people.
859 2012-07-05 18:07:13 <Azureus777> I repeat i am not scammer ok? I am not doing this to become as famous as bitcoin is, so dont be affraid ok? And stop calling me scammer ok? i dont think this is ethical
860 2012-07-05 18:08:57 <BlueMatt> I dont think anyone is required to spend their time researching, thinking about, and doing everything but writing the actual code for your project
861 2012-07-05 18:09:01 <gmaxwell> Azureus777: luke will call anyone creating an alt currency a scammer by default, at least if it doesn't offer something clearly useful. Don't take it personally.
862 2012-07-05 18:09:02 <BlueMatt> or interested in
863 2012-07-05 18:09:55 <gmaxwell> (and scam intentions or not, a lot of alt currencies are just worthless efforts that move funds around for no real gain for mainkind. They're usually pretty boring, so I don't think luke's cynicism is unjustified, though he could be a little more clear about it)
864 2012-07-05 18:10:00 <Azureus777> by default? i dont think it is correct to name people by default, anyway i see i am getting disapointed again:(
865 2012-07-05 18:10:05 <gmaxwell> s/mainkind/mankind/
866 2012-07-05 18:11:05 <Hunner> Most things come out of being curious and tenacious. Why not assume that is a possible motive?
867 2012-07-05 18:12:08 <gmaxwell> The motivations aren't really in question in any case. "Sorry, but we're not going to do your homework for you!" is sufficient.