1 2013-03-04 00:00:01 <MagicalTux> [07:54:27] <sipa> gmaxwell: we'd use IP-based blocking! you'll never get around that! <- actually spambots get around that
   2 2013-03-04 00:00:07 <doublec> Luke-Jr: then they'd get requests to hve a mode that stops when it reaches the highest balance...
   3 2013-03-04 00:00:18 <Luke-Jr> doublec: XD
   4 2013-03-04 00:00:20 <gmaxwell> MagicalTux: yea, sipa was being sarcastic. :P
   5 2013-03-04 00:00:21 <sipa> MagicalTux: i would hope that you of all people would see the irony :)
   6 2013-03-04 00:00:21 <TD> MagicalTux: i think he was kidding ;)
   7 2013-03-04 00:00:43 <MagicalTux> I've just read the part where you hilighted me, so I was missing context
   8 2013-03-04 00:00:49 <sipa> ha :)
   9 2013-03-04 00:00:59 <sipa> petertodd: anyway, my worry is not transactions in the mempool
  10 2013-03-04 00:01:10 <sipa> petertodd: my worry is transactions in a block that gets disconnected
  11 2013-03-04 00:01:12 <TD> Luke-Jr: no, it doesn't fundamentally change the 1 tx 1 bet model. that'd be too specific to SD
  12 2013-03-04 00:01:32 <TD> MagicalTux: i was wondering if you were going to comment on my question about fatca or if that's pretty much off limits to discuss publicly ...
  13 2013-03-04 00:01:39 <sipa> petertodd: as those may not be able to enter the mempool again
  14 2013-03-04 00:02:11 <Luke-Jr> TD: not really.. especially if implemented as a "connect, send deposit, multiple interactions, request refund of remaining balance"
  15 2013-03-04 00:02:31 <gmaxwell> Luke-Jr: perhaps $someone should offer to write evoorhees a SD replacement that does what you described. It has been suggested that sane behavior is being inhibited by technical staff charging for efforts.
  16 2013-03-04 00:02:35 <TD> Luke-Jr: well sure. they could hold deposits for their users. they could do that today. they feel that one of the reasons they got so huge is they don't do that.
  17 2013-03-04 00:02:36 <TD> it's a feature
  18 2013-03-04 00:02:36 <petertodd> sipa: Ah, right, that same discussion we had. Well then why not have a "nBestHeightEverSeen" that can only increase, and use that as the basis for the IsFinal() check?
  19 2013-03-04 00:02:50 <TD> gmaxwell: i just theorized. i don't know what arrangement they have
  20 2013-03-04 00:03:03 <Luke-Jr> TD: if it's done transparently so users don't notice..
  21 2013-03-04 00:03:12 <gmaxwell> TD: thus suggested, — well, I thought that too. FWIW.
  22 2013-03-04 00:03:18 <Luke-Jr> gmaxwell: too bad $someone can't be anyone in the US without stupid liability
  23 2013-03-04 00:03:21 <sipa> petertodd: or disable IsStandard for transactions being disconnected
  24 2013-03-04 00:03:31 <gmaxwell> (I also believed that to be the case for some of the past large mining pools)
  25 2013-03-04 00:03:33 <MagicalTux> [08:43:12] <TD> MagicalTux: i was wondering if you were going to comment on my question about fatca or if that's pretty much off limits to discuss publicly ... <- we can't really discuss this publicly
  26 2013-03-04 00:03:49 <TD> MagicalTux: ok, thanks anyway
  27 2013-03-04 00:04:11 <petertodd> sipa: Yeah, I'm a bit dubious about doing that, because that behavior makes it easier to recover from a non-standard transaction that causes a fork.
  28 2013-03-04 00:04:23 <sipa> uhu, agree
  29 2013-03-04 00:04:30 <gmaxwell> sipa: it's .. really hard for the longer chain to be lower in height. Do we really care if isstandard is consistent around that case?
  30 2013-03-04 00:04:44 d4de has quit (Ping timeout: 245 seconds)
  31 2013-03-04 00:04:50 <petertodd> sipa: I think we're better off only disabling the IsFinal() check.
  32 2013-03-04 00:05:03 <petertodd> sipa: Or moving it out of IsStandard()
  33 2013-03-04 00:05:20 <sipa> gmaxwell: well, i think bitcoin is designed to never ever let a confirmed transaction get removed from the chain, absent any attack
  34 2013-03-04 00:05:48 d4de has joined
  35 2013-03-04 00:06:09 <gmaxwell> hm. okay. Good point. ugh. but thats a great argument for isstandard not applying at the reorg.
  36 2013-03-04 00:06:19 <sipa> anyway, i like the idea of using nLockTime in practice, but i feared this could challenge that rule
  37 2013-03-04 00:07:18 <gmaxwell> I still like my transactions-are-checkpoints-for-fees notion. :P too bad there isn't a softforking way to implement it.
  38 2013-03-04 00:07:24 <petertodd> gmaxwell: I'd be interested to know if blockchain.info stores transactions related to your wallet when they see them. Unfortunately while the ref client does, I wouldn't be surprised if a lot of other wallets make the assumption that if a transaction is broadcast to the network, it's guaranteed to get confirmed.
  39 2013-03-04 00:07:40 <petertodd> gmaxwell: what's that one?
  40 2013-03-04 00:08:38 <gmaxwell> petertodd: thats where every txn includes a [height][truncated block hash] as part of the signed data. And the fees for the txn can only be collected in blockchains where the hash matches at the specified height. Otherwise the txn can still be mined, but the fees are burnt.
  41 2013-03-04 00:09:02 <gmaxwell> and then wallets would automatically include a checkpoint in every transaction for some height a few blocks back.
  42 2013-03-04 00:09:03 <petertodd> gmaxwell: Ah, right, yes I like that idea.
  43 2013-03-04 00:09:43 coblee_ has joined
  44 2013-03-04 00:09:46 <gmaxwell> so an attacker loses the fees for the block consensus they've undone.
  45 2013-03-04 00:10:06 <sipa> oh, nice
  46 2013-03-04 00:10:27 <gmaxwell> we could almost have it, except the scriptsig isn't signed.
  47 2013-03-04 00:11:22 <gmaxwell> variations are possible, e.g. only get half the fees or something, so there is still an incentive to included it... dunno how to reason about that.
  48 2013-03-04 00:11:33 <petertodd> gmaxwell: How would signing the scriptSig allow that?
  49 2013-03-04 00:11:52 <sipa> ;;genrate 200
  50 2013-03-04 00:11:53 <gribble> The expected generation output, at 200.0 Mhps, given difficulty of 4367876.00084, is 0.0230275088548 BTC per day and 0.000959479535618 BTC per hour.
  51 2013-03-04 00:12:24 freakazoid has joined
  52 2013-03-04 00:12:28 <gmaxwell> petertodd: we could just insert that data into the scriptsig. Then make a softforking rule to burn those fees when it doesn't match.
  53 2013-03-04 00:12:56 coblee has quit (Ping timeout: 252 seconds)
  54 2013-03-04 00:12:56 coblee_ is now known as coblee
  55 2013-03-04 00:14:12 <petertodd> gmaxwell: Right, so equally you could do it with another signature around that data in the scriptSig, and the rule needs a way to determine if a client requests that behavior.
  56 2013-03-04 00:14:25 <petertodd> gmaxwell: Do it when some ECDSA opcodes or the like get added.
  57 2013-03-04 00:14:35 <gmaxwell> that would make it a lot bigger though.
  58 2013-03-04 00:15:11 <petertodd> Yeah, but the extra size can be removed later with a second soft fork. (IE, version 1 to 3 is a hard-fork, 1-2 and 2-3 is soft)
  59 2013-03-04 00:15:12 <sipa> and it's data that needs to be maintained in the UTXO set
  60 2013-03-04 00:16:04 <gmaxwell> sipa: hm? scriptsigs are not in the utxo set.
  61 2013-03-04 00:16:17 <sipa> ehm... brainfart
  62 2013-03-04 00:16:30 TD has quit (Quit: TD)
  63 2013-03-04 00:16:52 <gmaxwell> but yes, extra blockchain data. (though thats also why I said truncated— I'd assumed they'd be just 32 bit chunks)
  64 2013-03-04 00:17:21 <petertodd> gmaxwell: Another ugly way to do it would be to hi-jack the sequence numbers for this...
  65 2013-03-04 00:18:15 <gmaxwell> yuck
  66 2013-03-04 00:18:53 <petertodd> Heh, they are the one bit of (nearly) free-form data that is signed.
  67 2013-03-04 00:18:55 bitcoinbulletin has quit (Quit: bitcoinbulletin)
  68 2013-03-04 00:19:06 TD has joined
  69 2013-03-04 00:19:50 <gmaxwell> I feel like replacement is one of the few places the system was strangely overoptimistic about miner honesty.
  70 2013-03-04 00:20:10 <petertodd> Yeah, frankly I can't see replacement ever working in practice.
  71 2013-03-04 00:20:36 <petertodd> We'll soon wind up with "Assurance Pool", so you can "assure" your transactions get mined...
  72 2013-03-04 00:23:13 <sipa> i'm not sure that can be avoided
  73 2013-03-04 00:23:37 <petertodd> Me neither, especially as the subsidy falls.
  74 2013-03-04 00:23:41 <gmaxwell> sure, but thats why replacement is .. not very valuable.
  75 2013-03-04 00:24:09 <gmaxwell> E.g. if you can't count on miners to pay attention, you can only use replacement where no one would have an incentive to force the mining of an older txn.
  76 2013-03-04 00:24:21 <petertodd> (if you didn't get the joke, "Assurance Pool" only assures your transactions, rather than an "unfortunate" double-spend gets mined...)
  77 2013-03-04 00:24:21 <gmaxwell> (or a conflicting txn)
  78 2013-03-04 00:24:47 <sipa> joke?
  79 2013-03-04 00:24:50 <petertodd> Such a pool could even use non-fee payment too, like hash tokens and the like.
  80 2013-03-04 00:25:21 <petertodd> sipa: You wouldn't need the assurance if not for Assurance Pool... like mafia protection money.
  81 2013-03-04 00:26:15 <gmaxwell> external fees are lame. one thing I wish could be done would be that all the txn that go into a block pay their fees into a fee pool, and then the subsidy takes half the fee pool. (the remainder is passed on to following blocks).. this would much better reflect paying for burrying a txn.
  82 2013-03-04 00:26:36 <gmaxwell> sadly, people would just make their fees 0 and pay for fees externally or with special outputs.
  83 2013-03-04 00:26:59 <petertodd> gmaxwell: Carrying fees forward is something that really should have been done.
  84 2013-03-04 00:27:02 <gmaxwell> (luke even implemented taking fees in the form of a special output at one point!)
  85 2013-03-04 00:27:07 <gmaxwell> petertodd: I just explained why it can't be.
  86 2013-03-04 00:27:32 <gmaxwell> at least not assuming optimal greedy-rational miners.
  87 2013-03-04 00:28:37 bitcoinbulletin has joined
  88 2013-03-04 00:28:47 <petertodd> gmaxwell: I'm not convined there. The value to person sending the tx is less if only one miner gets the fee than if more than one, so even with out-of-band fee payment, people will still have reasons to use carry-forward fees.
  89 2013-03-04 00:28:57 <petertodd> *convinced
  90 2013-03-04 00:30:20 <gmaxwell> I mean, I suppose that when subsidy is small miners would prefer to extend the chain with the most carry forward available.
  91 2013-03-04 00:31:38 <petertodd> No, I think it'd be large miners who would have the biggest incentives there. They're the ones most likely to get those carry forward fees, so they can take higher risks re: orphans and so on.
  92 2013-03-04 00:32:02 <petertodd> Although with small blocks maybe the risks are so low, and fixed, that the strategy is the same for everyone.
  93 2013-03-04 00:32:05 <ProfMac> ;;genrate 10
  94 2013-03-04 00:32:06 <gribble> The expected generation output, at 10.0 Mhps, given difficulty of 4367876.00084, is 0.00115137544274 BTC per day and 4.79739767809e-05 BTC per hour.
  95 2013-03-04 00:35:16 <ProfMac> do the miners choose which tx they include in their blocks?
  96 2013-03-04 00:35:46 <gmaxwell> Of course.
  97 2013-03-04 00:36:34 <gmaxwell> Or do you propose a consensus system to decide which they must include?  Oh I know, we could use a blockchain.  hm.. how will the metaminers decide which txn to include? :P
  98 2013-03-04 00:37:32 <petertodd> How about we encrypt every transaction with time-lock crypto, so only after the tx is included in a block can it be decrypted and validated?
  99 2013-03-04 00:37:51 <petertodd> ...and the PoW can be a timelock decryptor...
 100 2013-03-04 00:40:21 CodeShark has joined
 101 2013-03-04 00:49:50 swappermall has quit (Read error: Connection reset by peer)
 102 2013-03-04 00:52:55 swappermall has joined
 103 2013-03-04 00:53:15 swappermall is now known as Guest79014
 104 2013-03-04 00:55:11 coolsa has quit (Read error: Connection reset by peer)
 105 2013-03-04 00:56:30 coolsa has joined
 106 2013-03-04 00:58:15 TD has quit (Quit: TD)
 107 2013-03-04 01:16:04 RazielZ has quit (Ping timeout: 246 seconds)
 108 2013-03-04 01:18:45 chicagobitcoins has quit (Remote host closed the connection)
 109 2013-03-04 01:19:07 hattorihanzo has joined
 110 2013-03-04 01:19:10 freakazoid has quit (Ping timeout: 245 seconds)
 111 2013-03-04 01:22:28 Diablo-D3 has joined
 112 2013-03-04 01:31:23 emryss has joined
 113 2013-03-04 01:32:03 emryss has quit (Read error: Connection reset by peer)
 114 2013-03-04 01:32:49 emryss has joined
 115 2013-03-04 01:34:20 emryss has quit (Client Quit)
 116 2013-03-04 01:34:24 dvide has quit (Ping timeout: 250 seconds)
 117 2013-03-04 01:35:24 bock has quit (Quit: Verlassend)
 118 2013-03-04 01:37:06 yoast has joined
 119 2013-03-04 01:39:13 Guest79014 has quit (Read error: Connection reset by peer)
 120 2013-03-04 01:39:36 yoast has left ()
 121 2013-03-04 01:39:53 MobGod has joined
 122 2013-03-04 01:39:54 MobGod has quit (Changing host)
 123 2013-03-04 01:39:54 MobGod has joined
 124 2013-03-04 01:48:21 toffoo has joined
 125 2013-03-04 01:56:30 wladston has joined
 126 2013-03-04 02:01:32 abrkn has joined
 127 2013-03-04 02:04:43 wladston has left ()
 128 2013-03-04 02:05:23 <Quazgaa> trying to understand the output of bfgminer... would the "Best share" stat equal "Diff" for a solved block
 129 2013-03-04 02:07:16 <Luke-Jr> >=
 130 2013-03-04 02:08:50 edwincheese has quit (Read error: Connection reset by peer)
 131 2013-03-04 02:09:58 edwincheese has joined
 132 2013-03-04 02:15:48 <Quazgaa> ah
 133 2013-03-04 02:18:54 d4de has quit (Ping timeout: 245 seconds)
 134 2013-03-04 02:20:28 d4de has joined
 135 2013-03-04 02:28:13 emryss has joined
 136 2013-03-04 02:34:57 edwincheese has quit (Read error: Connection reset by peer)
 137 2013-03-04 02:35:46 edwincheese has joined
 138 2013-03-04 02:36:00 bitit has joined
 139 2013-03-04 02:38:51 coolsa has quit (Read error: Connection reset by peer)
 140 2013-03-04 02:38:57 Shealan has quit (Quit: Leaving.)
 141 2013-03-04 02:39:30 coolsa has joined
 142 2013-03-04 02:44:43 swappermall has joined
 143 2013-03-04 02:45:03 swappermall is now known as Guest18144
 144 2013-03-04 02:47:24 djoot has quit (Quit: leaving)
 145 2013-03-04 02:48:33 mappum has quit (Ping timeout: 248 seconds)
 146 2013-03-04 02:50:03 tcatm has quit (Quit: No Ping reply in 180 seconds.)
 147 2013-03-04 02:50:20 tcatm has joined
 148 2013-03-04 02:50:21 tcatm has quit (Changing host)
 149 2013-03-04 02:50:21 tcatm has joined
 150 2013-03-04 02:50:53 saivann has quit ()
 151 2013-03-04 02:51:55 CodeShark has quit (Remote host closed the connection)
 152 2013-03-04 02:52:51 BenderCoin has quit (Read error: Operation timed out)
 153 2013-03-04 02:53:11 BenderCoin has joined
 154 2013-03-04 02:55:24 Guest18144 has quit (Quit: ChatZilla 0.9.90 [Firefox 19.0/20130218103317])
 155 2013-03-04 02:58:39 mappum has joined
 156 2013-03-04 03:00:38 BTCTrader2 has quit (Quit: BTCTrader2)
 157 2013-03-04 03:01:34 <muhoo> which version of bitcoimj does multibit use? it looks like it hasn't been updated in over a year
 158 2013-03-04 03:03:05 Impaler has joined
 159 2013-03-04 03:07:05 djoot has joined
 160 2013-03-04 03:13:00 Goonie has quit (Ping timeout: 260 seconds)
 161 2013-03-04 03:14:47 Apexseals has quit (Remote host closed the connection)
 162 2013-03-04 03:15:10 Apexseals has joined
 163 2013-03-04 03:15:53 D34TH has quit (Quit: Leaving)
 164 2013-03-04 03:19:38 BTCTrader2 has joined
 165 2013-03-04 03:23:59 BTCTrader2 has quit (Ping timeout: 252 seconds)
 166 2013-03-04 03:24:17 Zarutian has quit (Quit: Zarutian)
 167 2013-03-04 03:25:16 abrkn has quit (Ping timeout: 240 seconds)
 168 2013-03-04 03:26:26 Impaler has quit (Remote host closed the connection)
 169 2013-03-04 03:30:35 sebicas has joined
 170 2013-03-04 03:32:26 fiesh has quit (Ping timeout: 246 seconds)
 171 2013-03-04 03:35:37 fiesh has joined
 172 2013-03-04 03:40:50 djoot has quit (Quit: leaving)
 173 2013-03-04 03:41:41 vigilyn2 has joined
 174 2013-03-04 03:41:42 vigilyn3 has quit (Read error: Connection reset by peer)
 175 2013-03-04 03:42:01 BTCTrader2 has joined
 176 2013-03-04 03:43:44 edwincheese has quit (Quit: edwincheese)
 177 2013-03-04 03:51:21 FredEE has joined
 178 2013-03-04 03:52:27 LargoG has quit (Remote host closed the connection)
 179 2013-03-04 03:54:23 copumpkin has quit (Ping timeout: 252 seconds)
 180 2013-03-04 03:55:25 copumpkin has joined
 181 2013-03-04 03:59:11 mappum has quit (Ping timeout: 255 seconds)
 182 2013-03-04 04:02:37 djoot has joined
 183 2013-03-04 04:09:11 jurov has quit (Ping timeout: 246 seconds)
 184 2013-03-04 04:10:02 abrkn has joined
 185 2013-03-04 04:11:20 jurov has joined
 186 2013-03-04 04:13:23 <muhoo> oic, they don't actually use master as master :-(
 187 2013-03-04 04:13:28 <muhoo> weird
 188 2013-03-04 04:14:57 <muhoo> looks like master is actually 0.5
 189 2013-03-04 04:15:05 <muhoo> and.. they've forked the hell out of bitcoinj. wow.
 190 2013-03-04 04:17:32 <muhoo> well, that gives me a bit of courage. maybe i will just replace wallet with my own code then, and be done with it.
 191 2013-03-04 04:17:38 copumpkin has quit (Ping timeout: 245 seconds)
 192 2013-03-04 04:18:19 copumpkin has joined
 193 2013-03-04 04:22:08 emryss has quit (Ping timeout: 255 seconds)
 194 2013-03-04 04:24:04 <muhoo> ooh, replayableblockstore. awesome, hmm, might be easier to base my stuff on multibit instead.
 195 2013-03-04 04:29:58 emryss has joined
 196 2013-03-04 04:32:04 [7] has quit (Disconnected by services)
 197 2013-03-04 04:32:13 TheSeven has joined
 198 2013-03-04 04:34:54 swulf-- has joined
 199 2013-03-04 04:35:11 <swulf--> Why would it be that "bitcoind getbalance specific-account" yields a higher balance result than "bitcoind getbalance" ? I have a situation where I have nearly double the bitcoins in specific-account than I do if i just do getbalance without a parameter
 200 2013-03-04 04:38:21 <Luke-Jr> swulf--: another account has a negative amount perhaps?
 201 2013-03-04 04:38:32 <swulf--> hmm
 202 2013-03-04 04:38:36 <swulf--> ok
 203 2013-03-04 04:38:44 * swulf-- scans the other hundreds of accounts
 204 2013-03-04 04:39:20 <swulf--> ah there we go
 205 2013-03-04 04:39:23 <clav8_> listaccounts
 206 2013-03-04 04:39:27 <clav8_> will show it
 207 2013-03-04 04:39:40 <swulf--> yup
 208 2013-03-04 04:39:46 <swulf--> account "" has negative value in it
 209 2013-03-04 04:39:49 <swulf--> weird
 210 2013-03-04 04:39:53 <swulf--> why is that?
 211 2013-03-04 04:40:06 <Luke-Jr> you probably used sendtoaddress or some other account-unaware method
 212 2013-03-04 04:40:14 <swulf--> ahhh
 213 2013-03-04 04:40:17 <swulf--> yes
 214 2013-03-04 04:40:35 HM has quit (Ping timeout: 252 seconds)
 215 2013-03-04 04:40:46 <swulf--> sendfrom would be better to keep track of account balances then?
 216 2013-03-04 04:41:09 freakazoid has joined
 217 2013-03-04 04:41:53 <swulf--> "move"ing a negative amount seems disallowed.  that's too bad
 218 2013-03-04 04:41:56 <Luke-Jr> swulf--: yes, or sendmany
 219 2013-03-04 04:42:05 <Luke-Jr> swulf--: move a positive amount ;)
 220 2013-03-04 04:42:28 <swulf--> good idea
 221 2013-03-04 04:43:13 coolsa has quit (Ping timeout: 248 seconds)
 222 2013-03-04 04:44:03 <swulf--> cool. thanks!
 223 2013-03-04 04:44:23 HM has joined
 224 2013-03-04 05:01:17 Mad7Scientist has quit (Ping timeout: 255 seconds)
 225 2013-03-04 05:01:51 Mad7Scientist has joined
 226 2013-03-04 05:05:23 Diablo-D3 has quit (Quit: This computer has gone to sleep)
 227 2013-03-04 05:06:35 _dr has quit (Ping timeout: 246 seconds)
 228 2013-03-04 05:09:04 mappum has joined
 229 2013-03-04 05:09:12 sega01 has quit (Ping timeout: 250 seconds)
 230 2013-03-04 05:12:45 _dr has joined
 231 2013-03-04 05:13:44 xjrn has joined
 232 2013-03-04 05:14:10 xjrn has quit (Client Quit)
 233 2013-03-04 05:29:54 puhc has joined
 234 2013-03-04 05:30:00 sebicas has quit (Quit: sebicas)
 235 2013-03-04 05:36:20 clav8_ has quit ()
 236 2013-03-04 05:41:24 swappermall_ has joined
 237 2013-03-04 05:43:08 [u]{s}[e]{r} has joined
 238 2013-03-04 05:48:47 HM has quit (Ping timeout: 252 seconds)
 239 2013-03-04 05:50:35 Muis has joined
 240 2013-03-04 05:51:44 Muis__ has quit (Ping timeout: 272 seconds)
 241 2013-03-04 05:52:27 HM has joined
 242 2013-03-04 06:03:58 ThomasV has joined
 243 2013-03-04 06:08:42 gritcoin has joined
 244 2013-03-04 06:10:48 CodeShark has joined
 245 2013-03-04 06:12:01 HM has quit ()
 246 2013-03-04 06:18:28 swappermall_ has quit (Ping timeout: 256 seconds)
 247 2013-03-04 06:21:16 FredEE has quit (Quit: FredEE)
 248 2013-03-04 06:23:09 grau has joined
 249 2013-03-04 06:30:40 FredEE has joined
 250 2013-03-04 06:32:10 FredEE has quit (Client Quit)
 251 2013-03-04 06:43:34 RazielZ has joined
 252 2013-03-04 06:44:45 owowo has quit (Quit: sayonara)
 253 2013-03-04 06:57:36 grau has quit (Remote host closed the connection)
 254 2013-03-04 06:58:21 Benjojo has quit (Ping timeout: 245 seconds)
 255 2013-03-04 06:59:01 terry has quit (Ping timeout: 264 seconds)
 256 2013-03-04 07:00:18 eoss has joined
 257 2013-03-04 07:00:50 splnkr has quit (Quit: splnkr)
 258 2013-03-04 07:11:51 <muhoo> huh, what, how is it possible i'd never heard of multibitmerchant until now.
 259 2013-03-04 07:12:28 <muhoo> looks like there is already a complete store backend, based on bitcoinj (though old, 0.5.0).
 260 2013-03-04 07:12:56 <weex> the ability to get the word out about ones product or service is not evenly distributed
 261 2013-03-04 07:13:20 veerboot has joined
 262 2013-03-04 07:13:57 <muhoo> clearly
 263 2013-03-04 07:14:04 <muhoo> weex: is that what you use for coindl?
 264 2013-03-04 07:14:10 <weex> nope
 265 2013-03-04 07:14:25 <weex> looking at bitcoinj though
 266 2013-03-04 07:14:42 <muhoo> did you modify the bitcoind then?
 267 2013-03-04 07:14:56 <weex> never really ran java for any web services but bitcoinj seemingly does so much it's tempting
 268 2013-03-04 07:15:03 <weex> no, no mods required
 269 2013-03-04 07:15:19 <weex> just careful connectivity
 270 2013-03-04 07:15:45 <muhoo> really? but there's no feature in bitcoind to watch transactions, AFAICT. that's how i ended up with bitcoinj.
 271 2013-03-04 07:16:14 <weex> what about listtransactions?
 272 2013-03-04 07:16:24 <muhoo> polling, sure
 273 2013-03-04 07:16:27 <muhoo> but not a callback
 274 2013-03-04 07:16:39 * muhoo hates polling
 275 2013-03-04 07:16:48 <weex> hate it all you want, workforme
 276 2013-03-04 07:16:54 <weex> worksforme*
 277 2013-03-04 07:17:25 <weex> was bitcoinj hard to setup and keep running?
 278 2013-03-04 07:17:27 Mandrius has joined
 279 2013-03-04 07:17:40 <muhoo> if you know java, not really.
 280 2013-03-04 07:17:42 <gmaxwell> muhoo: uh,   -walletnotify=<cmd>    Execute command when a wallet transaction changes (%s in cmd is replaced by TxID)
 281 2013-03-04 07:18:03 <gmaxwell> (though blocknotify is really sufficient for most things)
 282 2013-03-04 07:18:14 <muhoo> gmaxwell: wallet tx, not all tx. gah, that's why i missed it.
 283 2013-03-04 07:18:49 <muhoo> i actually only need wallet tx anyway, but i guess i'd not realized that at the time.
 284 2013-03-04 07:18:53 freakazoid has quit (Ping timeout: 245 seconds)
 285 2013-03-04 07:19:36 <gmaxwell> well, blocknotify lets you just go run listtransactions when there is a new block. So it's 'polling' based but only once per ten minutes.
 286 2013-03-04 07:19:57 <muhoo> well i have to say i feel like a complete idiot. i've spent months trying to build somethign from low-level building blocks that could have been hacked together easily with existing components.
 287 2013-03-04 07:20:03 buddyrandom has joined
 288 2013-03-04 07:20:31 puhc has quit (Quit: Leaving)
 289 2013-03-04 07:20:59 puhc has joined
 290 2013-03-04 07:21:05 <weex> i happened to have the bitcoinj security model doc open in my browser from last night
 291 2013-03-04 07:21:19 <muhoo> yes, it's SPV
 292 2013-03-04 07:21:45 <muhoo> tho there's a new feature to run it as full node. haven't tried that yet.
 293 2013-03-04 07:22:29 brwyatt is now known as brwyatt|Away
 294 2013-03-04 07:26:45 <muhoo> well, i'm just dumbfounded. stunned by how much time i've wasted. looks like there's nothing here for me after all. shit.
 295 2013-03-04 07:26:53 eoss has quit (Remote host closed the connection)
 296 2013-03-04 07:27:35 puhc has quit (Quit: Leaving)
 297 2013-03-04 07:28:15 puhc has joined
 298 2013-03-04 07:30:21 swulf-- has quit (Read error: Connection reset by peer)
 299 2013-03-04 07:31:03 swulf-- has joined
 300 2013-03-04 07:31:39 bitit has quit (Ping timeout: 276 seconds)
 301 2013-03-04 07:33:23 <weex> muhoo:
 302 2013-03-04 07:34:02 <weex> sorry, i was going to ask what you were trying to do, i can't imagine the time you spent was totally wasted
 303 2013-03-04 07:35:48 <muhoo> not totally wasted, but i learned a lot of stuff that it now appears there's not a lot of potential to turn into income :-)
 304 2013-03-04 07:36:38 <muhoo> anyway, going to sleep on this and decide in the morning whether to continue.
 305 2013-03-04 07:36:43 <muhoo> thanks all
 306 2013-03-04 07:38:51 grau has joined
 307 2013-03-04 07:45:57 grau has quit (Remote host closed the connection)
 308 2013-03-04 07:51:57 toffoo has quit ()
 309 2013-03-04 07:56:37 <sturles> gmaxwell: -walletnotify?  bitcoind 0.8.0 --help doesn't mention it.
 310 2013-03-04 07:59:18 <gmaxwell> It was merged after 0.8.0 was cut (the pull was there for a long time before). The blocknotify has been around for a long time.
 311 2013-03-04 08:01:54 kritCoin_ has joined
 312 2013-03-04 08:02:00 <kritCoin_> hello
 313 2013-03-04 08:02:26 gritcoin has quit (Quit: gritcoin)
 314 2013-03-04 08:02:41 <kritCoin_> !ticker
 315 2013-03-04 08:02:42 <gribble> BTCUSD ticker | Best bid: 34.57769, Best ask: 34.78000, Bid-ask spread: 0.20231, Last trade: 34.78000, 24 hour volume: 17357.95129435, 24 hour low: 33.80000, 24 hour high: 34.80000, 24 hour vwap: 34.30602
 316 2013-03-04 08:05:35 random_cat has joined
 317 2013-03-04 08:05:47 <kritCoin_> !ticker --last
 318 2013-03-04 08:05:47 <gribble> 34.85000
 319 2013-03-04 08:07:38 epscy has quit (Ping timeout: 245 seconds)
 320 2013-03-04 08:09:00 epscy has joined
 321 2013-03-04 08:09:28 <sturles> gmaxwell: Thanks!  Will test.
 322 2013-03-04 08:11:06 CaptainBlaze has joined
 323 2013-03-04 08:11:50 Adifex has joined
 324 2013-03-04 08:13:15 grau has joined
 325 2013-03-04 08:13:43 Adifex has quit (Client Quit)
 326 2013-03-04 08:21:23 kritCoin_ has quit (Quit: Page closed)
 327 2013-03-04 08:34:43 swulf-- has quit (Ping timeout: 245 seconds)
 328 2013-03-04 08:35:01 swulf-- has joined
 329 2013-03-04 08:36:09 Guest64161 is now known as Hasimir
 330 2013-03-04 08:36:10 Hasimir has quit (Changing host)
 331 2013-03-04 08:36:10 Hasimir has joined
 332 2013-03-04 08:42:48 ThomasV has quit (Quit: Quitte)
 333 2013-03-04 08:44:08 Adifex has joined
 334 2013-03-04 08:47:21 ovidiusoft has joined
 335 2013-03-04 08:48:22 Goonie has joined
 336 2013-03-04 08:49:20 nowan_ has quit (Ping timeout: 250 seconds)
 337 2013-03-04 08:51:18 Adifex has quit (Quit: Adifex)
 338 2013-03-04 08:51:27 nowan has joined
 339 2013-03-04 08:51:34 Adifex has joined
 340 2013-03-04 08:58:01 Prattler has joined
 341 2013-03-04 08:58:58 terry has joined
 342 2013-03-04 09:02:48 [u]{s}[e]{r} has quit ()
 343 2013-03-04 09:16:51 Benjojo has joined
 344 2013-03-04 09:20:13 TD has joined
 345 2013-03-04 09:27:19 copumpkin has quit (Ping timeout: 252 seconds)
 346 2013-03-04 09:28:03 copumpkin has joined
 347 2013-03-04 09:28:37 n0g4n0o has quit (Quit: Blessed are the peacemakers, For they shall be called sons of God. Blessed are those who are persecuted for righteousness' sake, For theirs is the kingdom of heaven. Blessed are you when they revile and persecute you, and say all kinds of evil against yo)
 348 2013-03-04 09:32:09 AtashiCon has quit (Quit: AtashiCon)
 349 2013-03-04 09:35:52 swulf-- has left ()
 350 2013-03-04 09:36:06 mappum has quit (Ping timeout: 264 seconds)
 351 2013-03-04 09:39:35 kritCoin_ has joined
 352 2013-03-04 09:40:43 Shealan has joined
 353 2013-03-04 09:55:03 AtashiCon has joined
 354 2013-03-04 10:01:58 TD has quit (Quit: TD)
 355 2013-03-04 10:02:29 Acciaio has joined
 356 2013-03-04 10:03:34 <Acciaio> Hi all, after how many time an unconfirmed transaction will be deleted?
 357 2013-03-04 10:03:43 <Acciaio> how can I recover that bitcoins?
 358 2013-03-04 10:05:39 <Adifex> All bitcoin transactions are irreversible.
 359 2013-03-04 10:08:15 <gmaxwell> Acciaio: to recover coins stuck by unconfirmed transactions you must rebuild the wallet. Make a backup and use the salvagewallet option to bitcoin
 360 2013-03-04 10:08:24 <gmaxwell> Acciaio: how did you end up with stuck unconfirmed transactions?
 361 2013-03-04 10:08:25 <Acciaio> adifex but the transaction is unconfirmed and it is no more present https://blockchain.info/it/tx/ee082cd20d6bd35380d1311f84ca4ff967d72821aa12ca3f354718c3a5208e66
 362 2013-03-04 10:09:14 <Acciaio> thanks gmaxwell I'm trying this!
 363 2013-03-04 10:09:33 <Acciaio> the transaction was similar to https://blockchain.info/it/tx/2ce912fd4ce2438ff899c1d55cbc2c2653a86cf8ac1be08a3d38d153618a7cfe
 364 2013-03-04 10:10:38 <weex> it does have a fee
 365 2013-03-04 10:12:52 <Acciaio> yes 0.0005 fee
 366 2013-03-04 10:13:15 <Acciaio> I'm always sending 0.0005 fee
 367 2013-03-04 10:13:37 <gmaxwell> Acciaio: how long have you waited?
 368 2013-03-04 10:14:07 TD has joined
 369 2013-03-04 10:14:21 <Acciaio> since 28 feb but I don't know exactly when it was dropped
 370 2013-03-04 10:15:45 <gmaxwell> Acciaio: did you build the transaction using the raw transaction api? the reference client shouldn't build a transaction which won't get mined.
 371 2013-03-04 10:16:32 <Acciaio> gmaxwell, yes it was build using rawtransaction
 372 2013-03-04 10:17:07 <Acciaio> and never got a reject(from client) when I send some fee
 373 2013-03-04 10:20:02 <gmaxwell> sendrawtransaction will only reject transactions that your node won't even relay. It does not reject low fee transactions which is wouldn't mine.
 374 2013-03-04 10:21:05 <gmaxwell> If  your transaction does not qualify as free and don't have at least 0.0001 BTC /KB of fee it is pretty unlikely to get mined.
 375 2013-03-04 10:22:25 <gmaxwell> Acciaio: absolute fee value isn't imporant, what miners care about is btc per kb.
 376 2013-03-04 10:23:22 <Graet> http://pastebin.com/82u4nNce was from reference client version 0.6.3, does salvage wallet work on older versions or will i need to update?
 377 2013-03-04 10:24:31 <kritCoin_> gmaxwell, you are employed by a bitcoin company?
 378 2013-03-04 10:24:49 <sipa> Graet: since 0.7.0 or .1 iirc
 379 2013-03-04 10:24:56 <Graet> ty sipa
 380 2013-03-04 10:25:22 Sydna has joined
 381 2013-03-04 10:25:41 Sydna has left ()
 382 2013-03-04 10:25:46 <Acciaio> ty gmaxwell  I'm changing my sender to meet 0.0001/KB
 383 2013-03-04 10:27:35 ielo has quit (Ping timeout: 260 seconds)
 384 2013-03-04 10:27:40 <gmaxwell> kritCoin_: no.
 385 2013-03-04 10:28:29 one_zero has quit ()
 386 2013-03-04 10:33:06 <TD> gmaxwell do you ever sleep? :)
 387 2013-03-04 10:33:35 Raccoon has quit (Quit: Ì%Ɯ-¯ãÙH¬Pgg'hå×ãT”bë..Zž`Xjµà±]+~#B¹U¢(®;³ª~ZµQO!AMá`ûϪ#ù7¬€)
 388 2013-03-04 10:33:44 <gmaxwell> Just finished doing my taxes ... and time to go to sleep, in fact. :P
 389 2013-03-04 10:34:14 <TD> sleep and taxes
 390 2013-03-04 10:34:16 <TD> the two inevitabilities
 391 2013-03-04 10:34:20 <sipa> it's only 2:30 am for him, give hin a rest
 392 2013-03-04 10:34:24 <TD> enjoy :)
 393 2013-03-04 10:34:34 <TD> i suppose so
 394 2013-03-04 10:35:11 <Graet> no rest for ppl with nicks starting with g!!!
 395 2013-03-04 10:35:18 <Graet> i have found :P
 396 2013-03-04 10:35:35 <sipa> thank goodness
 397 2013-03-04 10:35:51 <Graet> good choice sipa "D
 398 2013-03-04 10:37:30 <petertodd> sipa: Oh, and on top of that, I think transaction selection in CreateNewBlock has been wrong all along re: IsFinal()
 399 2013-03-04 10:41:33 <sipa> petertodd: :o
 400 2013-03-04 10:43:27 <petertodd> sipa: Basically, CreateNewBlock() calls IsFinal() directly, which compares against nBlockHeight, while AcceptBlock() compares against the height of the block itself, IE, there is an off-by-one.
 401 2013-03-04 10:43:42 <petertodd> sipa: Really bad for fidelity bonds... the sacrifices are meaningless until this is fixed.
 402 2013-03-04 10:46:38 <sipa> petertodd: just checked; you're right
 403 2013-03-04 10:46:49 <petertodd> sipa: Yeah, checking it myself on testnet to be sure.
 404 2013-03-04 10:48:14 BTCOxygen has joined
 405 2013-03-04 10:49:16 BTCOxygen has quit (Ping timeout: 256 seconds)
 406 2013-03-04 10:49:22 <sipa> petertodd: that's a pity... it means it will take months before a nLockTime in wallets can be added
 407 2013-03-04 10:49:23 BTCOxygen is now known as 1!~BTCOxygen@unaffiliated/mroxy/bot/btcoxygen|BTCOxygen
 408 2013-03-04 10:49:42 <sipa> except a weaker off-by-one version
 409 2013-03-04 10:49:59 Adifex has quit (Quit: Adifex)
 410 2013-03-04 10:50:15 <petertodd> Yup, I'm thinking maybe do the off-by-one to shake out bugs, and then fix that eventually - fee sniping isn't going to be an issue for awhile.
 411 2013-03-04 10:50:29 <petertodd> I'm really glad I foudn that for fidelity bonds though, some savvy miner could have gotten every one for himself.
 412 2013-03-04 10:51:01 ProfMac has quit (Quit: Page closed)
 413 2013-03-04 10:54:53 <kritCoin_> http://www.wilmott.com/jobsboard/career/26048/Research-Scientist-Financial-Markets-Ny-Or-Ca-Basic-Salary-275k-Bonus-Region-Ny-Ca
 414 2013-03-04 10:55:02 <kritCoin_> shocked
 415 2013-03-04 10:55:12 <kritCoin_> jobs in finance are crazy
 416 2013-03-04 10:55:24 <petertodd> damn, that's crazy, that code is satoshi's, last changed Nov 23 2010
 417 2013-03-04 10:56:59 <petertodd> Kinda bizzare too, normally you'd fix it by changing what IsFinal() means, but, this is a potential hard-fork bug if it gets screwed up so...
 418 2013-03-04 10:57:18 bitit has joined
 419 2013-03-04 10:57:56 sgornick has quit (Ping timeout: 256 seconds)
 420 2013-03-04 10:58:05 <sipa> petertodd: but CreateNewBlock is wrong too, no?
 421 2013-03-04 10:58:44 <petertodd> Yes, but that's the thing, only AcceptBlock (and by extension IsFinal()) is the hard-fork risk, so you want to fix it everywhere but there.
 422 2013-03-04 10:58:48 <sipa> it also just calls IsFinal, while it should call IsFinal(nBestHeight+1) (ie the height of the block being created)
 423 2013-03-04 10:58:54 <petertodd> For instance the GUI is broken
 424 2013-03-04 10:59:05 <petertodd> Yup, that's what I did to fix it.
 425 2013-03-04 10:59:18 <petertodd> I'm waiting for a block to be mined on testnet to be sure that really does fix it
 426 2013-03-04 10:59:39 <sipa> i only see a pullreq for changing IsStandard
 427 2013-03-04 10:59:42 <sipa> oh, ok
 428 2013-03-04 11:00:22 <bitnumus> hi again, i have had Luceo bail on a project, is there anybody here that may be interested in about 30-40hours of work, that they can squeeze in 2-3weeks ?
 429 2013-03-04 11:00:41 <petertodd> ...yup, I think that works. getblockcount is 55599, and tx a984595e4fff9d98d45d5a53bee1cd80a749e3260cf80f49dd97ca267b2dd9be has nLockTime 55598
 430 2013-03-04 11:01:11 <petertodd> sipa: Anyway, I'll update the pull-req, but this needs more testing.
 431 2013-03-04 11:03:28 <sipa> petertodd: yes the hars rule is "nLockTime specifies the block up to and including which the tx cannot be mined", and that rule can't change
 432 2013-03-04 11:03:51 <sipa> so IsStandard and CreateNewBlick should be adapted to fit that rule
 433 2013-03-04 11:03:59 <sipa> which is what you propose, afaics
 434 2013-03-04 11:04:32 <petertodd> yup
 435 2013-03-04 11:13:57 kicek has joined
 436 2013-03-04 11:14:26 knotwork has quit (Ping timeout: 246 seconds)
 437 2013-03-04 11:19:00 Diablo-D3 has joined
 438 2013-03-04 11:20:12 cdecker_ has quit (Remote host closed the connection)
 439 2013-03-04 11:28:23 knotwork has joined
 440 2013-03-04 11:28:30 rbecker is now known as RBecker
 441 2013-03-04 11:31:06 Diablo-D3 has quit (Quit: do coders dream of sheep()?)
 442 2013-03-04 11:31:16 <Acciaio> 1kb=1024b or 1000b?
 443 2013-03-04 11:31:25 Diablo-D3 has joined
 444 2013-03-04 11:33:34 <Quazgaa> technically speaking 1kb == 1000b and 1kib == 1024b
 445 2013-03-04 11:33:47 <Acciaio> fee calculating speaking?
 446 2013-03-04 11:34:03 <Acciaio> ah ok
 447 2013-03-04 11:34:05 <Acciaio> so 1000
 448 2013-03-04 11:34:12 <Acciaio> thanks
 449 2013-03-04 11:34:24 <kritCoin_> how do you pronounce kib and kb?
 450 2013-03-04 11:34:29 <kritCoin_> the difference?
 451 2013-03-04 11:37:02 <_dr> kilobyte kibibyte
 452 2013-03-04 11:37:23 <_dr> and swap the order ;)
 453 2013-03-04 11:37:54 <_dr> see http://xkcd.com/394/ for any further questions
 454 2013-03-04 11:39:05 RBecker is now known as rbecker
 455 2013-03-04 11:39:32 Scrat_l has joined
 456 2013-03-04 11:39:37 Scrat is now known as Guest90966
 457 2013-03-04 11:40:00 Scrat_l is now known as Scrat
 458 2013-03-04 11:42:26 Guest90966 has quit (Ping timeout: 252 seconds)
 459 2013-03-04 11:43:04 bitit has quit (Remote host closed the connection)
 460 2013-03-04 11:47:29 t7 has joined
 461 2013-03-04 11:48:42 BTCTrader2 has quit (Quit: BTCTrader2)
 462 2013-03-04 11:57:28 Grishnakh_ has quit (Read error: Connection reset by peer)
 463 2013-03-04 11:58:00 Grishnakh_ has joined
 464 2013-03-04 11:59:28 Quazgaa has quit (Read error: Operation timed out)
 465 2013-03-04 12:05:39 HM has joined
 466 2013-03-04 12:11:36 drizztbsd has joined
 467 2013-03-04 12:12:25 <Graet> --salvagewallet worked a treat, cheers gmaxwell , sipa :D
 468 2013-03-04 12:12:45 Quazgaa has joined
 469 2013-03-04 12:13:35 asoltys has quit (Ping timeout: 246 seconds)
 470 2013-03-04 12:13:58 GMP has quit (Read error: Connection reset by peer)
 471 2013-03-04 12:18:43 dub has quit (Ping timeout: 272 seconds)
 472 2013-03-04 12:19:00 CaptainBlaze has quit (Quit: CaptainBlaze)
 473 2013-03-04 12:20:40 asoltys has joined
 474 2013-03-04 12:24:33 dub has joined
 475 2013-03-04 12:28:21 drizztbsd has quit (Quit: Konversation terminated!)
 476 2013-03-04 12:28:22 <kritCoin_> Could part of the miner code be implemented on a shark DSP ?
 477 2013-03-04 12:28:49 <kritCoin_> they are very low cost , low energy and used by the military for shell trajectory calculations
 478 2013-03-04 12:29:07 <kritCoin_> anyone?
 479 2013-03-04 12:29:36 <kritCoin_> http://www.analog.com/en/processors-dsp/sharc/products/index.html
 480 2013-03-04 12:29:43 <kritCoin_> The SHARC® Processor family dominates the floating-point DSP market with exceptional core and memory performance and outstanding I/O throughput. For as little as 319 MFLOPS/dollar, SHARC brings floating-point processing performance to applications where dynamic range is key.
 481 2013-03-04 12:30:00 <kritCoin_> usable?
 482 2013-03-04 12:30:24 occulta has joined
 483 2013-03-04 12:30:51 <sipa> FLOPs are useless, we need integer operations
 484 2013-03-04 12:31:33 <kritCoin_> so trunc it, jezus
 485 2013-03-04 12:31:37 <kritCoin_> duh
 486 2013-03-04 12:31:42 <sipa> ?
 487 2013-03-04 12:32:33 <sipa> also, 319 MFLOPS/dollar is significantly more expensive than GPUs
 488 2013-03-04 12:32:43 <kritCoin_> really?
 489 2013-03-04 12:32:59 <kritCoin_> I thought GPU's didnt do MFLOPS
 490 2013-03-04 12:33:00 <sipa> high-end GPUs do close to (or already over?) TFLOPS
 491 2013-03-04 12:33:28 <kritCoin_> "you need integer operations" how how are GPU's usefull then?
 492 2013-03-04 12:33:37 <sipa> just as a comparison
 493 2013-03-04 12:33:44 <sipa> they are not used at all for bitcoin
 494 2013-03-04 12:33:50 <kritCoin_> GPU's?
 495 2013-03-04 12:33:56 <kritCoin_> not used for bitcoin?
 496 2013-03-04 12:33:57 <kritCoin_> ok
 497 2013-03-04 12:34:01 <sipa> no, floating point operations
 498 2013-03-04 12:34:16 <kritCoin_> Sharcs do a lit bit more then FP
 499 2013-03-04 12:34:22 <sipa> *sigh*
 500 2013-03-04 12:34:24 <sipa> so do GPUs
 501 2013-03-04 12:34:37 <kritCoin_> so why are you making this an issue then
 502 2013-03-04 12:34:51 <sipa> because FLOPS are typically the only performance number listed
 503 2013-03-04 12:35:03 <kritCoin_> so it isnt an issue then
 504 2013-03-04 12:35:13 <sipa> so just as a comparison, assuming the speed ratio between integer and floating point ops is similar
 505 2013-03-04 12:35:20 <kritCoin_> (usability for mining)
 506 2013-03-04 12:35:37 <kritCoin_> thats was not your initial objection, but never mind
 507 2013-03-04 12:35:50 Zarutian has joined
 508 2013-03-04 12:35:53 <kritCoin_> why assume?
 509 2013-03-04 12:35:57 <kritCoin_> you dont know this at all
 510 2013-03-04 12:36:05 <kritCoin_> the SHARC is pretty cheap
 511 2013-03-04 12:36:27 <sipa> so, assuming 1) the chips you list do the necessary operations 2) they have performance proportional to the FLOPS they indicate, they are far more expensive than GPUs for the same mining power
 512 2013-03-04 12:36:36 <kritCoin_> and very little energy consumption, maybe slower, so what, plug in more sharcs
 513 2013-03-04 12:36:41 <sipa> assuming they'd only do floating point ops, they are useless at all
 514 2013-03-04 12:36:43 <kritCoin_> Sorry I dont assume
 515 2013-03-04 12:36:50 <kritCoin_> why would i do that?
 516 2013-03-04 12:36:52 <sipa> then look it up and prove me wrong, please!
 517 2013-03-04 12:37:03 <kritCoin_> excuse me?
 518 2013-03-04 12:37:17 <kritCoin_> I cant go around disproving people who assume stuff in the moment
 519 2013-03-04 12:37:25 <sipa> ok
 520 2013-03-04 12:37:47 <sipa> then don't :)
 521 2013-03-04 12:38:11 <CodeShark> lol
 522 2013-03-04 12:40:14 <kritCoin_> funny how loosing an argument results in stalking behavior from the counterparty
 523 2013-03-04 12:40:23 <kritCoin_> sad really
 524 2013-03-04 12:40:42 <_dr> kritCoin_: at least sipa has an idea what he's talking about
 525 2013-03-04 12:41:01 <kritCoin_> how would you know, he just "assumes" , right?
 526 2013-03-04 12:41:15 <_dr> since for GPUs there no 1:1 relation of flops ~ integer performance as on CPUs
 527 2013-03-04 12:41:22 <kritCoin_> and you just said its kindof ok to stalk after running out of arguments, ... not good
 528 2013-03-04 12:42:11 <sipa> wait, stalk?
 529 2013-03-04 12:42:41 <kritCoin_> Common Architectural Features      32/40-Bit IEEE Floating-Point Math     32-Bit Fixed-Point Multipliers with 64-Bit Product & 80-Bit Accumulation
 530 2013-03-04 12:42:52 <_dr> who cares about FP?
 531 2013-03-04 12:42:53 <kritCoin_> it does a lil more then FPM
 532 2013-03-04 12:42:56 <_dr> as sipa said, we need int
 533 2013-03-04 12:43:00 <sipa> i just consider it extremely unlikely that such a chip would be useful for mining, and i gave my reasoning why - yes it makes a lot of assumptions, but i think these are very reasonable ones
 534 2013-03-04 12:43:16 <kritCoin_> http://www.analog.com/en/processors-dsp/sharc/products/index.html
 535 2013-03-04 12:43:17 <_dr> if you want FP performance buy a nvidia k20 or an intel mic, they both do 1TFLOP DP
 536 2013-03-04 12:43:32 <kritCoin_> _dr, please follow the thread
 537 2013-03-04 12:43:49 <sipa> you don't have to believe me, and if i'm wrong, and you build a succesfull hardware mining production business out of it, i'll be very glad to be wrong
 538 2013-03-04 12:44:04 <kritCoin_> ok
 539 2013-03-04 12:47:05 LargoG has joined
 540 2013-03-04 12:47:05 <_dr> fyi, a dsp does dsp stuff, it's more or less some sort of asic for signal processing. i'd be surprised if it has a vector integer pipeline
 541 2013-03-04 12:48:32 <CodeShark> DSPs are optimized for I/O throughput and reproducible performance
 542 2013-03-04 12:49:17 <kritCoin_> its pretty generic
 543 2013-03-04 12:49:27 <kritCoin_> read the datasheet on sharc
 544 2013-03-04 12:50:00 <_dr> i don't want to offend you
 545 2013-03-04 12:50:05 <kritCoin_> Once wrap my head around the code a bit I might order a sharc starters kit
 546 2013-03-04 12:50:17 <kritCoin_> ok _dr tell me
 547 2013-03-04 12:50:20 <_dr> but it's like assuming that maybe your pocket calculator has a sha2 pipeline, maybe we should check it out!
 548 2013-03-04 12:50:47 <kritCoin_> hahaha
 549 2013-03-04 12:50:54 <kritCoin_> ok
 550 2013-03-04 12:50:58 <_dr> almost no off the shelf hardware has large int vector pipelines, which is what you need for fast mining
 551 2013-03-04 12:51:02 <_dr> apart for crypto asics
 552 2013-03-04 12:51:31 <kritCoin_> I am reading the stuff in SHARC  , gimme a moment please
 553 2013-03-04 12:51:35 * CodeShark imagines programming sha2 using an old HP calculator
 554 2013-03-04 12:51:39 <_dr> look at nvidia, they boast a stunning 3 TFLOPS of single precision floating point operations per second
 555 2013-03-04 12:51:48 <kritCoin_> crap, they flew to the moon on that
 556 2013-03-04 12:51:58 <_dr> and for mining, they suck, because they have almost no integer piplelines, and why would they have it on a gpu?
 557 2013-03-04 12:54:25 <kritCoin_> Oh man, this baby as some serious commands
 558 2013-03-04 12:54:39 <kritCoin_> ok, it flattens out on 300-400 MFLOPS
 559 2013-03-04 12:54:59 <_dr> heh :)
 560 2013-03-04 12:56:06 <_dr> that makes my macbook air about 220x faster than your dsp
 561 2013-03-04 12:56:11 <_dr> good luck mining on that :)
 562 2013-03-04 12:57:45 <HM> _dr: it's not all lost
 563 2013-03-04 12:57:53 <_dr> (that's only the macbook's CPU btc)
 564 2013-03-04 12:57:57 <_dr> btw
 565 2013-03-04 12:57:59 <HM> Dan Bernstein implemented some EC stuff utilising float ops
 566 2013-03-04 12:58:06 <HM> i saw the code myself
 567 2013-03-04 12:58:10 <HM> clever hacks
 568 2013-03-04 12:58:18 <_dr> HM: too bad we need sha256, not ec
 569 2013-03-04 12:58:37 <_dr> and hashing is based on two things: replacement and permutation
 570 2013-03-04 12:58:42 <HM> good point
 571 2013-03-04 12:58:47 <_dr> none of which you can do effectively with floats
 572 2013-03-04 12:59:09 <kritCoin_> it has an integer pipeline 32 bits
 573 2013-03-04 12:59:36 <_dr> HM: i think it's a good idea someone is doing research on that, but until i see some good results i choose to believe that it's not possible to do such a thing fast
 574 2013-03-04 12:59:41 <kritCoin_> the sharc is kind of a generic cpu aswell more of a hybrid
 575 2013-03-04 13:00:01 <CodeShark> should we start building shrines to the sharc and praying before it?
 576 2013-03-04 13:00:40 <kritCoin_> is just assesment, but please not base it on assumptions as I was just read ing the datasheets
 577 2013-03-04 13:00:45 <kritCoin_> I actually just read it
 578 2013-03-04 13:01:18 <kritCoin_> the energy consumption is stupidly low,
 579 2013-03-04 13:02:04 <_dr> if it's not lower than 0.01W my ivy bridge cpu will beat it kritCoin_ :)
 580 2013-03-04 13:02:08 <kritCoin_> ok...., i need to get farmiliar with the code anyway and order a simulation kit, see what happens
 581 2013-03-04 13:02:39 <kritCoin_> ok _dr,..
 582 2013-03-04 13:03:02 <kritCoin_> what about the vector pipeline of INtel that was supposed to live on the northbridge?
 583 2013-03-04 13:03:11 <kritCoin_> is that out yet?
 584 2013-03-04 13:03:17 <_dr> kritCoin_: wtf are you talking about
 585 2013-03-04 13:03:23 <_dr> the northbridge is the interface to memory
 586 2013-03-04 13:03:26 <kritCoin_> lemmi give you a link
 587 2013-03-04 13:03:36 bitit has joined
 588 2013-03-04 13:03:36 <_dr> which has been removed, and lives on-die since nehalem
 589 2013-03-04 13:04:08 <_dr> instruction pipelines live inside the cpu core d'oh
 590 2013-03-04 13:04:10 <kritCoin_> yeah, the place it there, i think some kind of DMA stuff, wait, i get th elink
 591 2013-03-04 13:04:20 <_dr> no, please don't :)
 592 2013-03-04 13:04:22 <CodeShark> I sense troll :)
 593 2013-03-04 13:04:39 <CodeShark> my troll alarm keeps going off
 594 2013-03-04 13:04:55 <_dr> i feel like the time Diablo-D3 told me AVX is nothing more than two fused SSE registers from two cores, to net performance stays the same :)
 595 2013-03-04 13:05:19 <kritCoin_> http://software.intel.com/sites/default/files/319433-014.pdf
 596 2013-03-04 13:06:01 <HM> a page number might be useful
 597 2013-03-04 13:06:09 <CodeShark> nah, just read the whole thing
 598 2013-03-04 13:06:11 <kritCoin_> No that was my nerd-alert filter
 599 2013-03-04 13:06:17 <CodeShark> we'll give you a couple minutes, HM
 600 2013-03-04 13:07:20 <kritCoin_> your troll detector must be doing overtime after you got the future designs from intel
 601 2013-03-04 13:07:31 <kritCoin_> "planned enhancements
 602 2013-03-04 13:07:33 <HM> CodeShark: does that include the time required for ramen?
 603 2013-03-04 13:08:09 <CodeShark> your CPU has ramen pipelines?
 604 2013-03-04 13:08:33 <HM> i wish
 605 2013-03-04 13:08:46 twobitcoins has quit (Read error: Connection reset by peer)
 606 2013-03-04 13:09:10 twobitcoins has joined
 607 2013-03-04 13:09:25 <kritCoin_> ump to: navigation, search  Advanced Vector Extensions (AVX) is an extension to the x86 instruction set architecture for microprocessors from Intel and AMD proposed by Intel in March 2008 and first supported by Intel with the Sandy Bridge processor shipping in Q1 2011 and later on by AMD with the Bulldozer processor shipping in Q3 2011.  AVX provides new features, new instructions and a new coding scheme.
 608 2013-03-04 13:09:35 <kritCoin_> http://en.wikipedia.org/wiki/Advanced_Vector_Extensions
 609 2013-03-04 13:10:00 <kritCoin_> How old is your macbook  dr_?
 610 2013-03-04 13:11:06 <HM> man x86 has a lot of baggage. almost 3 years from announce to shelves
 611 2013-03-04 13:11:24 <kritCoin_> my point,
 612 2013-03-04 13:11:34 <Graet> and bulldozer hashes really badly too
 613 2013-03-04 13:11:39 nus has quit (Read error: Connection reset by peer)
 614 2013-03-04 13:11:57 nus has joined
 615 2013-03-04 13:12:23 <kritCoin_> If  you have a new mac you have something that was designedin 2005/2004 maybe
 616 2013-03-04 13:12:50 abrkn has quit (Ping timeout: 256 seconds)
 617 2013-03-04 13:13:00 drizztbsd has joined
 618 2013-03-04 13:13:00 drizztbsd has quit (Changing host)
 619 2013-03-04 13:13:00 drizztbsd has joined
 620 2013-03-04 13:14:00 <kritCoin_> http://en.wikipedia.org/wiki/Haswell_%28microarchitecture%29  (planned 2012, lol)
 621 2013-03-04 13:14:32 <kritCoin_> http://en.wikipedia.org/wiki/Broadwell_%28microarchitecture%29 (2014, maybe)
 622 2013-03-04 13:15:13 <kritCoin_> ( I like http://software.intel.com/en-us/blogs/2011/06/13/haswell-new-instruction-descriptions-now-available)
 623 2013-03-04 13:16:31 <kritCoin_> This is downright revolutionary. Finally the SIMD instruction set will feature a vector equivalent of every (relevant) scalar instruction. This allows parallelizing lots of code loops, in theory offering up to an eightfold increase in throughput computing performance. GPGPU is history, as Haswell will prove that mainstream CPUs can combine the power of GPUs into a superior homogeneous architecture. Top
 624 2013-03-04 13:16:49 phungus has quit (Ping timeout: 276 seconds)
 625 2013-03-04 13:16:49 ageis has quit (Ping timeout: 245 seconds)
 626 2013-03-04 13:16:56 EPiSKiNG- has quit (Ping timeout: 246 seconds)
 627 2013-03-04 13:17:08 datagutt has joined
 628 2013-03-04 13:18:59 <Diablo-D3> [08:04:54] <_dr> i feel like the time Diablo-D3 told me AVX is nothing more than two fused SSE registers from two cores, to net performance stays the same :)
 629 2013-03-04 13:19:04 <Diablo-D3> I said Intel's implementation is
 630 2013-03-04 13:19:21 <Diablo-D3> however you don't notice it until you try to execute two avx instructions at the same time, one on one core, one on the other
 631 2013-03-04 13:19:47 <Diablo-D3> _dr: what SIMD does on CISC arcs is explicitly state various things
 632 2013-03-04 13:20:05 <Diablo-D3> the decoder/scheduler part of the pipeline needs to do much less work and can more optimally queue micro ops for execution
 633 2013-03-04 13:20:08 <Diablo-D3> avx IS a good thing
 634 2013-03-04 13:20:22 <Diablo-D3> so if you have avx available use it
 635 2013-03-04 13:20:54 <Diablo-D3> but its not as big of an optimization boost as doing sun/ibm type hardware threading
 636 2013-03-04 13:21:05 CodeShark has quit (Remote host closed the connection)
 637 2013-03-04 13:23:02 <Diablo-D3> where you have a decoder/scheduler handling multiple threads concurrently to intelligently work around various issues such as blocking, cache/system ram fetching, badly optimized code, and the fact CPUs have integer AND fp alus (ergo on a single thread single core design, if you have a thread that is all integer, that core's fp alus go idle; on a hardware thread design, if you have one integer thread and one fp thread,
 638 2013-03-04 13:23:02 <Diablo-D3>  both ales get used as if you had two cores all along)
 639 2013-03-04 13:23:40 <Diablo-D3> intel seems to also be using sun/ibm hw threads now as well in the new i7s that can do hyperthreading
 640 2013-03-04 13:24:11 <HM> hyperthreading has been around since forever
 641 2013-03-04 13:24:17 <HM> the pentium 4s had hyperthreading
 642 2013-03-04 13:24:22 <Diablo-D3> (intel hw threads isn't the same thing, thats used in p4 hyper threading and it parses a fixed number of ops per go, and basically it does nothing useful because it doesn't intelligently solve around blocking and fetch latency
 643 2013-03-04 13:24:24 <Diablo-D3> )
 644 2013-03-04 13:24:41 <Diablo-D3> (so yes, intel isn't using intel hw threads anymore. lawlz)
 645 2013-03-04 13:25:18 <Diablo-D3> the only reason modern x86 cpus are so goddamned fast is because of the speculative execution engines
 646 2013-03-04 13:25:44 <Diablo-D3> p4's was fucking retarded and required massive setup time on a speculation fail
 647 2013-03-04 13:25:53 agricocb has quit (Quit: Leaving.)
 648 2013-03-04 13:26:18 <Diablo-D3> on sun/ibm hw threads, among other things, you have multiple speculation states
 649 2013-03-04 13:26:18 <HM> i wouldn't say it was retarded
 650 2013-03-04 13:26:23 <Diablo-D3> HM: it was retarded.
 651 2013-03-04 13:26:36 <Diablo-D3> compared to the fact IBM already solved this particular problem _in the 70s_, Intel has no excuse
 652 2013-03-04 13:26:38 <HM> having an expensive pipeline incentivises good speculation
 653 2013-03-04 13:26:45 <HM> which is useful to have on newer cores as well
 654 2013-03-04 13:26:51 <HM> it's not like the technology is wasted
 655 2013-03-04 13:27:03 <Diablo-D3> yes, but you can't control the speculation programmably in P4s
 656 2013-03-04 13:27:10 <Diablo-D3> so Intel screwed that pooch
 657 2013-03-04 13:27:28 <kritCoin_> Diablo-D3, I think you can give "hints" in intel , no?
 658 2013-03-04 13:27:44 <HM> there are memory prefetch instructions
 659 2013-03-04 13:27:46 <Diablo-D3> kritCoin_: you can, and gcc even supports those now
 660 2013-03-04 13:27:53 <Diablo-D3> HM: no, you can hint branch paths too
 661 2013-03-04 13:28:06 <HM> i think that just effects code ordering
 662 2013-03-04 13:28:07 <kritCoin_> about goddamn time
 663 2013-03-04 13:28:12 <HM> e.g. compiler decisions
 664 2013-03-04 13:28:19 <Diablo-D3> yeah, its mostly compile time
 665 2013-03-04 13:28:32 <Diablo-D3> but loading more hot code into caches increases execution performance
 666 2013-03-04 13:28:41 <Diablo-D3> and also increases the performance of a highly speculative cpu
 667 2013-03-04 13:28:42 <HM> there was a recent linux kernel mailing list thread that basically pointed out that all their LIKELY/UNLIKELY macros made no difference
 668 2013-03-04 13:29:05 <Diablo-D3> HM: it depends entirely on the cpu and the program
 669 2013-03-04 13:29:09 <HM> it's more that there may be a target one day that benefits more greatly, especially on embedded
 670 2013-03-04 13:29:20 <Diablo-D3> atoms probably lap that shit up
 671 2013-03-04 13:29:34 <Diablo-D3> atoms are so fucking horrid its not funny
 672 2013-03-04 13:29:34 <HM> ARM has conditional instructions
 673 2013-03-04 13:29:39 <HM> which is very cool
 674 2013-03-04 13:29:45 <Diablo-D3> newer ARMs are really good at speculation
 675 2013-03-04 13:29:55 <Diablo-D3> they're not nearly as speculative as modern x86 though
 676 2013-03-04 13:29:56 kritCoin_ has quit (Quit: Page closed)
 677 2013-03-04 13:30:03 <Diablo-D3> go look at a picture of, say, bulldozer
 678 2013-03-04 13:30:22 <Diablo-D3> HUGE GIANT DECODER AND SCHEDULER BLOCK, tiny little integer alus, medium sized fp alus
 679 2013-03-04 13:30:39 <Diablo-D3> the only thing bigger on the die is the giant blob of L3
 680 2013-03-04 13:31:19 <gavinandresen> darn, I was just about to tell kritCoin_ that the sharc will be the BESTEST MOST AWESOMEST MINER EVAH!  And he should spend all his money buying as many of them as possible before everybody else figures it out
 681 2013-03-04 13:31:22 <HM> right but that's natural
 682 2013-03-04 13:31:39 <HM> most programs dont' benefit from quadrupling your integer performance
 683 2013-03-04 13:31:42 <Diablo-D3> what I don't get is why amd stopped at only 2 hardware threads
 684 2013-03-04 13:32:03 <Diablo-D3> they feed the ALUs in a round robin queue
 685 2013-03-04 13:32:23 <HM> well AMD own ATI
 686 2013-03-04 13:32:29 <Diablo-D3> yes, Im well aware of that
 687 2013-03-04 13:32:40 <Diablo-D3> what Im saying is, they already have gone the ibm/sun route
 688 2013-03-04 13:32:43 <HM> they probably see their GPUs as integer workhorses and hope to sell both
 689 2013-03-04 13:32:53 <Diablo-D3> they could have done one core with 4x4 integer ALUs and 4x2 fp ALUs
 690 2013-03-04 13:32:57 <HM> even if it's a low end integrated chip
 691 2013-03-04 13:33:07 <Diablo-D3> and then done an absolutely massive decoder/scheduler that takes 8 hw threads
 692 2013-03-04 13:33:27 * HM shrugs
 693 2013-03-04 13:33:41 <Diablo-D3> you could have up to 16 integer ops execute in a single cycle in a single thread
 694 2013-03-04 13:34:06 <Diablo-D3> having massive instruction level parallelism like that in traditional computing would be amazing
 695 2013-03-04 13:34:34 <HM> would it
 696 2013-03-04 13:34:59 <HM> i can't see it making my web browser any faster
 697 2013-03-04 13:35:10 <Diablo-D3> heh, nothing is making your web browser any faster
 698 2013-03-04 13:35:14 <Diablo-D3> they were written by retards
 699 2013-03-04 13:35:22 <Diablo-D3> firefox and chrome both have some of the worst code I've ever seen
 700 2013-03-04 13:35:40 <HM> well FF is old and has a crufty architecture
 701 2013-03-04 13:35:47 <HM> i doubt chrome though, the code isn't that bad
 702 2013-03-04 13:36:00 <Diablo-D3> chrome is just as crufty
 703 2013-03-04 13:36:12 <Diablo-D3> remember, apple started webkit using konqueror's html engine
 704 2013-03-04 13:36:28 <Diablo-D3> khtml is older than firefox
 705 2013-03-04 13:36:44 <Diablo-D3> not as old as the entire monolithic netscape codebase
 706 2013-03-04 13:36:50 <Diablo-D3> but its still rather goddamned old
 707 2013-03-04 13:36:54 <HM> yes but that's webkit
 708 2013-03-04 13:36:58 <HM> that's not chromium
 709 2013-03-04 13:37:01 <Diablo-D3> chrome is a webkit browser.
 710 2013-03-04 13:37:23 t7 has quit (Read error: Connection reset by peer)
 711 2013-03-04 13:37:33 <Diablo-D3> intact, now that opera is also webkit
 712 2013-03-04 13:37:34 <HM> Google do stick to their C++ coding style quite well
 713 2013-03-04 13:37:38 <Diablo-D3> *infact
 714 2013-03-04 13:37:43 <HM> i'm not sure about webkit
 715 2013-03-04 13:37:55 <Diablo-D3> theres only two engines left
 716 2013-03-04 13:38:06 <Diablo-D3> gecko and whatever microsoft calls trident now
 717 2013-03-04 13:38:09 t7 has joined
 718 2013-03-04 13:38:19 <_dr> 14:18  Diablo-D3$ [08:04:54] <_dr> i feel like the time Diablo-D3 told me AVX is nothing more than two fused SSE registers from two cores, to net performance stays the same :)
 719 2013-03-04 13:38:22 <_dr> 14:18  Diablo-D3$ I said Intel's implementation is
 720 2013-03-04 13:38:25 <_dr> ARGH! :)
 721 2013-03-04 13:38:45 <_dr> vector register width has been increased from 128bit (sse 4.2) to 256bit (avx)
 722 2013-03-04 13:38:54 <_dr> these registers live inside a core, they are not shared among cores
 723 2013-03-04 13:38:57 <Diablo-D3> _dr: tl;dr given the performance boost of AVX, use it, even if its not implemented the best way
 724 2013-03-04 13:39:07 <Diablo-D3> _dr: yes, the registers are in the core
 725 2013-03-04 13:39:16 <_dr> Diablo-D3: the performance boost is x2
 726 2013-03-04 13:39:27 <Diablo-D3> the performance boost CAN be x2
 727 2013-03-04 13:39:46 <_dr> peak sp/dp perf IS x2
 728 2013-03-04 13:39:52 <Diablo-D3> yeah, peak
 729 2013-03-04 13:40:02 <_dr> yeah, and i measured it. and net = peak
 730 2013-03-04 13:40:13 <_dr> if your problem allows it, the speedup is 2x
 731 2013-03-04 13:40:27 <Diablo-D3> which is why I said use it if you have it available
 732 2013-03-04 13:40:32 <Diablo-D3> its just not a magic bullet _yet_
 733 2013-03-04 13:40:49 <_dr> of course not, autovectorization is pretty much non-existant
 734 2013-03-04 13:40:53 <_dr> except in gpus :_
 735 2013-03-04 13:42:36 <Diablo-D3> well it depends
 736 2013-03-04 13:43:01 <Diablo-D3> but yeah, I'll agree, gcc nor clang is all that good at it
 737 2013-03-04 13:43:06 <_dr> nothing is
 738 2013-03-04 13:43:11 <_dr> not even icc
 739 2013-03-04 13:43:24 <_dr> you can have a look at ispc at githut, it's the best thing there is
 740 2013-03-04 13:43:25 <Diablo-D3> traditional computing isn't suited for it, to be honest
 741 2013-03-04 13:43:33 <_dr> but the abstraction comes at the cost of efficiency
 742 2013-03-04 13:43:51 <HM> AVX is still fp isn't it?
 743 2013-03-04 13:43:54 <Diablo-D3> this is why I'm seriously considering in this thing I'm writing just to use OpenCL explicitly in everything parallel
 744 2013-03-04 13:44:00 <HM> are there plans for 256bit int ops?
 745 2013-03-04 13:44:01 <_dr> if i write assembly for avx i get 100% performance; if i use ispc or cude, i can write really easy codes, but at max. 50-75% peak max
 746 2013-03-04 13:44:10 <_dr> HM: yes in HNI aka avx2
 747 2013-03-04 13:44:11 <Diablo-D3> hm: no, its double wide sse3
 748 2013-03-04 13:44:18 <Diablo-D3> or 4 or whatever
 749 2013-03-04 13:44:21 <Diablo-D3> does integer too
 750 2013-03-04 13:44:23 <_dr> HM: xeon phi (aka intel mic) even has 512bit ints
 751 2013-03-04 13:44:33 <Diablo-D3> _dr: phi isn't what you'd call an x86 cpu though
 752 2013-03-04 13:44:38 <Scrat> which version of BDB is the linux binary built with?
 753 2013-03-04 13:44:39 <_dr> yes it is
 754 2013-03-04 13:44:42 <_dr> Diablo-D3: it's a x86 cpu
 755 2013-03-04 13:44:45 <_dr> 61 of them
 756 2013-03-04 13:44:51 <Diablo-D3> it "is" kind of
 757 2013-03-04 13:44:55 <_dr> no it IS
 758 2013-03-04 13:45:00 <Scrat> no u
 759 2013-03-04 13:45:11 <Diablo-D3> _dr: I can execute x86 asm on it, yes
 760 2013-03-04 13:45:13 <HM> Scrat: i'm not sure. i built it against the wrong one the other week, i think  it compiles against most versions
 761 2013-03-04 13:45:15 <Diablo-D3> do I want to? fuck no.
 762 2013-03-04 13:45:23 <HM> Scrat: it may only be important if you have an existing db?
 763 2013-03-04 13:45:26 <Diablo-D3> I will use intel's OpenCL compiler on that goddamned thing
 764 2013-03-04 13:45:53 <_dr> Diablo-D3: well, do you want x86 asm on a cpu? no you want avx
 765 2013-03-04 13:46:11 <_dr> so not wanting to run x86 asm but ICMI on KNC makes it a non-x86 cpu, right...
 766 2013-03-04 13:46:12 <Scrat> HM: even then detachdb does the trick
 767 2013-03-04 13:46:26 <Diablo-D3> _dr: I consider all SIMD extension sets to be part of x86
 768 2013-03-04 13:46:26 <sipa> Scrat: 4.8
 769 2013-03-04 13:46:49 <Diablo-D3> _dr: its just that intel's compiler for phi is going to be a hell of a lot better suited than treating it as a 64 core cpu
 770 2013-03-04 13:46:51 <Scrat> sipa: ty!
 771 2013-03-04 13:46:53 <HM> Scrat: ty, never knew that
 772 2013-03-04 13:47:02 <_dr> Diablo-D3: no it's not
 773 2013-03-04 13:47:10 <Diablo-D3> dude, Ive looked at it
 774 2013-03-04 13:47:13 <Diablo-D3> its clearly meant for opencl
 775 2013-03-04 13:47:17 kritCoin_ has joined
 776 2013-03-04 13:47:26 <kritCoin_> sorry was afk
 777 2013-03-04 13:47:28 <sipa> HM: BDB is backward-incompatible in the most gruesome way
 778 2013-03-04 13:47:38 <HM> oh?
 779 2013-03-04 13:47:38 <_dr> Diablo-D3: so have i
 780 2013-03-04 13:47:44 <Diablo-D3> the ONLY thing I'd be interested in on that is async messaging across cores using the high bandwidth bus
 781 2013-03-04 13:47:48 <_dr> Diablo-D3: in fact i'm logged into a machine right now
 782 2013-03-04 13:48:12 <_dr> and i can tell you what icc does with c code for it :)
 783 2013-03-04 13:48:14 <sipa> HM: the log files are often even forward-incompatible - you're supposed to actively migrate your databases from one version to another
 784 2013-03-04 13:48:30 <HM> still better than a SQL dump
 785 2013-03-04 13:49:03 <_dr> and all compilers produce shitty code for the platform
 786 2013-03-04 13:49:40 <HM> sipa: http://howfuckedismydatabase.com/
 787 2013-03-04 13:49:52 rdymac has joined
 788 2013-03-04 13:51:27 <Diablo-D3> "did you just tell me to go fuck myself?" "I do believe I did, Bob"
 789 2013-03-04 13:51:52 <Diablo-D3> _dr: they're using icc for that? bleh
 790 2013-03-04 13:52:16 <Diablo-D3> _dr: does gcc support the intrinsics for bizzaro-SSE yet?
 791 2013-03-04 13:53:08 <_dr> i haven't tried
 792 2013-03-04 13:53:21 <_dr> but i don't think so
 793 2013-03-04 13:53:33 <_dr> just check a recent immintrin.h for __m512
 794 2013-03-04 13:54:52 <_dr> i mean, the hardware is ''officially'' released; but as there's almost no documentation available yet, the folks at gcc will have a hard time to include support atm
 795 2013-03-04 13:55:26 <_dr> binutil's gas can't even handle the zmmX registers
 796 2013-03-04 13:56:05 <HM> Linux already uses AVX in its software raid implementation
 797 2013-03-04 13:56:31 <_dr> that's ymmX
 798 2013-03-04 13:56:44 <HM> yeah, i know
 799 2013-03-04 13:57:02 Acciaio has quit (Remote host closed the connection)
 800 2013-03-04 13:57:27 <HM> it's pretty mindboggling that soon everyone will have a cpu capable of 256bit ops
 801 2013-03-04 13:57:46 <HM> i guess they felt the same way in the 8 bit era about 64bit
 802 2013-03-04 13:58:08 <_dr> time for haswell to arrive :)
 803 2013-03-04 13:58:09 <HM> in 10 years we'll be talking about our kilobit ops :P
 804 2013-03-04 13:58:15 <_dr> it'll be the convergence of GPU & CPU
 805 2013-03-04 13:58:28 Jynx has joined
 806 2013-03-04 13:59:31 <_dr> it'll have vector mask registers like IMCI already does, which is essentially what gpu's do to do predication on threads
 807 2013-03-04 14:00:04 darksk1ez has quit (Ping timeout: 276 seconds)
 808 2013-03-04 14:00:22 <_dr> the instruction set is already available through various channels (e.g. the free intel compiler student edition with manuals)
 809 2013-03-04 14:02:07 <HM> how are SHA ASICs doing atm?
 810 2013-03-04 14:02:14 <HM> Are Avalon still the only people shipping?
 811 2013-03-04 14:03:11 <gavinandresen> lol:  http://xkcd.com/1181/
 812 2013-03-04 14:03:26 Acciaio has joined
 813 2013-03-04 14:03:33 <HM> lol truth
 814 2013-03-04 14:05:02 <_dr> heh http://download.intel.com/embedded/processor/whitepaper/327457.pdf
 815 2013-03-04 14:05:04 occulta has quit (Quit: KVIrc 4.1.3 Equilibrium http://www.kvirc.net/)
 816 2013-03-04 14:05:30 <_dr> seems like intel already has a sha256 for haswell
 817 2013-03-04 14:05:39 <_dr> if they put some serious crypto hardware in there i wouldn't be surprised if they can compete with the first generation asics
 818 2013-03-04 14:05:52 <_dr> since intel chips tend to have a lot of cores and are clocked about 10x higher
 819 2013-03-04 14:06:11 agricocb has joined
 820 2013-03-04 14:07:17 * HM sells his BTC and buys Intel stock
 821 2013-03-04 14:07:19 <HM> shovels!
 822 2013-03-04 14:08:31 <HM> I'm a bit baffled as to why you'd put crypto primitives in a CPU
 823 2013-03-04 14:08:39 <HM> like AES-NI
 824 2013-03-04 14:08:51 <HM> it seems like short term thinking
 825 2013-03-04 14:09:09 <Scrat> I have a feeling sipa might want to internet slap me again, but this works really well http://pastebin.com/60ujKcrT
 826 2013-03-04 14:09:32 <_dr> HM: i don't think it'll take intel a lot of time to swap the hardware for something else once aes is broken (whenever that'll be)
 827 2013-03-04 14:09:44 <HM> yes but the instruction set lives on forever
 828 2013-03-04 14:10:16 <HM> your x86 cpu will run 386 code from 1986
 829 2013-03-04 14:10:41 <HM> i know most software will have to either JIT these instructions or make them conditional anyhow, with cpuid
 830 2013-03-04 14:10:45 <_dr> hm, i'm no expert but 1. it's just an addition to the core x86 isa and 2. legacy cost is pretty much irrelevant since the HW will take about >0.00001mm^2 or something i guess :)
 831 2013-03-04 14:10:51 <HM> but it still seems like a burden
 832 2013-03-04 14:11:49 <_dr> HM: do they keep ALL their little ISA additions in future cpus?
 833 2013-03-04 14:11:57 <_dr> i can't seem to find a counter-example
 834 2013-03-04 14:11:59 <HM> I have no idea
 835 2013-03-04 14:12:04 <HM> I would hope so
 836 2013-03-04 14:12:19 <_dr> sse has mmx, avx has sse; imci doesn't have avx, but that's the xeon phi
 837 2013-03-04 14:13:04 <HM> it's desirable because you want your compiler to target a reasonably recent cpu and then let the binaries rot
 838 2013-03-04 14:13:30 <_dr> that's true, then i guess my second argument from before can solve the problem :)
 839 2013-03-04 14:13:30 <HM> I mean, it's only recently that Windows assumed SSE
 840 2013-03-04 14:13:41 <HM> I think Windows 8 is the first version that won't run on the Pentium 2?
 841 2013-03-04 14:13:50 <HM> and Linux just ditched 386 compat
 842 2013-03-04 14:15:08 <_dr> i think the usefulness outweighed the legacy cost.
 843 2013-03-04 14:15:27 <_dr> i know that the ISA and stuff is a pretty small fraction on recent cores
 844 2013-03-04 14:15:45 <_dr> it's all caches and the out of order logic
 845 2013-03-04 14:16:06 Raccoon has joined
 846 2013-03-04 14:16:13 <HM> I'm surprised we don't have mini FPGAs in modern machines
 847 2013-03-04 14:16:29 <_dr> there are advocates of the 10x10 approach
 848 2013-03-04 14:17:04 <_dr> instead of having one general purpose core, have ten different specialised, less-general-purpose cores (and they claim that 'all' problems can be reduced to one of the ten applications)
 849 2013-03-04 14:17:17 <_dr> but CPUs are working way too good to just ditch them, i guess
 850 2013-03-04 14:17:25 <HM> I don't agree with that
 851 2013-03-04 14:17:57 <HM> Microsoft had a nice research project where they talked about ditching virtual memory
 852 2013-03-04 14:18:07 <HM> instead software verification was used to isolate memory
 853 2013-03-04 14:18:19 <HM> i think it was a custom C# runtime
 854 2013-03-04 14:18:54 <HM> Chromes NaCl is sort of the same thing
 855 2013-03-04 14:19:13 <HM> i think x86 makes this difficult though, for instance Google had to align all instructions
 856 2013-03-04 14:19:34 <TD> you're thinking of singularity
 857 2013-03-04 14:19:35 <_dr> hm, i should read up on that, if only i had the time
 858 2013-03-04 14:19:40 <TD> coolest OS research done in recent times, imho
 859 2013-03-04 14:20:38 <HM> I am a firm believer that complexity should be kept in software where it can do the most damage
 860 2013-03-04 14:20:41 <HM> :}
 861 2013-03-04 14:24:57 rdymac has quit (Quit: This computer has gone to sleep)
 862 2013-03-04 14:30:08 bitit has quit (Remote host closed the connection)
 863 2013-03-04 14:30:49 abrkn has joined
 864 2013-03-04 14:31:45 bitit has joined
 865 2013-03-04 14:42:04 Hashdog has joined
 866 2013-03-04 14:53:32 bitit has quit (Remote host closed the connection)
 867 2013-03-04 14:54:55 bitit has joined
 868 2013-03-04 14:58:52 LargoG has quit (Ping timeout: 276 seconds)
 869 2013-03-04 14:59:13 LargoG has joined
 870 2013-03-04 15:04:02 yellowhat has joined
 871 2013-03-04 15:04:07 yellowhat has quit (Read error: Connection reset by peer)
 872 2013-03-04 15:05:26 yellowhat has joined
 873 2013-03-04 15:05:26 yellowhat has quit (Remote host closed the connection)
 874 2013-03-04 15:07:09 B0g4r7_ has quit (Remote host closed the connection)
 875 2013-03-04 15:07:37 FredEE has joined
 876 2013-03-04 15:07:51 <Luke-Jr> http://sourceforge.net/projects/bitcoin/files/Bitcoin/blockchain/ <-- who is maintaining this and why? O.o
 877 2013-03-04 15:08:02 rdymac has joined
 878 2013-03-04 15:10:18 K1773R is now known as OFF!~K1773Rfre@www.darkgamex.ch|K1773R
 879 2013-03-04 15:20:39 sgornick has joined
 880 2013-03-04 15:22:09 <DarkGhost`> you don't like that Luke-Jr?
 881 2013-03-04 15:22:42 <sipa> i wouldn't call it "maintained"
 882 2013-03-04 15:22:51 <sipa> bootstrap.dat took over, and is a much saner solution
 883 2013-03-04 15:22:58 EPiSKiNG has joined
 884 2013-03-04 15:23:30 ageis has joined
 885 2013-03-04 15:26:34 phungus has joined
 886 2013-03-04 15:31:28 <Luke-Jr> sipa: last update this month
 887 2013-03-04 15:31:38 <Luke-Jr> oh, a year ago
 888 2013-03-04 15:31:39 <Luke-Jr> nm
 889 2013-03-04 15:31:39 FredEE has quit (Quit: FredEE)
 890 2013-03-04 15:32:45 rlifchitz has quit (Ping timeout: 246 seconds)
 891 2013-03-04 15:32:55 abrkn has quit (Ping timeout: 245 seconds)
 892 2013-03-04 15:35:14 <jgarzik> Luke-Jr: could put the torrent file in there
 893 2013-03-04 15:35:39 <Luke-Jr> jgarzik: and update the README
 894 2013-03-04 15:38:09 EPiSKiNG is now known as EPiSKiNG-
 895 2013-03-04 15:41:42 <kritCoin_> I dont think Bitcoin can grow to mainstream
 896 2013-03-04 15:42:33 <kritCoin_> it cant handle it coz, if you want to repleace every transaction in all currencies with BT transaction the blockchain would be idioticly large
 897 2013-03-04 15:44:10 <jgarzik> kritCoin: take it to #bitcoin
 898 2013-03-04 15:44:35 FredEE has joined
 899 2013-03-04 15:45:03 rlifchitz has joined
 900 2013-03-04 15:45:17 <Luke-Jr> kritCoin_: you're assuming all transactions occur on the blockchain, which is not the case today and will probably be less in the future
 901 2013-03-04 15:46:06 <gavinandresen> Luke-Jr: take it to #bitcoin
 902 2013-03-04 15:53:11 <Scrat> I probably gave sipa a heart attack. is it ok if I block the main thread for a few milliseconds? http://pastebin.com/60ujKcrT
 903 2013-03-04 15:55:00 drizztbsd has quit (Quit: Konversation terminated!)
 904 2013-03-04 15:57:31 drizztbsd has joined
 905 2013-03-04 15:57:31 drizztbsd has quit (Changing host)
 906 2013-03-04 15:57:31 drizztbsd has joined
 907 2013-03-04 15:58:54 <Scrat> I don't want to spawn new threads (like the original walletnotify) and I want to catch the tx later if my backend is down when it arrives
 908 2013-03-04 16:01:07 <kritCoin_> Luke-Jr: where do most transactions occur, reading the wiki I was under assumption that the blockchain contained all transactions of the currency
 909 2013-03-04 16:01:51 <Luke-Jr> kritCoin_: for some reason, jgarzik and gavinandresen want you to discuss this on #bitcoin despite it being more on-topic here and that channel being busy with other discussions
 910 2013-03-04 16:01:54 <kritCoin_> my wallet has references into the blockchain, no blockchain , = empty wallet
 911 2013-03-04 16:03:19 <kritCoin_> I agree Luke-Jr, this is totally a coding question as the binrary wallet format explicitly has references into the blockchain,,.., if transactions are off blockchain...., where do they "live"?
 912 2013-03-04 16:04:17 <Luke-Jr> kritCoin_: the point is to take it to #bitcoin
 913 2013-03-04 16:05:15 <kritCoin_> bitcoin-dev is only fot assembly questions
 914 2013-03-04 16:05:17 <kritCoin_> )
 915 2013-03-04 16:06:04 <kritCoin_> If you can answer it in #bitcoin Luke-jr it would be cool
 916 2013-03-04 16:07:00 FredEE has quit (Quit: FredEE)
 917 2013-03-04 16:07:19 copumpkin has quit (Ping timeout: 252 seconds)
 918 2013-03-04 16:08:36 copumpkin has joined
 919 2013-03-04 16:08:38 Zarutian has quit (Quit: Zarutian)
 920 2013-03-04 16:11:14 FredEE has joined
 921 2013-03-04 16:15:33 egecko has quit (Quit: ~ Trillian Astra - www.trillian.im ~)
 922 2013-03-04 16:15:34 FredEE has quit (Ping timeout: 245 seconds)
 923 2013-03-04 16:16:02 ProfMac has joined
 924 2013-03-04 16:17:06 Zarutian has joined
 925 2013-03-04 16:18:14 da2ce7 has joined
 926 2013-03-04 16:18:49 Acciaio has quit (Ping timeout: 276 seconds)
 927 2013-03-04 16:20:10 d4de has quit (Ping timeout: 245 seconds)
 928 2013-03-04 16:20:11 da2ce7_d has quit (Ping timeout: 256 seconds)
 929 2013-03-04 16:25:01 Descry has joined
 930 2013-03-04 16:27:29 copumpkin has quit (Ping timeout: 252 seconds)
 931 2013-03-04 16:28:14 copumpkin has joined
 932 2013-03-04 16:29:12 freakazoid has joined
 933 2013-03-04 16:42:39 freakazoid has quit (Ping timeout: 245 seconds)
 934 2013-03-04 16:48:03 rdymac has quit (Quit: This computer has gone to sleep)
 935 2013-03-04 16:49:55 ovidiusoft has quit (Ping timeout: 250 seconds)
 936 2013-03-04 16:51:38 Sorroh has joined
 937 2013-03-04 16:53:21 <Sorroh> does anybody know where I can download a trustworthy download of the complete blockchain, in order to skip my client's synchronisation process?
 938 2013-03-04 16:53:31 t7 has quit (Quit: Leaving)
 939 2013-03-04 16:53:40 owowo has joined
 940 2013-03-04 16:54:13 <sipa> Sorroh: search the forum for bootstrap.dat, and use 0.8.0
 941 2013-03-04 16:54:16 <Luke-Jr> Sorroh: it doesn't have to be trustworthy, there's a torrent
 942 2013-03-04 16:54:40 <jgarzik> Sorroh: the aforementioned torrent, https://bitcointalk.org/index.php?topic=145386.0
 943 2013-03-04 16:54:44 <sipa> the client still fully validates that when importing
 944 2013-03-04 16:55:47 <Sorroh> thanks a lot guys!
 945 2013-03-04 16:57:18 <BlueMatt> gmaxwell: ping
 946 2013-03-04 16:58:04 <gmaxwell> BlueMatt: pong?
 947 2013-03-04 16:58:21 <Scrat> is it ok if I block the main thread for a few milliseconds? http://pastebin.com/60ujKcrT
 948 2013-03-04 16:58:26 <BlueMatt> gmaxwell: fee stuff.
 949 2013-03-04 16:58:28 discrete has joined
 950 2013-03-04 16:58:37 <discrete> ;;ticker
 951 2013-03-04 16:58:37 <gribble> BTCUSD ticker | Best bid: 35.41000, Best ask: 35.45000, Bid-ask spread: 0.04000, Last trade: 35.41000, 24 hour volume: 29560.40897771, 24 hour low: 33.99871, 24 hour high: 35.45000, 24 hour vwap: 34.78664
 952 2013-03-04 16:58:50 sivu has joined
 953 2013-03-04 16:58:51 <BlueMatt> gmaxwell: is splitting ever a legitimate strategy given a reasonably sane priority system even if it doesnt use value
 954 2013-03-04 16:59:27 <Luke-Jr> Scrat: that looks like it might introduce a race if you check for the 1
 955 2013-03-04 16:59:42 <sivu> ehlo world
 956 2013-03-04 16:59:51 <Scrat> Luke-Jr: not checking for the 1, just checking the directory for changes
 957 2013-03-04 17:00:11 <Scrat> with inotify
 958 2013-03-04 17:00:34 <gmaxwell> I don't think it is, given an otherwise sane priority. But I haven't yet convinced myself of it for sure.. Thats why I was throwing out a pure uxto priority. bytes-in-the-uxto can be a scarce resource you consume.
 959 2013-03-04 17:01:29 <BlueMatt> yea
 960 2013-03-04 17:02:04 <Scrat> then I defer its txid to my main backend process which will check if it has been already entered in my DB (this may fire multiple times per transaction)
 961 2013-03-04 17:03:38 Sorroh has quit (Quit: ChatZilla 0.9.90 [Firefox 19.0/20130215130331])
 962 2013-03-04 17:03:57 <Luke-Jr> hmm, Bitcoin-Qt has no way to access the current difficulty? O.o
 963 2013-03-04 17:04:03 <Luke-Jr> (console doesn't count)
 964 2013-03-04 17:04:29 <gmaxwell> BlueMatt: someone was saying in here that bitcoinj used a ping interval of 200ms the other day? is that so?
 965 2013-03-04 17:04:54 <BlueMatt> I dont think its that bad, but I think its reasonably high
 966 2013-03-04 17:05:15 <TD> default is 5sec
 967 2013-03-04 17:05:29 <TD> that said, some app may change it of ocurse
 968 2013-03-04 17:05:34 <BlueMatt> (plus pings after goups of blocks if you are using bloom)
 969 2013-03-04 17:05:41 <TD> right
 970 2013-03-04 17:05:44 <gmaxwell> (200ms * 124 peers = 620 pps = about 400kbit/sec before adding in whatever the ping itself takes)
 971 2013-03-04 17:07:18 <TD> there might be a bug. it's not supposed to be 200msec
 972 2013-03-04 17:07:30 <TD> let me check
 973 2013-03-04 17:08:13 <TD> 200msec sounds like what might be observed if it pinged immediately
 974 2013-03-04 17:08:19 <TD> (upon getting a pong)
 975 2013-03-04 17:08:51 <gmaxwell> 23:18 < muhoo> wow, the ping interval, at least on bitcoinj, looks like about 200ms
 976 2013-03-04 17:08:57 <gmaxwell> thats all I'm going on there.
 977 2013-03-04 17:09:15 <Scrat> what would a double spent transaction look like if you getttransaction its txid? error -5 ?
 978 2013-03-04 17:10:03 <gmaxwell> Scrat: if it's in your wallet, you'll get the transaction with confirms=0.
 979 2013-03-04 17:10:12 <sipa> Scrat: unless you do the doublespend itself, it will never get in your wallet
 980 2013-03-04 17:10:47 <gmaxwell> sipa: huh? not so. e.g. someone makes the 'double spend' first from your point of view, but the network reverses it.
 981 2013-03-04 17:11:20 <TD> hm, it does look like there's a bug with that
 982 2013-03-04 17:11:25 <sipa> gmaxwell: right, but then it's not really the double spend from your point of view, but the original :p
 983 2013-03-04 17:12:01 <gmaxwell> TD: please fix asap? this could be debilitating for the network if a widely deployed client does this.
 984 2013-03-04 17:12:28 <TD> a ping every 200msec is not even going to stress a pocket calculator, but yeah, i'll see what's going on there
 985 2013-03-04 17:12:42 B0g4r7_ has joined
 986 2013-03-04 17:12:44 <gmaxwell> TD: uh. it would run my node out of bandwidth.
 987 2013-03-04 17:12:51 <gmaxwell> if it were done by all my peers.
 988 2013-03-04 17:13:11 <TD> how many peers do you have?
 989 2013-03-04 17:13:13 zackham has quit (Quit: zackham)
 990 2013-03-04 17:13:16 <gmaxwell> 125.
 991 2013-03-04 17:13:37 <gmaxwell> 09:05 < gmaxwell> (200ms * 124 peers = 620 pps = about 400kbit/sec before adding in whatever the ping itself takes)
 992 2013-03-04 17:16:38 <TD> a ping is just the base overhead of a bitcoin message plus a 4 byte nonce. so i guess that's something like 32 bytes for a hash, 12 bytes for command, 4 bytes for nonce, 44 bytes? ish?
 993 2013-03-04 17:17:26 <gmaxwell> plus 20 bytes of IP, and about 28 bytes of TCP. (I was computing 20+28+32 since I didn't remember what the pong itself was)
 994 2013-03-04 17:17:55 <gmaxwell> I thought we had nagle disabled, if not that will help. looking.
 995 2013-03-04 17:18:37 toffoo has joined
 996 2013-03-04 17:18:42 <TD> the pong is the same thing as a ping, size-wise
 997 2013-03-04 17:19:04 <gmaxwell> hm. I don't see it. so maybe we're not disabling it. small blessings.
 998 2013-03-04 17:19:06 <TD> how about making the default 1 second. 44 bytes once a second is trivial compared to the cost of running a full node, even if you have a large number of peers.
 999 2013-03-04 17:19:55 <TD> the algorithm can use improvement anyway, the delta should be some kind of limited backoff rather than constant
1000 2013-03-04 17:20:08 <TD> right, found the bug
1001 2013-03-04 17:21:38 <gmaxwell> TD: Any reason it needs to be that fast? The network radius is many seconds now. I was thinking of conferring DOS when pings come faster than once per second, so actually sending at once per second would be right on the wire.
1002 2013-03-04 17:21:39 gritcoin has joined
1003 2013-03-04 17:23:10 bitit has quit (Ping timeout: 276 seconds)
1004 2013-03-04 17:23:54 sebicas has joined
1005 2013-03-04 17:24:43 <TD> because responsiveness matters, a ton. it's trying to figure out which of the peers it was able to find can shovel it the chain fastest, ie, is not overloaded
1006 2013-03-04 17:24:56 BTCTrader2 has joined
1007 2013-03-04 17:24:57 <TD> because this happens whenever you start the clients, it needs to be fast (<2-3 seconds)
1008 2013-03-04 17:25:28 <TD> pings are super cheap. if we go up to even a pathetic level of traffic like 5-6 transactions per second, invs will be far more expensive bandwidth and cpu wise
1009 2013-03-04 17:25:45 <TD> anyway, once it's synced the chain the value of knowing how responsive your peers are goes down a lot
1010 2013-03-04 17:25:51 rdymac has joined
1011 2013-03-04 17:25:54 <TD> hence, adaptive ping times would be useful
1012 2013-03-04 17:26:06 gritcoin has quit (Client Quit)
1013 2013-03-04 17:26:35 <gmaxwell> INVs end up batched. I just pointed out that because of varrious overhead those 'super cheap' pings are getting up to a half a megabit per second outbound for a node with all connections in use.
1014 2013-03-04 17:26:39 <TD> seeing as pings are not very large, we could just disable checksumming for them and that'd eliminate the cpu load.
1015 2013-03-04 17:26:53 <gmaxwell> You can measure responsiveness with a single ping. You don't need many pings for for that.
1016 2013-03-04 17:27:26 <TD> unfortunately nodes responsiveness can fluctuate wildly given their single threaded nature. some other client connects and suddenly responsiveness tanks.
1017 2013-03-04 17:27:43 <discrete> ;;genrate 930
1018 2013-03-04 17:27:44 <gribble> The expected generation output, at 930.0 Mhps, given difficulty of 4367876.00084, is 0.107077916175 BTC per day and 0.00446157984062 BTC per hour.
1019 2013-03-04 17:27:52 <TD> also, if your node "goes away", you want to find out so you can replace it. anyway, one second is pretty reasonable.
1020 2013-03-04 17:30:57 <gmaxwell> TD: nothing on the network happens within a one second time frame. The time it takes to get a message across the network is multiple seconds. Any sane peer has multiple connections. 1 second is still on the order of (20+28+12+32)*8*125 = 92kbit/sec for a node with 125 peers.
1021 2013-03-04 17:31:10 <gmaxwell> I do not think this is reasonable in the general case.
1022 2013-03-04 17:31:30 discrete has quit ()
1023 2013-03-04 17:32:03 <gmaxwell> esp unlike blocks and such, pings are not reduced by recieving one first from another peer.
1024 2013-03-04 17:32:20 sebicas has quit (Quit: sebicas)
1025 2013-03-04 17:32:30 kicek has quit (Read error: Connection reset by peer)
1026 2013-03-04 17:32:31 discrete has joined
1027 2013-03-04 17:32:46 <TD> even with pings, these connections are much cheaper than a full node connection because they're bloom filtered. your bandwidth usage will still be lower, much lower, with 125 spv clients than 125 full clients
1028 2013-03-04 17:33:18 <TD> so yes, whilst pinging behaviour could be smarter and i just fixed that bug, the exact time to use is something that can be debated forever. right now a far more important improvement is to actually store and use addr broadcasts
1029 2013-03-04 17:33:24 <gmaxwell> I'm thinking your behavior should be more like 1 per N seconds. such that you send out one per second to all of your peers togeather.
1030 2013-03-04 17:33:48 grau_ has joined
1031 2013-03-04 17:34:23 <gmaxwell> I'm giving you a chance to convince me that 1/sec is actually needed, because I'm writing code to blacklist peers that ping that fast right now.
1032 2013-03-04 17:34:29 sebicas has joined
1033 2013-03-04 17:35:28 Descry has quit (Quit: Leaving)
1034 2013-03-04 17:35:28 <sipa> gmaxwell: i'm sure TD would welcome a patch to improve the ping behaviour :)
1035 2013-03-04 17:35:37 grau has quit (Ping timeout: 252 seconds)
1036 2013-03-04 17:35:48 <TD> there are much more important things to do than that. even if by some incredible co-incidence *all* your peers end up being SPV, 100 kbit/second is less than ISDN speed.
1037 2013-03-04 17:36:13 <TD> i'll make it adaptive at some point so it monitors responsiveness more when it needs to and less when it doesn't. until then we can live with it.
1038 2013-03-04 17:36:34 <gmaxwell> TD: it's also as much as we'll spend transmitting the blockchain on average when all our peers are full nodes.
1039 2013-03-04 17:36:58 <gmaxwell> (or at least within a small factor of it)
1040 2013-03-04 17:37:17 <TD> assuming small blocks :)
1041 2013-03-04 17:37:19 b4tt3r135 has joined
1042 2013-03-04 17:37:49 <TD> anyway, if I want to DoS a node, there are much better ways than ping floods.
1043 2013-03-04 17:38:02 <gmaxwell> Assuming the _current_ behavior. Come on, this is abusive. This serves little benefit for the SPV client, and _no_ benefit for the full node operators.
1044 2013-03-04 17:38:43 <gmaxwell> the purpose of DoSing on this is so that bugs like this one don't end up seriously impacting the network, less than intentional attacks. As you note, a getheader flood or something would be a better attack.
1045 2013-03-04 17:38:45 <TD> it does have benefit for full nodes - if they get overloaded and can't keep up, some of their peers will disconnect, thus reducing the load
1046 2013-03-04 17:38:54 <TD> it's a form of continuous load balancing, albeit a crude one
1047 2013-03-04 17:39:11 nOgAnOo has joined
1048 2013-03-04 17:39:11 nOgAnOo has quit (Client Quit)
1049 2013-03-04 17:39:26 <gmaxwell> ...
1050 2013-03-04 17:39:27 <TD> it'd be better for the node itself to do that, of course. but they don't
1051 2013-03-04 17:39:29 nOgAnOo has joined
1052 2013-03-04 17:39:51 <TD> would you be happy with a 2 second ping time?
1053 2013-03-04 17:41:42 <gmaxwell> It wouldn't run afoul of the DoS I was planning on putting in. Though I'd think defaulting to 10 would make more sense. If you want to measure responsiveness... just time your infrequent pings. If you send one per ten seconds, you can disconnect peers that take— say— >2 seconds to respond too often.
1054 2013-03-04 17:42:40 <gmaxwell> (I was contemplating >1/sec as what adds DoS to a node, so a 1/sec ping would bump into it some, 0.5/sec wouldn't.)
1055 2013-03-04 17:44:06 <TD> i'll change it to every 2 seconds for now and raise it further when i revisit the networking code and make use of pinging smarter. it was on my todo list anyway, right now we don't wait to start syncing the chain but start and then abandon/restart if a much faster peer comes along. so there's lots of ways to make it better. but storing addr broadcasts to more evenly distribute the load is next for networking
1056 2013-03-04 17:44:37 <gmaxwell> OKAY.
1057 2013-03-04 17:45:16 * sipa puts in earplugs
1058 2013-03-04 17:45:29 <gavinandresen> two…   but eleven is my favorite number….  (then I run away and hide)
1059 2013-03-04 17:46:02 <HM> if it helps, although it's not portable, you can query the tcp socket for the round trip time
1060 2013-03-04 17:46:31 <HM> getsocketopt() ...SOL_TCP, tcp_info struct
1061 2013-03-04 17:46:37 <HM> it's a moving average
1062 2013-03-04 17:47:44 <Luke-Jr> gavinandresen: you should make Base11Coin
1063 2013-03-04 17:48:31 <sipa> oh, so you can send elevenhundred eleventy eleven coins!
1064 2013-03-04 17:49:54 i2pRelay has quit (Remote host closed the connection)
1065 2013-03-04 17:50:16 <HM> fun fact: 'hundred' hasn't always meant 100
1066 2013-03-04 17:50:33 i2pRelay has joined
1067 2013-03-04 17:51:42 <HM> i think 11^3 is pushing it a bit though sipa
1068 2013-03-04 17:52:06 <HM> err 11^2
1069 2013-03-04 17:52:08 <HM> maybe not
1070 2013-03-04 17:52:27 <HM> https://en.wikipedia.org/wiki/Hundred_(word) 11^2 is pretty close to a "long hundred" :P
1071 2013-03-04 18:04:59 egecko has joined
1072 2013-03-04 18:05:14 <TD> HM:  that IS a fun fact
1073 2013-03-04 18:07:33 <HM> :}
1074 2013-03-04 18:07:39 <TD> there's a lot of pull requests sort of stacked up
1075 2013-03-04 18:07:58 <BlueMatt> TD: theres always a lot of pulls stacked up
1076 2013-03-04 18:08:05 chmod755 has joined
1077 2013-03-04 18:08:09 <TD> well, exactly
1078 2013-03-04 18:08:32 <TD> it's not really great to let them bitrot there
1079 2013-03-04 18:09:12 <BlueMatt> bitcoin should hire a code-review intern
1080 2013-03-04 18:09:24 <BlueMatt> which is kinda a oxymoron, but...maybe grad student
1081 2013-03-04 18:11:43 <TD> "It just isn't very fun. We've learned from game currencies that people like a little inflation in their economies. But Bitcoin is built to deflate. And we've just seen, culturally, people don't like deflation."
1082 2013-03-04 18:11:54 <BlueMatt> google summer of code project: review all our pull requests and make thorough recommendations for each one
1083 2013-03-04 18:12:21 <BlueMatt> TD: where is this?
1084 2013-03-04 18:12:33 <TD> http://www.guardian.co.uk/business/2013/mar/04/bitcoin-currency-of-vice
1085 2013-03-04 18:12:48 <TD> more brilliant PR. well, at least it points out that dollars are no better
1086 2013-03-04 18:12:53 <TD> hope the new website can go live soon
1087 2013-03-04 18:14:09 <Luke-Jr> what new website?
1088 2013-03-04 18:14:44 <TD> see the forum thread
1089 2013-03-04 18:15:20 <Luke-Jr> where? why are people still using the trollforums for productive efforts? <.<
1090 2013-03-04 18:15:36 <BlueMatt> the bitcoinfoundation forums
1091 2013-03-04 18:15:41 <Luke-Jr> oh, those forums
1092 2013-03-04 18:16:02 <gavinandresen> … threads on both forums, now… but foundation thread is much more productive
1093 2013-03-04 18:16:37 <BlueMatt> although the content is much better, I have to admit I find the new design slightly less professional
1094 2013-03-04 18:17:48 <Luke-Jr> https://bitcoinfoundation.org/forum/index.php?/topic/94-upgrade-bitcoinorg/ <-- this one?
1095 2013-03-04 18:17:55 <BlueMatt> yes
1096 2013-03-04 18:18:14 <gmaxwell> I think the design is considerably less professional, but I'm not sure what to suggest. "yuck" doesn't help.
1097 2013-03-04 18:18:16 <Luke-Jr> not so sure about promoting a webwallet prominently
1098 2013-03-04 18:18:28 <phantomcircuit> jgarzik, is https://github.com/jgarzik/python-bitcoinrpc maintained?
1099 2013-03-04 18:18:44 <wumpus> great, the bitcoinfoundation forum is not readable without signing up, and it's not possible to sign up
1100 2013-03-04 18:19:08 <BlueMatt> wumpus: it auto-registers you with a membership
1101 2013-03-04 18:19:17 <Luke-Jr> wumpus: https://bitcoinfoundation.org/support
1102 2013-03-04 18:19:18 <phantomcircuit> BlueMatt, ILLUMINATI CONSPIRACY
1103 2013-03-04 18:19:22 * phantomcircuit runs
1104 2013-03-04 18:19:26 <wumpus> oh, so I should already have an account
1105 2013-03-04 18:19:32 <BlueMatt> yes
1106 2013-03-04 18:19:50 <BlueMatt> gmaxwell: yes, that is exactly where I am
1107 2013-03-04 18:20:20 coolsa has joined
1108 2013-03-04 18:20:23 <HM> I'd be ok with Bitcoin being the currency of vice
1109 2013-03-04 18:20:26 <BlueMatt> Im an awful designed, but...I dunno, the fact that everything is white is maybe a place to start...what if the background of the center (where text is) were something slightly off-white
1110 2013-03-04 18:20:34 <BlueMatt> s/designed/designer/
1111 2013-03-04 18:20:43 <wumpus> I see, somehow missed the mail
1112 2013-03-04 18:21:19 <TD> basically anything is better than the current website
1113 2013-03-04 18:21:21 iwilcox has joined
1114 2013-03-04 18:21:29 <TD> a 404 not found page would be better than the current website
1115 2013-03-04 18:21:41 <BlueMatt> TD: in content, but the current site has a (somewhat) professional-looking design
1116 2013-03-04 18:21:44 <TD> here's a funny thing. try clicking "News" on bitcoin.org - what do you see?
1117 2013-03-04 18:21:53 <gmaxwell> TD: I think the current website is reasonably good. It's clean, maybe a little text busy.
1118 2013-03-04 18:22:01 <BlueMatt> the value of which cannot be understated when trying to get people to trust you with their money
1119 2013-03-04 18:22:16 <TD> the current website is nowhere near professional. and (pet peeve) the text is all grey on white.
1120 2013-03-04 18:22:41 <TD> i agree the new site looks a little childish, perhaps because of the simple icons
1121 2013-03-04 18:22:44 <BlueMatt> TD: some of it is grey on grey
1122 2013-03-04 18:23:02 <gmaxwell> TD: and the enter the matrix javascript drug trip in the background?
1123 2013-03-04 18:23:05 <gmaxwell> :P
1124 2013-03-04 18:23:11 <TD> that's cool :)
1125 2013-03-04 18:23:29 <BlueMatt> well, despite it being TD's pet peve, going in chrome's developer bar and setting the background-color of #content to something slightly grey-er actually helps a lot to my eye
1126 2013-03-04 18:23:40 <gmaxwell> It needs a few under construction digging guys too. And that dancing baby gif. :P
1127 2013-03-04 18:23:48 b4tt3r135 has quit (Ping timeout: 245 seconds)
1128 2013-03-04 18:23:56 <wumpus> lol
1129 2013-03-04 18:23:57 <TD> anyway, i can think of at least one organization that became very successful with a home page dominated by childish looking primary-colours :)
1130 2013-03-04 18:24:00 <petertodd> "For Enthusiasts" <- I hate that term.
1131 2013-03-04 18:24:04 <BlueMatt> the matrixy thing doesnt look terrible if you bring the content out some with some grey background
1132 2013-03-04 18:24:27 <wumpus> gmaxwell: it needs a java applet with water ripple effect :P
1133 2013-03-04 18:24:56 <gmaxwell> wumpus: Yes! thats the internet I remember!.
1134 2013-03-04 18:24:58 i2pRelay has quit (Remote host closed the connection)
1135 2013-03-04 18:25:23 <TD> http://www2.warnerbros.com/spacejam/movie/jam.htm
1136 2013-03-04 18:25:25 <BlueMatt> oh, and the chose your wallet buttons hover off the side of the white background and into the matrix
1137 2013-03-04 18:25:42 <TD> BlueMatt: yeah, i'm not sure why the background is not itself white
1138 2013-03-04 18:27:18 <jgarzik> phantomcircuit: yes
1139 2013-03-04 18:27:39 <phantomcircuit> gavinandresen, where are the bylaws for the foundation?
1140 2013-03-04 18:28:03 <gmaxwell> TD: in any case, certantly it's not primary colors that create the poor impression. If I were able to articulate it, I suppose I'd be a designer.
1141 2013-03-04 18:28:09 Hashdog has quit (Quit: Leaving.)
1142 2013-03-04 18:28:13 freakazoid has joined
1143 2013-03-04 18:28:33 Hashdog has joined
1144 2013-03-04 18:28:54 <TD> it's still important to get it out there and launched. the graphics design can be tweaked later. the most important thing is content
1145 2013-03-04 18:29:15 <TD> bad PR will continue until it's brain-dead simple for journalists to get information that dispels their preconceptions.
1146 2013-03-04 18:29:20 <gmaxwell> TD: I'm confused then. Why is this important? If it's not an improvement why does it need to be launched?
1147 2013-03-04 18:29:23 <TD> the new content doesn't do a whole lot for that.
1148 2013-03-04 18:29:25 <TD> it is an improvement
1149 2013-03-04 18:29:32 <TD> the problem with the existing site is content (or lack of it)
1150 2013-03-04 18:29:46 <TD> at least this new site has some, and it does a pretty good job of hitting the high level points
1151 2013-03-04 18:29:57 <gmaxwell> Such as?
1152 2013-03-04 18:29:58 <TD> i'd want to introduce a press center at some point, and port the faq from the wiki to the main site
1153 2013-03-04 18:30:11 <TD> why would a business want to use bitcoins? how does it work? how can i get started/
1154 2013-03-04 18:30:38 <TD> stuff we take for granted but the current website doesn't even discuss
1155 2013-03-04 18:30:42 <gavinandresen> phantomcircuit: https://github.com/pmlaw/The-Bitcoin-Foundation-Legal-Repo/tree/master/Bylaws
1156 2013-03-04 18:30:49 <gavinandresen> phantomcircuit: "patches welcome"
1157 2013-03-04 18:30:49 i2pRelay has joined
1158 2013-03-04 18:31:13 <gmaxwell> TD: it does, mostly through directing to weusecoins.
1159 2013-03-04 18:31:16 <TD> the foundations website is hosed
1160 2013-03-04 18:31:26 <petertodd> gmaxwell: Color choice means a lot for designating trust, consider: www.cibc.ca, www.tdcanadatrust.com, www.rbcroyalbank.com, www.paypal.com, www.pcfinancial.ca, citibank.com <- every single one has a color scheme consisting of at most two colors, and generally strong colors like reds and blues, rainbows just say "fun" which isn't want we want to build trust
1161 2013-03-04 18:31:42 <petertodd> Bitcoin needs some seriousness to it
1162 2013-03-04 18:31:45 rdymac has quit (Quit: This computer has gone to sleep)
1163 2013-03-04 18:32:27 <gmaxwell> petertodd: and you thought your artschool training was all fluff.
1164 2013-03-04 18:32:27 <jgarzik> TD: +1 RE "why/how/how/etc" questions
1165 2013-03-04 18:32:27 <TD> i guess you could submit an issue or a pull req to make the top nav-bar less colourful
1166 2013-03-04 18:32:40 <jgarzik> TD: Businesses need a sane introduction, !wiki
1167 2013-03-04 18:32:53 holorga has joined
1168 2013-03-04 18:32:55 <jgarzik> TD: Educated users need a less ideological, more nuts-n-bolts introduction
1169 2013-03-04 18:33:12 <petertodd> gmaxwell: Heh, businesses live and die on shitty design and messaging.
1170 2013-03-04 18:33:24 <gmaxwell> We also need to be very careful that we're not encouraging unlawful activity, or feeding irrational exuberance. People will be irrational on their own, the site being conservative is an asset.
1171 2013-03-04 18:33:35 <jgarzik> +1
1172 2013-03-04 18:34:08 <kritCoin_> should this belong in #bitcoin instead #bitcoin-dev
1173 2013-03-04 18:34:14 <TD> that said, bank websites, are extremely variable ….. check out this nonsense https://wellsoffice.wellsfargo.com/ceoportal/signon/loader.jsp
1174 2013-03-04 18:34:14 <petertodd> gmaxwell: One of the things I like about the current bitcoin.org is it's both conservative, the simple blue and yellow, and yet doesn't look like it's trying to sell anything at all.
1175 2013-03-04 18:34:31 <wumpus> petertodd: agreed
1176 2013-03-04 18:34:34 <jouke> +1
1177 2013-03-04 18:34:42 <kritCoin_> should this belong in #bitcoin instead of #bitcoin-dev
1178 2013-03-04 18:34:52 <BlueMatt> TD: doesnt mean we have to be
1179 2013-03-04 18:35:13 <kritCoin_> Oh, I was ordered off by jarzik for talking about hashchains
1180 2013-03-04 18:35:14 <gmaxwell> I haven't reviewed the 'new' site since the last time we talked about it here, but the prior version had some serious messaging flaws that Saivann Carignan did not understand.
1181 2013-03-04 18:35:15 <wumpus> TD: yeah that one's hilarious
1182 2013-03-04 18:35:24 <kritCoin_> weird,.., mods are so wimsical
1183 2013-03-04 18:35:26 <petertodd> Similarly, the bitcoinfoundation.org website talks about "transacting on your own terms", and nothing more. Although I think the references to increases in value in the blog postings on the site are a bad move.
1184 2013-03-04 18:35:31 <TD> probably, making the text smaller would make it look a bit more professional
1185 2013-03-04 18:35:44 <TD> kritCoin_: yes, we're not very consistent. that said, developing the website is as important as the code, imo
1186 2013-03-04 18:35:46 <gavinandresen> kritCoin_ : anybody who's contributed more than 100 hours of coding gets to break the rules 20% of the time.
1187 2013-03-04 18:35:57 <gavinandresen> … that's in the Official Rules.
1188 2013-03-04 18:36:10 <kritCoin_> TD, yeah so is bruching my teeth, but i dont go into it here do i?
1189 2013-03-04 18:36:12 <petertodd> TD: heh, that wells fargo site doesn't even load for me.
1190 2013-03-04 18:36:20 <TD> petertodd: make sure it's https
1191 2013-03-04 18:36:22 <OneMiner> A selling point for me is that I see bitcoin as devoid of passion or idology. It's merely a system, nearly mechanical in nature and therefore worthy of near absolute trust (barring math failing or something like that).
1192 2013-03-04 18:36:23 <TD> for some reason pure http hangs
1193 2013-03-04 18:36:27 <TD> ironically
1194 2013-03-04 18:36:36 <petertodd> TD: yup, https still hangs for me
1195 2013-03-04 18:36:52 <TD> odd
1196 2013-03-04 18:36:59 <holorga> hey, people, quick question on bitcoind api.. are transactions in a list I receive when I call listtransactions ordered by when my client found out about them, or by the 'official network transaction order'.. I'm asking this as I'd like to know if their order can change..
1197 2013-03-04 18:37:06 <petertodd> TD: Oh, and reminds me, bitcoin.org really should be https too, it's another one of those subtle trust-builders. Ideally with a green cert, but I know they are expensive.
1198 2013-03-04 18:37:28 <kritCoin_> I am going to try putting the blockchain in oracle
1199 2013-03-04 18:37:30 <TD> it needs to be moved off github for that, i suspect
1200 2013-03-04 18:37:31 <TD> but yes.
1201 2013-03-04 18:37:41 <kritCoin_> with the new OCI features it will blow i think
1202 2013-03-04 18:37:41 <BlueMatt> holorga: its not in the order of blocks, Im not even sure if its in the order received and not just random
1203 2013-03-04 18:37:43 <TD> anyway, home time
1204 2013-03-04 18:37:53 <phantomcircuit> gavinandresen, lol i printed the page and firefox decided to print it as like 0.5 pt font because of the long lines
1205 2013-03-04 18:37:54 <phantomcircuit> sigh
1206 2013-03-04 18:38:43 <holorga> BlueMatt: do you know if their order can change? I'm asking as I'd like to implement a 'gimme all transactions up to a specific one' call on top of this api.. something a lot of people did implement I suppose.
1207 2013-03-04 18:39:06 <BlueMatt> Id have to read the code, but I have to leave for class in like 1 minute, sorry
1208 2013-03-04 18:39:13 <helo> it's probably not safe to rely on the order never changing
1209 2013-03-04 18:39:18 <kritCoin_> 36.7
1210 2013-03-04 18:39:21 <kritCoin_> woot
1211 2013-03-04 18:39:21 <holorga> kk tnx
1212 2013-03-04 18:39:41 <phantomcircuit> holorga, there's an api call to get transactions after a block number
1213 2013-03-04 18:39:43 <kritCoin_> !ticker --last
1214 2013-03-04 18:39:43 <gribble> 36.51985
1215 2013-03-04 18:39:49 <phantomcircuit> assuming there isn't a large reorg you can use that
1216 2013-03-04 18:40:03 <jgarzik> kritCoin_: take the noise to #bitcoin
1217 2013-03-04 18:40:41 TD has quit (Quit: TD)
1218 2013-03-04 18:40:50 <holorga> phantomcircuit: oh, that could work, tnx
1219 2013-03-04 18:41:30 <phantomcircuit> holorga, getblockcount -> getblockhash -> listsinceblock
1220 2013-03-04 18:41:46 zooko has joined
1221 2013-03-04 18:42:10 kritCoin_ has quit (Quit: Page closed)
1222 2013-03-04 18:42:17 <phantomcircuit> holorga, listsince (getblockhash (getblockcount - 10))
1223 2013-03-04 18:42:20 <phantomcircuit> something like that
1224 2013-03-04 18:42:31 <phantomcircuit> obviously rpc calls aren't nestable but you get the idea
1225 2013-03-04 18:43:24 <gmaxwell> kritCoin_: I consider bitcoin.org part of the project development.  It's just more software, though it's 100% UI. Some software is softer than others.
1226 2013-03-04 18:44:10 <holorga> phantomcircuit: yeah, that's good. and I can decide how far back in blocks I want to look in order to update confirmation numbers for transactions I already know about. great. thanks.
1227 2013-03-04 18:44:26 <phantomcircuit> holorga, yup
1228 2013-03-04 18:44:46 <phantomcircuit> you could also keep track of which block index transactions were in
1229 2013-03-04 18:44:51 <phantomcircuit> and then do something like
1230 2013-03-04 18:44:58 <phantomcircuit> select max(block_index) from transactions;
1231 2013-03-04 18:45:08 <phantomcircuit> but probably that's not necessary :)
1232 2013-03-04 18:47:10 <holorga> phantomcircuit: yeah, I'll do that, I'll keep 'last block seen' somewhere, so that I can turn off my wallet watcher and be sure that it will fetch all the changes that happened when it was down..
1233 2013-03-04 18:48:24 <phantomcircuit> holorga, yeah i just run through the last 10k transactions on the first run
1234 2013-03-04 18:48:30 <phantomcircuit> it's not efficient but it's very simple
1235 2013-03-04 18:49:28 <holorga> phantomcircuit: :) yeah, I used to do something like that, I want to do it properly this time.
1236 2013-03-04 18:54:20 i2pRelay has quit (Remote host closed the connection)
1237 2013-03-04 18:55:01 i2pRelay has joined
1238 2013-03-04 18:55:53 sebicas has quit (Quit: sebicas)
1239 2013-03-04 18:56:51 <BlueMatt> somehow I remember someone pointing to a ml post which suggested the use of proof-of-work to create a consensus system long before satoshi, but I cant find it.  Anyone have a link to such a post lying around?
1240 2013-03-04 18:57:58 Acciaio has joined
1241 2013-03-04 18:58:19 bock has joined
1242 2013-03-04 18:59:13 <BlueMatt> or evidence thereof?
1243 2013-03-04 19:00:45 <Acciaio> Hi all, there is a way to stop bitcoind  retransmitting unconfirmable transactions?
1244 2013-03-04 19:02:32 <gmaxwell> Acciaio: run salvage wallet on the wallet in question (after backing it up).
1245 2013-03-04 19:02:53 <gavinandresen> … a purge unconfirmed transactions feature would be nice to have....
1246 2013-03-04 19:04:02 <Prattler> if only we had any devs around!
1247 2013-03-04 19:04:04 <Prattler> if only...
1248 2013-03-04 19:04:06 <petertodd> Salvage wallet? What's this new feature?
1249 2013-03-04 19:04:32 <sipa> petertodd: basically: wipe wallet, try to import anything that looks like a secret key from the old one, rescan
1250 2013-03-04 19:04:37 <gmaxwell> What I think had been suggested before would be that it should be possible to put transactions into a "request-cancel" state where it stops rebroadcasting and then if it doesn't confirm in XX hours, it goes into a "canceled state" where it frees up the inputs.
1251 2013-03-04 19:04:37 <petertodd> gavinandresen: I noticed namecoin seems to have a delete tx feature.
1252 2013-03-04 19:04:57 <gavinandresen> well if namecoin jumps off a bridge...
1253 2013-03-04 19:05:33 <sipa> gmaxwell: yes, i've suggested a refactor of the wallet code before, where we don't use a vfSpent vector per wallet entry, but use a mempool-like mapTxNext in the wallet to find which have been spent
1254 2013-03-04 19:05:46 <sipa> gmaxwell: that means we can disable/enable transactions on the fly
1255 2013-03-04 19:06:09 <petertodd> gavinandresen: "But they have a very nice parachute to do it with!"
1256 2013-03-04 19:06:28 nOgAnOo has quit (Quit: Blessed are the peacemakers, For they shall be called sons of God. Blessed are those who are persecuted for righteousness' sake, For theirs is the kingdom of heaven. Blessed are you when they revile and persecute you, and say all kinds of evil against yo)
1257 2013-03-04 19:06:45 <petertodd> sipa: Ah I see. I've got a few wallets messed up from my nLockTime and large tx experiments...
1258 2013-03-04 19:06:53 <gmaxwell> Though.. hm. I guess it should show  "requested-cancel" -> "pending-cancel" -> "canceled"  the first is no rebroadcast, the second is the inputs get respent, the third is that its finally conflicted in the chain.
1259 2013-03-04 19:06:54 D34TH has joined
1260 2013-03-04 19:06:55 D34TH has quit (Changing host)
1261 2013-03-04 19:06:55 D34TH has joined
1262 2013-03-04 19:07:20 <gmaxwell> (and if its already double spent in the chain, I suppose you can go straight from "requested-cancel" to "canceled" instantly)
1263 2013-03-04 19:08:10 <gmaxwell> the yucky thing is that you don't want to cancel some txn, pay someone with another txn, and then have the first go through anyways.
1264 2013-03-04 19:08:32 <sipa> gmaxwell: it'd be useful to make any transaction that spends an input of your also in your wallet - not because it affects you wallet directly, but it would make the mapTxNext tracing system find conflicts
1265 2013-03-04 19:08:41 <petertodd> Putting some sort of reasonable "in chain tx must have n confirms" would help I think.
1266 2013-03-04 19:08:41 <sipa> so you can do negative confirmations & stuff
1267 2013-03-04 19:09:17 <petertodd> (prevent users from touching tx's until we're sure they're never going to confirm)
1268 2013-03-04 19:11:13 <sipa> yeah
1269 2013-03-04 19:12:16 FredEE has joined
1270 2013-03-04 19:12:31 drizztbsd has quit (Remote host closed the connection)
1271 2013-03-04 19:12:43 <phantomcircuit> petertodd, you can only be sure once there is a conflict
1272 2013-03-04 19:12:55 <phantomcircuit> especially since eligius will mine things that can appear will never confirm
1273 2013-03-04 19:12:56 <sipa> you can never be _sure_ :)
1274 2013-03-04 19:13:01 Acciaio has quit (Ping timeout: 276 seconds)
1275 2013-03-04 19:13:01 random_cat has quit (Ping timeout: 276 seconds)
1276 2013-03-04 19:13:03 <phantomcircuit> but with 1% of the mining power it can take a while
1277 2013-03-04 19:13:12 <petertodd> sipa: Seems that salvage wallet can only take one bad wallet, and create another from it; it'd be good to be able to take multiple wallets, and merge all the private keys into one clean wallet.
1278 2013-03-04 19:13:13 whizter has joined
1279 2013-03-04 19:13:13 <phantomcircuit> sipa, not just theoretically
1280 2013-03-04 19:13:17 <phantomcircuit> actually
1281 2013-03-04 19:13:27 <gmaxwell> Thats why I was suggesting it only showed canceled once its conflicted. :(
1282 2013-03-04 19:13:39 <phantomcircuit> gmaxwell, a good suggestion :)
1283 2013-03-04 19:14:00 <sipa> gmaxwell: conflicted with something that has at least N confirmations
1284 2013-03-04 19:14:02 <petertodd> gmaxwell: That's my thoughts too, just don't let the user touch it until it has enough confirms that we're sure nothing is going to happen. (also should check that chain height is reasonable)
1285 2013-03-04 19:14:08 <gmaxwell> (frown because that means you can't ever go from pending to cancled fast)
1286 2013-03-04 19:14:33 <gmaxwell> sipa: yes, N might be 1 though.
1287 2013-03-04 19:14:54 <petertodd> phantomcircuit: It's the weird thing about PoW blockchains: they can be trustworthy while people use them, but if they ever fall out of popularity, eventually an attacker can make the best chain be anything - serious implications for timestamping...
1288 2013-03-04 19:15:37 <Prattler> (someone please create an LTC attack pool that I can join)
1289 2013-03-04 19:16:07 <gmaxwell> Prattler: I don't know what an "attack pool" is but I'm pretty sure it doesn't belong in here.
1290 2013-03-04 19:16:53 Acciaio has joined
1291 2013-03-04 19:19:11 random_cat has joined
1292 2013-03-04 19:26:26 bitafterbit has joined
1293 2013-03-04 19:34:23 HM has quit ()
1294 2013-03-04 19:34:44 ashams has joined
1295 2013-03-04 19:34:44 ashams has quit (Changing host)
1296 2013-03-04 19:34:44 ashams has joined
1297 2013-03-04 19:38:24 HM has joined
1298 2013-03-04 19:38:31 <Scrat> gavinandresen: can't the foundation hire a professional to design bitcoin.org?
1299 2013-03-04 19:38:42 <Scrat> this sucks: https://bitcointalk.org/index.php?topic=148712.0
1300 2013-03-04 19:39:12 <HM> i agree, there's not nearly enough <blink> tags for my liking
1301 2013-03-04 19:39:35 <gavinandresen> Scrat: do you like the look of bitcoinfoundation.org ?
1302 2013-03-04 19:40:03 <gmaxwell> I like bitcoinfoundation.org.
1303 2013-03-04 19:40:05 <Scrat> I do
1304 2013-03-04 19:40:26 <Scrat> very clear, concise, not cheesy, the right amount of eye candy
1305 2013-03-04 19:40:29 <gavinandresen> ok.  Foundation could use the same design firm to do bitcoin.org …
1306 2013-03-04 19:40:42 <gavinandresen> I don't know nuthin about design.
1307 2013-03-04 19:40:50 <HM> you could do what nginx do. have .com for general commercial services/information and .org as a dev portal
1308 2013-03-04 19:41:03 <HM> frankly i think bitcoin.org looks fine
1309 2013-03-04 19:41:08 * gavinandresen wonders who "you" is
1310 2013-03-04 19:41:30 <HM> whoever is in charge
1311 2013-03-04 19:41:37 <gavinandresen> mmmm… yeah....
1312 2013-03-04 19:41:39 rdymac has joined
1313 2013-03-04 19:41:41 <gmaxwell> I'd probably use a somewhat sparser design than that even, but it at least fits my expectations for the sort of thing it should look like.
1314 2013-03-04 19:42:31 <HM> http://www.utorrent.com/ <-- this is basically what most visitors want to see :P
1315 2013-03-04 19:42:31 rdymac has quit (Read error: Connection reset by peer)
1316 2013-03-04 19:42:38 <HM> Big button. Get it now
1317 2013-03-04 19:42:57 <sipa> i wonder whether we need a separate domain specific for the reference client
1318 2013-03-04 19:43:06 zooko has quit (Remote host closed the connection)
1319 2013-03-04 19:43:10 <sipa> and have bitcoin.org be completely client-neutral
1320 2013-03-04 19:43:24 <gavinandresen> i think bitcoin.org should be client-neutral
1321 2013-03-04 19:43:37 <gavinandresen> I bought a SSL cert for bitcoincore.org just the other day.....
1322 2013-03-04 19:43:44 <sipa> oh, cool
1323 2013-03-04 19:44:07 <HM> if you wanted to make sure it was neutral you could actually not use it, just redirect it and hold it as stewards
1324 2013-03-04 19:44:20 <gavinandresen> … mostly for payment protocol testing purposes, but we could fire up a VM/website
1325 2013-03-04 19:44:52 <gavinandresen> (there I go… "we" ….)
1326 2013-03-04 19:45:39 hsy has quit (Read error: Operation timed out)
1327 2013-03-04 19:46:08 <gavinandresen> interesting aside:  I tried to get a free startssl cert for bitcoincore.com, but application was rejected because they don't give free certs for finance-related sites
1328 2013-03-04 19:46:13 <BlueMatt> yet another one to design though...
1329 2013-03-04 19:46:47 TD has joined
1330 2013-03-04 19:47:47 <gavinandresen> BlueMatt: how are we doing on IP addresses and our SnelServer machine?  Have we run out yet?
1331 2013-03-04 19:48:31 <BlueMatt> gavinandresen: theoretically out, but we could fix that if we need another
1332 2013-03-04 19:48:44 <gavinandresen> … or I could just buy another, they're not expensive
1333 2013-03-04 19:48:46 <BlueMatt> (ie nat+port forward for jenkins/dnsseed, saving 2 ips)
1334 2013-03-04 19:48:54 <BlueMatt> well, its not hard, its up to you
1335 2013-03-04 19:49:05 <BlueMatt> actually, could nat for testnet node too
1336 2013-03-04 19:49:09 <BlueMatt> none of the ports collide
1337 2013-03-04 19:49:14 <gavinandresen> ok, I'll buy another.  I might bug you to setup a bitcoincore.org VM if I can't figure out how to do it myself
1338 2013-03-04 19:50:28 i2pRelay has quit (Remote host closed the connection)
1339 2013-03-04 19:50:52 <gavinandresen> … although before I do that… anybody feel like being the sysadmin for a bitcoincore.org ?  I'm imagining it'll host static web page and https binary downloads
1340 2013-03-04 19:51:01 i2pRelay has joined
1341 2013-03-04 19:51:07 <BlueMatt> if you wish, though two ip tables rules + the nat mode for networking  isnt hard (the jekyll build vm already uses nat)
1342 2013-03-04 19:52:27 <gavinandresen> BlueMatt: isn't hard, until you go away and somebody else has to figure out how everything is setup…. (yeah, yeah, we can document how it is done but documentation tends to get neglected / out of date)
1343 2013-03-04 19:52:42 <BlueMatt> alright, well feel free to buy them then
1344 2013-03-04 19:52:44 swappermall has joined
1345 2013-03-04 19:54:54 <gavinandresen> Will do.  Here's a good religious topic for conversation:  apache2 or nginx?  Go!
1346 2013-03-04 19:55:06 <sipa> lighttpd?
1347 2013-03-04 19:55:46 zooko has joined
1348 2013-03-04 19:56:26 <jgarzik> gavinandresen: nginx or lighttpd > apache
1349 2013-03-04 19:56:31 <BlueMatt> lighttpd
1350 2013-03-04 19:56:32 <jgarzik> gavinandresen: even though I am a long time apache user
1351 2013-03-04 19:56:36 <jgarzik> I think the world has moved on
1352 2013-03-04 19:57:26 <gavinandresen> I don't know nuthin about lighttpd…   one of y'all needs to be the bitcoincore sysadmin....
1353 2013-03-04 19:57:38 <Scrat> lighttpd is dead
1354 2013-03-04 19:58:02 <Scrat> use nginx
1355 2013-03-04 19:58:24 <HM> nginx is nice
1356 2013-03-04 19:58:28 rbecker is now known as RBecker
1357 2013-03-04 19:58:43 <gavinandresen> see, I knew this would be a good religious discussion.
1358 2013-03-04 19:58:53 <HM> tbh though, apache is always a good choice
1359 2013-03-04 19:59:24 kriqCoin has joined
1360 2013-03-04 19:59:38 <phantomcircuit> Scrat, long live lighttpd
1361 2013-03-04 19:59:39 * phantomcircuit runs
1362 2013-03-04 20:00:10 <kriqCoin> apache?
1363 2013-03-04 20:00:44 <phantomcircuit> kriqCoin, ahahaahahaaaaa
1364 2013-03-04 20:01:12 <kriqCoin> )))
1365 2013-03-04 20:01:16 <kriqCoin> wasuo phantomcircuit
1366 2013-03-04 20:01:25 <kriqCoin> wasup phantomcircuit
1367 2013-03-04 20:01:35 <phantomcircuit> i have ran out of cookies
1368 2013-03-04 20:01:36 <BlueMatt> IIS
1369 2013-03-04 20:02:21 <sipa> i know! we can build a webserver into bitcoind, so everyone can run the bitcoin.org site locally!
1370 2013-03-04 20:02:41 <BlueMatt> better encode the content in the blockchain
1371 2013-03-04 20:02:52 <sipa> oh, yes, of course
1372 2013-03-04 20:02:57 datagutt has quit (Quit: kthxbai)
1373 2013-03-04 20:02:58 <sipa> sitecoin
1374 2013-03-04 20:03:53 <kriqCoin> power of the swarm
1375 2013-03-04 20:04:06 <kriqCoin> doesnt bittorrent do this already
1376 2013-03-04 20:04:26 <kriqCoin> maybe we should just merge a bitcoin client with a torrent client and be done with it
1377 2013-03-04 20:04:41 <Scrat> just add DHT everywhere
1378 2013-03-04 20:04:46 <sipa> gavinandresen: by the way, what do you still want in 0.8.1? it seems there isn't much movement right now, so i wouldn't be opposed to doing a 0.8.1 rc soon
1379 2013-03-04 20:04:51 <Scrat> toasters, refridgerators, car engines
1380 2013-03-04 20:05:05 <kriqCoin> "bitcoin everywhere"
1381 2013-03-04 20:05:08 <gavinandresen> sipa: fix for the crash-at-shutdown bug would be nice
1382 2013-03-04 20:05:31 <sipa> which one?
1383 2013-03-04 20:05:32 OneMiner has quit (Quit: If your not living on the edge, you're taking up too much space)
1384 2013-03-04 20:06:11 <gavinandresen> sipa: the one that I get on my Mac every once in a while… race condition with threads not shutting down promptly
1385 2013-03-04 20:06:20 <kriqCoin> just to ask about this lighthttpd
1386 2013-03-04 20:06:28 <kriqCoin> its written in C i see
1387 2013-03-04 20:06:40 <kriqCoin> but how do you generate content? php? C? perl?
1388 2013-03-04 20:06:56 <kriqCoin> or it only serves static pages?
1389 2013-03-04 20:07:07 <sipa> kriqCoin: it's a web server, not a content management system
1390 2013-03-04 20:07:40 <sipa> like any other webserver, it serves static pages, CGI, or probably directly from scripts in several languages
1391 2013-03-04 20:07:56 <kriqCoin> you dont understand, all webservers are "content" generators
1392 2013-03-04 20:07:58 <kriqCoin> LOL
1393 2013-03-04 20:08:16 <kriqCoin> "content generators"  not "content management"  the latter is a diff beast
1394 2013-03-04 20:08:46 <sipa> gavinandresen: hmm, is there an issue about it?
1395 2013-03-04 20:09:03 FredEE_ has joined
1396 2013-03-04 20:09:05 <gavinandresen> sipa: I thought so….  but maybe not....
1397 2013-03-04 20:09:24 <sipa> the only issue i personally experienced was fixed late in 0.8.0
1398 2013-03-04 20:11:09 <gavinandresen> bah. I'll open an issue if I see it again, I had a private report of it happening
1399 2013-03-04 20:12:54 <gavinandresen> sipa: https://github.com/bitcoin/bitcoin/issues/2204
1400 2013-03-04 20:13:20 <sipa> ah, that one
1401 2013-03-04 20:13:22 FredEE has quit (Ping timeout: 248 seconds)
1402 2013-03-04 20:13:23 FredEE_ is now known as FredEE
1403 2013-03-04 20:13:58 <gavinandresen> … although I doubt the private report person was running with -server=1 ...
1404 2013-03-04 20:15:29 <sipa> sure it wasn't someone running rc1?
1405 2013-03-04 20:15:37 Lolcust has quit (Quit: Nap time)
1406 2013-03-04 20:15:57 <sipa> using boost::thread for (at least) RPC, and doing a boost::thread::interrupt on shutdown sounds like a solution
1407 2013-03-04 20:16:23 <gavinandresen> sipa:  yes...
1408 2013-03-04 20:16:53 <sipa> "yes, sure it wasn't rc1" or "yes, that will solve it" ?
1409 2013-03-04 20:17:00 <gavinandresen> sipa: my TODO list looks like:  + get payment protocol stuff done enough so wumpus or somebody else can start making the GUI look pretty. And then tackle the thread stuff
1410 2013-03-04 20:17:26 <gavinandresen> not sure it was rc1, and yes, ::thread::interrupt should solve it
1411 2013-03-04 20:18:47 OneMiner has joined
1412 2013-03-04 20:19:23 <gavinandresen> sipa: if you feel inspired to clean up the shutdown code, go for it.
1413 2013-03-04 20:19:32 <sipa> not really :)
1414 2013-03-04 20:19:37 <gavinandresen> awwww
1415 2013-03-04 20:20:43 <sipa> but it seems there will not be many changes still, and i think head has some nice improvements over 0.8.0 already
1416 2013-03-04 20:22:36 <sipa> then again, no real need or important bugfixes
1417 2013-03-04 20:22:59 <gavinandresen> I think we should wait until Sergio has a free weekend again.
1418 2013-03-04 20:23:05 <sipa> haha
1419 2013-03-04 20:23:06 <gavinandresen> :)
1420 2013-03-04 20:23:47 <gmaxwell> I wonder how much cpu usage you get with pessimal bloom filter settings?
1421 2013-03-04 20:23:54 <gmaxwell> (and scanning the chain)
1422 2013-03-04 20:25:26 root2 has quit (Ping timeout: 248 seconds)
1423 2013-03-04 20:25:41 <gmaxwell> What ever happened to getting a birth time in the private key format?
1424 2013-03-04 20:27:13 techlife has quit (Ping timeout: 264 seconds)
1425 2013-03-04 20:28:11 ielo has joined
1426 2013-03-04 20:33:23 <jgarzik> gmaxwell: open pull req
1427 2013-03-04 20:33:35 <jgarzik> at least for bitcoind
1428 2013-03-04 20:33:42 Lolcust has joined
1429 2013-03-04 20:33:46 BenderCoin has quit (Ping timeout: 245 seconds)
1430 2013-03-04 20:34:55 techlife has joined
1431 2013-03-04 20:34:56 techlife has quit (Max SendQ exceeded)
1432 2013-03-04 20:35:32 techlife has joined
1433 2013-03-04 20:35:37 techlife has quit (Max SendQ exceeded)
1434 2013-03-04 20:36:32 techlife has joined
1435 2013-03-04 20:36:33 techlife has quit (Max SendQ exceeded)
1436 2013-03-04 20:36:48 <sipa> i don't like a birth date in the private key itself
1437 2013-03-04 20:37:07 techlife has joined
1438 2013-03-04 20:37:08 techlife has quit (Max SendQ exceeded)
1439 2013-03-04 20:37:41 <sipa> but afaik that's not what the pull does
1440 2013-03-04 20:37:45 techlife has joined
1441 2013-03-04 20:37:46 techlife has quit (Max SendQ exceeded)
1442 2013-03-04 20:38:24 techlife has joined
1443 2013-03-04 20:38:25 techlife has quit (Max SendQ exceeded)
1444 2013-03-04 20:38:49 kriqCoin has quit (Ping timeout: 245 seconds)
1445 2013-03-04 20:39:20 techlife has joined
1446 2013-03-04 20:40:49 <gmaxwell> IIRC jgarzik's pull didn't have them import/export, but I think they should- simply because that is probably the most important usecase for them.
1447 2013-03-04 20:40:54 ProfMac has quit (Ping timeout: 245 seconds)
1448 2013-03-04 20:41:26 <sipa> yes agree
1449 2013-03-04 20:41:45 <sipa> but we should have an export/import wallet function, which includes such metadata
1450 2013-03-04 20:41:49 <jgarzik> https://github.com/bitcoin/bitcoin/pull/1863
1451 2013-03-04 20:41:51 totaleclipse has joined
1452 2013-03-04 20:41:53 <sipa> not stuff it into individual key data
1453 2013-03-04 20:42:01 <jgarzik> "Wallet birthdays": store key create time; calc whole-wallet birthday
1454 2013-03-04 20:43:47 paraipan has joined
1455 2013-03-04 20:44:08 * sipa still wants to find that bug in CNetMessage
1456 2013-03-04 20:44:18 <grau_> I plan to announce botsofproof "beta" with this week. Any objections or offers of review from your side?
1457 2013-03-04 20:44:47 ProfMac has joined
1458 2013-03-04 20:45:15 <sipa> grau_: i don't have time for that now, i'm afraid
1459 2013-03-04 20:46:56 <grau_> sipa: it will be a longer beta phase until about San Jose. Thanks for even considering.
1460 2013-03-04 20:49:53 TD has quit (Quit: TD)
1461 2013-03-04 20:57:08 ovidiusoft has joined
1462 2013-03-04 20:58:33 root2 has joined
1463 2013-03-04 20:58:53 TD has joined
1464 2013-03-04 21:07:07 root2_ has joined
1465 2013-03-04 21:07:28 <grau_> I removed block templater code, so bitsofproof will not be used for mining.
1466 2013-03-04 21:08:25 <gmaxwell> grau_: I'll give it a look this week, or at least if I don't get to it before you announce your beta I'll be sure to look at your beta.
1467 2013-03-04 21:08:54 <grau_> gmaxwell: Thanks I really appretiate it.
1468 2013-03-04 21:09:46 root2 has quit (Ping timeout: 252 seconds)
1469 2013-03-04 21:12:24 grau_ has quit (Remote host closed the connection)
1470 2013-03-04 21:12:31 <BlueMatt> wait, you reimplemented bitcoin in java (again?)
1471 2013-03-04 21:13:01 <BlueMatt> I mean dont get me wrong, node diversity is good, but having 2 full bitcoin implementations in java seems like a lot of uselessly duplicated effort
1472 2013-03-04 21:13:33 totaleclipse has quit (Quit: Leaving)
1473 2013-03-04 21:13:40 clr_ has joined
1474 2013-03-04 21:13:40 clr_ is now known as c00w
1475 2013-03-04 21:14:36 <petertodd> When people talk about BlueMatt's blocktester, where exactly is the code they are referring to?
1476 2013-03-04 21:15:08 <BlueMatt> https://github.com/TheBlueMatt/test-scripts
1477 2013-03-04 21:16:56 OneMiner has quit (Quit: ASCII a stupid question, get a stupid ANSI!)
1478 2013-03-04 21:17:21 ashams has quit (Remote host closed the connection)
1479 2013-03-04 21:17:27 <petertodd> Thanks, I'll add some more LockTime tests to it; I found an interesting off-by-one error re: IsFinal() in CreateNewBlock(), could use some more testing.
1480 2013-03-04 21:18:05 <BlueMatt> oh, fun
1481 2013-03-04 21:18:13 <sipa> i think petertodd is right
1482 2013-03-04 21:18:18 <BlueMatt> also, I think I added one more very small test to it like last week
1483 2013-03-04 21:18:29 <BlueMatt> let me make sure to push it so you can use the latest
1484 2013-03-04 21:19:15 <petertodd> Thanks
1485 2013-03-04 21:19:50 <sipa> ok, after seeding bootstrap.dat for a few days: uploaded 270 MB
1486 2013-03-04 21:19:52 <sipa> ... wow
1487 2013-03-04 21:20:11 <BlueMatt> does it need another seedbox?
1488 2013-03-04 21:20:12 <petertodd> I also gotta ask, how the heck do you run the thing? Do you do it in a VM setup for the task? I noticed a lot of hard-coded /home/ubuntu and what not.
1489 2013-03-04 21:20:13 <sipa> there must be much more seeders than downloaders
1490 2013-03-04 21:21:09 <BlueMatt> petertodd: the jenkins vm
1491 2013-03-04 21:21:16 <BlueMatt> (thats on the devteam dedicated server)
1492 2013-03-04 21:21:29 <BlueMatt> also runs pull-tester
1493 2013-03-04 21:21:29 hsy has joined
1494 2013-03-04 21:21:36 <petertodd> Ah, yeah, I don't have access to that.
1495 2013-03-04 21:21:46 <BlueMatt> petertodd: grab the copy at https://code.google.com/r/bluemattme-bitcoinj/source/detail?r=60ba9bf80d0222e062d2852a9399c180e58d17b5&name=newscripts
1496 2013-03-04 21:22:55 <BlueMatt> if you need changes to it feel free to write up the necessary code and I can do them, just think generic ubuntu-server on same dist as gitian uses
1497 2013-03-04 21:23:39 <petertodd> Cool, I'll give that a try.
1498 2013-03-04 21:23:58 <BlueMatt> if the script works in gitian, chances are it will work on jenkins
1499 2013-03-04 21:23:59 <petertodd> (I'll write up a README as I figure it out for doc purposes)
1500 2013-03-04 21:24:08 <BlueMatt> sure, thanks
1501 2013-03-04 21:24:32 mappum has joined
1502 2013-03-04 21:24:37 OneMiner has joined
1503 2013-03-04 21:25:01 Mad7Scientist has quit (Ping timeout: 255 seconds)
1504 2013-03-04 21:26:02 <gmaxwell> grau_: I'm looking that your reimplemention of the bitcoinj block tester, and I don't see how blocks that must not be accepted are annotated in the json.
1505 2013-03-04 21:27:14 <gmaxwell> I also see that you have the b48 and b64 test cases disabled? Did you ever talk to bluematt about your belief that the tests were wrong?
1506 2013-03-04 21:27:14 i2pRelay has quit (Remote host closed the connection)
1507 2013-03-04 21:27:33 Mad7Scientist has joined
1508 2013-03-04 21:32:57 ovidiusoft has quit (Read error: No route to host)
1509 2013-03-04 21:34:35 ovidiusoft has joined
1510 2013-03-04 21:35:22 i2pRelay has joined
1511 2013-03-04 21:38:15 guruvan has joined
1512 2013-03-04 21:39:16 whizter has quit ()
1513 2013-03-04 21:46:55 LargoG has quit (Remote host closed the connection)
1514 2013-03-04 21:50:18 LargoG has joined
1515 2013-03-04 21:56:45 <gmaxwell> grau_: Do you remember block failures?   E.g. if I send you an inefficiently seralized block that you'll reject because you test the actual seralization against the limit instead of a reseralization  when someone else gives you exactly the same block but efficiently seralized, will you accept it?
1516 2013-03-04 21:57:14 RBecker is now known as rbecker
1517 2013-03-04 21:58:30 <gmaxwell> Your rejection of b64 is borderline on a network forking rule, except bitcoin nodes fix the seralization while forwarding.
1518 2013-03-04 21:58:35 <gmaxwell> uh, I don't quite understand this:
1519 2013-03-04 21:58:35 <gmaxwell> BitcoinPeer.java:       private static final int MAX_BLOCK_SIZE = 1000000 + 80;
1520 2013-03-04 21:58:51 <petertodd> gmaxwell: ...
1521 2013-03-04 21:59:16 BenderCoin has joined
1522 2013-03-04 21:59:34 <Scrat> massive ORPHAN BLOCK spam in my debug.log, sup with that
1523 2013-03-04 21:59:39 <Scrat> like thousands of lines
1524 2013-03-04 21:59:52 <gmaxwell> Scrat: happens during initial sync when the network finds a block before you're done pulling.
1525 2013-03-04 22:02:01 B0g4r7 has quit (Ping timeout: 276 seconds)
1526 2013-03-04 22:03:10 <gmaxwell> petertodd: so he's only testing the incoming message. So the +80 includes the header.
1527 2013-03-04 22:04:52 <gmaxwell> means that if nodes relay an inefficiently seralized block bitcoin-qt will accept (since it resealizes before testing) while BOP will ban the peer.
1528 2013-03-04 22:05:14 HM has quit (Read error: Connection reset by peer)
1529 2013-03-04 22:05:21 <Scrat> gmaxwell: and it stopped syncing for a good 5 minutes
1530 2013-03-04 22:05:53 <sipa> Scrat: also, expected, unfortunately
1531 2013-03-04 22:06:07 <sipa> Scrat: the block sync mechanism gets confused when a block is announced while it's downloading
1532 2013-03-04 22:06:17 <sipa> it'll continue when there's a new block
1533 2013-03-04 22:06:24 MobGod has quit (Ping timeout: 245 seconds)
1534 2013-03-04 22:06:53 <petertodd> gmaxwell: Interesting! Yes, that sounds like a subtle fork risk for sure, although as you say, probably not a hardfork.
1535 2013-03-04 22:07:48 <gmaxwell> I .. think... it's okay so long as no relayer becomes popular that fails to reseralize the encoding.
1536 2013-03-04 22:08:11 <gmaxwell> If one did, BOP nodes would instantly partition themselves as soon as someone sent a boundary block with an inefficient seralization.
1537 2013-03-04 22:08:11 <Scrat> sipa: yeah I've seen you mention it a couple of times, never seen that happen before (maybe I wasn't looking at current block like a hawk before)
1538 2013-03-04 22:08:22 HM has joined
1539 2013-03-04 22:08:31 owowo has quit (Ping timeout: 276 seconds)
1540 2013-03-04 22:09:05 <petertodd> Yeah, I'm sure I've mentioned this before, but I really like the idea of adding more block and transaction distribution mechanisms as a way to avoid such propagation-based forks.
1541 2013-03-04 22:09:19 <petertodd> Just maybe not satelite at first. :)
1542 2013-03-04 22:10:06 <petertodd> re: IsFinal() IsStandard(), it might be worthwhile to measure propagation of transactions affected by that, to get an idea of what the thresholds are.
1543 2013-03-04 22:10:14 B0g4r7 has joined
1544 2013-03-04 22:10:15 <petertodd> (as nodes upgrade to 0.8)
1545 2013-03-04 22:12:20 zooko has quit (Ping timeout: 260 seconds)
1546 2013-03-04 22:12:47 owowo has joined
1547 2013-03-04 22:13:36 zooko has joined
1548 2013-03-04 22:14:29 bitafterbit has quit (Read error: Connection reset by peer)
1549 2013-03-04 22:16:25 stalled has quit (Ping timeout: 264 seconds)
1550 2013-03-04 22:17:15 root2_ is now known as root2
1551 2013-03-04 22:18:10 <BlueMatt> petertodd: awww, but sattilites are fun!
1552 2013-03-04 22:18:23 <BlueMatt> (also, I wish I had the time to get my ham license to trade blocks w/ gmaxwell)
1553 2013-03-04 22:18:48 zooko has quit (Ping timeout: 245 seconds)
1554 2013-03-04 22:18:50 TD has quit (Quit: TD)
1555 2013-03-04 22:18:55 owowo has quit (Ping timeout: 276 seconds)
1556 2013-03-04 22:19:00 <BlueMatt> petertodd: on a related note, Im doing some research on block propagation across the p2p net atm
1557 2013-03-04 22:20:13 <D34TH> that would be interesting
1558 2013-03-04 22:20:18 <D34TH> p2p over ham
1559 2013-03-04 22:20:30 inlikeflynn has quit (Ping timeout: 246 seconds)
1560 2013-03-04 22:20:37 <petertodd> BlueMatt: Ha, that's an excellent excuse to get a license. Also, because Bitcoin is not encrypted, just signed, it wouldn't run afoul of the no-encryption rules on the ham bands.
1561 2013-03-04 22:20:54 inlikeflynn has joined
1562 2013-03-04 22:20:57 <petertodd> BlueMatt: Oh cool, I better not duplicate your work then.
1563 2013-03-04 22:22:08 <gmaxwell> petertodd: may run afoul of commercial use restrictions. Though generally, at least in emergency situations relaying third party commercial messages is acceptable under FCC regs at least. International is subject to more complicated regs though.
1564 2013-03-04 22:22:56 <BlueMatt> gmaxwell: I would not call blocks commercial, they are public data, even though they facilitate commercial stuff
1565 2013-03-04 22:23:24 <BlueMatt> petertodd: well free feel to help, Ive only half-stared (planning on taking measurements and running simulations on university-provided real-world networks)
1566 2013-03-04 22:23:29 <petertodd> Hmm... that's complex... well, just relaying headers might help re: regs.
1567 2013-03-04 22:23:54 <gmaxwell> Thats mostly how the US regs are oriented, they're mostly concerned with the operator themselves having a pecuniary interest in the transmission (e.g. don't use ham instead of buying your business a proper commercial radio license)
1568 2013-03-04 22:24:16 <petertodd> BlueMatt: Something I do want to do, is make up a network *simulator*, that accurately models bandwidth and propagation delays; mainly I want to test my theories on perverse incentives re: large blocks.
1569 2013-03-04 22:24:46 <petertodd> BlueMatt: Helps too that my thesis for my fine arts degree happened to be a python project that I could re-use a lot of the code from...
1570 2013-03-04 22:25:23 owowo has joined
1571 2013-03-04 22:25:50 <gmaxwell> Did you attempt to recreate fine art in the shirt colors of flash mobs by leaving around ipods with different instructions on them?
1572 2013-03-04 22:28:02 <Acciaio> how can I stop my client from retrasmitting unconfirmable transactions?
1573 2013-03-04 22:28:38 <gmaxwell> How can I stop Acciaio from retransmitting an answered question?
1574 2013-03-04 22:29:06 inlikeflynn has quit (Read error: Connection reset by peer)
1575 2013-03-04 22:30:51 chmod755 has quit (Quit: chmod755)
1576 2013-03-04 22:31:32 ovidiusoft has quit (Ping timeout: 252 seconds)
1577 2013-03-04 22:31:59 <BlueMatt> petertodd: Im working on doing a simulator that uses geni to run "nodes" that just delay messages instead of processing on a large-scale network-study network
1578 2013-03-04 22:33:55 inlikeflynn has joined
1579 2013-03-04 22:40:47 stalled has joined
1580 2013-03-04 22:41:16 sgornick has quit (Ping timeout: 260 seconds)
1581 2013-03-04 22:45:57 guruvan has quit ()
1582 2013-03-04 22:45:59 sgornick has joined
1583 2013-03-04 22:48:59 <freewil> you know what might be nice is if bitcoind has a reserved slot option
1584 2013-03-04 22:49:12 <BlueMatt> reserved for?
1585 2013-03-04 22:49:26 <freewil> a client you always want to be able to connect
1586 2013-03-04 22:49:29 <freewil> regardless of maxconnections
1587 2013-03-04 22:49:31 <BlueMatt> done
1588 2013-03-04 22:49:34 <BlueMatt> (in 0.8)
1589 2013-03-04 22:49:39 <freewil> oh nice
1590 2013-03-04 22:49:50 <freewil> havent upgraded yet
1591 2013-03-04 22:50:51 owowo has quit (Remote host closed the connection)
1592 2013-03-04 22:51:25 agricocb has quit (Quit: Leaving.)
1593 2013-03-04 22:51:27 owowo has joined
1594 2013-03-04 22:59:52 Acciaio has quit (Ping timeout: 276 seconds)
1595 2013-03-04 23:00:34 owowo has quit (Remote host closed the connection)
1596 2013-03-04 23:01:08 owowo has joined
1597 2013-03-04 23:01:10 <kritCoin> http://bitcoinfpga.com/
1598 2013-03-04 23:01:14 <kritCoin> look sgood
1599 2013-03-04 23:01:17 <petertodd> gmaxwell: Ha, basically it was a transportation network simulator.
1600 2013-03-04 23:01:44 <petertodd> BlueMatt: Ah, so it'll be real software then right?
1601 2013-03-04 23:01:55 <BlueMatt> petertodd: umm, yes?
1602 2013-03-04 23:02:04 <petertodd> BlueMatt: I was planning on doing it abstractly, with a big Python model.
1603 2013-03-04 23:02:04 <BlueMatt> "real" software?
1604 2013-03-04 23:02:10 <BlueMatt> oh
1605 2013-03-04 23:02:18 <BlueMatt> not it uses the real internet
1606 2013-03-04 23:02:19 <petertodd> BlueMatt: geni lets you simulate networks right? With real software?
1607 2013-03-04 23:02:23 <BlueMatt> yes
1608 2013-03-04 23:03:14 <petertodd> Oh well, sounds like our two approaches don't overlap, so worthwhile to do both.
1609 2013-03-04 23:03:34 <BlueMatt> they dont? its the same thing just different approaches to it
1610 2013-03-04 23:04:09 <BlueMatt> anyway, the real work in both is coming up with realistic values for latency across individual nodes, which is the same thing
1611 2013-03-04 23:04:17 <petertodd> BlueMatt: Yes, but my way will give faster prototyping, at the expense of accuracy, and with you the opposite.
1612 2013-03-04 23:04:34 <petertodd> Freenet did the same thing way back when with network simulators to experiment with algorithms.
1613 2013-03-04 23:04:44 <petertodd> Of course, I also want to generate pretty videos from it...
1614 2013-03-04 23:04:50 <BlueMatt> heh
1615 2013-03-04 23:06:18 c00w has quit (Ping timeout: 245 seconds)
1616 2013-03-04 23:06:44 slush has quit (Ping timeout: 276 seconds)
1617 2013-03-04 23:07:12 <BlueMatt> petertodd: are you planning on trying to test for real-world latencies across individual nodes, or just punt and guess?
1618 2013-03-04 23:08:44 <petertodd> BlueMatt: A bit of both - remember that I don't just want to simulate the network as it is, but simulate what it could be. In particular, I want to see what increasing the block size would do.
1619 2013-03-04 23:09:10 <BlueMatt> yes, as do I, I just wondered where you will be getting your seed values for latency
1620 2013-03-04 23:10:11 <petertodd> Initially, I think I'll just assume that the network behaves ideally, and that nodes have fast enough CPU and disks to keep up, so latency will just be bandwidth and speed of light.
1621 2013-03-04 23:10:33 <petertodd> Main thing is to make the code pluggable, so a better model can be added later.
1622 2013-03-04 23:10:57 <petertodd> I'm more interested in scaling behavior, rather than absolute performance.
1623 2013-03-04 23:11:08 <BlueMatt> yes, ofc
1624 2013-03-04 23:11:28 <BlueMatt> ahh, ok, well, Im sure Ill be behind you then...Ill see if I can give you more real-world data as I go
1625 2013-03-04 23:11:33 <BlueMatt> anyway, have to go
1626 2013-03-04 23:11:43 <petertodd> Later
1627 2013-03-04 23:12:58 <petertodd> Oh, and I will say, of course, I think one advantage of my approach with a higher level simulator, is given I've been arguing against large blocks, it's much better if others can easily run the simulator and decide for themselves if it's realistic. Not so easy with something like gini where gettng access is hard.
1628 2013-03-04 23:13:02 agricocb has joined
1629 2013-03-04 23:16:24 brwyatt is now known as Away!~brwyatt@brwyatt.net|brwyatt
1630 2013-03-04 23:16:30 slush has joined
1631 2013-03-04 23:18:56 one_zero has joined
1632 2013-03-04 23:22:17 denisx has joined
1633 2013-03-04 23:30:25 owowo has quit (Quit: sayonara)
1634 2013-03-04 23:35:15 denisx has quit (Read error: Operation timed out)
1635 2013-03-04 23:36:03 denisx has joined
1636 2013-03-04 23:41:43 kritCoin has quit (Quit: ChatZilla 0.9.90 [Firefox 20.0/20130227063501])
1637 2013-03-04 23:42:50 owowo has joined
1638 2013-03-04 23:43:29 PhantomSpark has joined
1639 2013-03-04 23:46:42 coblee_ has joined
1640 2013-03-04 23:47:05 CodeShark has joined
1641 2013-03-04 23:49:49 coblee has quit (Ping timeout: 250 seconds)
1642 2013-03-04 23:49:50 coblee_ is now known as coblee