1 2011-10-17 00:13:04 <theymos> I don't see where there is an off-by-1 problem with difficulty adjustments. I see adjustments over 145152-147167 and 147168-149183 inclusive. These ranges are 2015 blocks in size, but there isn't a block left out of the range.
  2 2011-10-17 00:14:28 <gmaxwell> theymos: ... yea it's an off by one.
  3 2011-10-17 00:14:44 <theymos> Which block is being left out?
  4 2011-10-17 00:14:47 <gmaxwell> theymos: consider. You lie in 147167 and say it happened two hours late.
  5 2011-10-17 00:15:06 <theymos> 147167 is included in the calculation as far as I can tell.
  6 2011-10-17 00:15:10 <gmaxwell> Then the 147168-149183 doesn't compensate for the difficulty offset that created.
  7 2011-10-17 00:16:46 <gmaxwell> So, say you split the chain at the point it was three weeks ago. At the end of the first difficulty window you say the block happened now+2hr (three weeks after when it properly should have). You'll end up with 1/4 the difficulty being computed for the next cycle.
  8 2011-10-17 00:17:32 <gmaxwell> And you never have to _pay_ for the fudged timestamp because the one after the window will be correctly timed (the minimum uses the median of 11, so you never get forced to use a future time anywhere else)
  9 2011-10-17 00:19:08 <gmaxwell> theymos: did I make that make sense?
 10 2011-10-17 00:19:48 <gmaxwell> instead of thinking about what blocks are included in the calculation, think about what GAPS are included, and you'll see we're leaving out a gap.
 11 2011-10-17 00:25:17 <odysseus654> i'm guessing that with something like bitcoin it really isn't that easy to change this kind of rule...
 12 2011-10-17 00:26:06 <theymos> gmaxwell: I see how that is an attack. What is it that's "off by one", though?
 13 2011-10-17 00:27:05 <gmaxwell> theymos: if the lower bound went one lower it would capture the gap which is being missed.  I don't really know if it was a coding off by one or just a design flaw though. I can't see into the developer's mind. ;)
 14 2011-10-17 00:30:12 <CIA-101> bitcoin: Con Kolivas * r09f6736ab00b cgminer/main.c: Style cleanups.
 15 2011-10-17 00:30:14 <CIA-101> bitcoin: Con Kolivas * r6d004ddfadc1 cgminer/main.c: Write unix configuration to .cgminer/cgminer.conf by default and prompt to overwrite if given a filename from the menu that exists.
 16 2011-10-17 00:32:53 <theymos> If the lower edge went one lower, wouldn't the broken time value just make the next retarget off as well? The time value would still be wrong.
 17 2011-10-17 00:34:58 <gmaxwell> It would make it off in the opposite direction.
 18 2011-10-17 00:35:03 <gmaxwell> So the average of the two is correct.
 19 2011-10-17 00:35:11 <gmaxwell> And you gain nothing much from your little game.
 20 2011-10-17 00:35:58 <gmaxwell> As it is now you can do this over and over again and get whatever difficulty you want without driving the timestamps into the far future.
 21 2011-10-17 00:36:17 <theymos> OK, thanks for explaining it. Looks like all 2016 blocks should have their time data used in the difficulty calculation.
 22 2011-10-17 00:36:39 <gmaxwell> So I can fork the chain, mine it down to difficulty one, and if I have >50% mine the ever lasting shit out of it until my sum diff beats the network ... release it and cause a big split and massive inflation.
 23 2011-10-17 01:00:14 <CIA-101> bitcoin: Con Kolivas * r9a0b7096230e cgminer/ (README example-cfg.json example.conf main.c): Update documentation about new configuration file with an example file.
 24 2011-10-17 01:10:21 <CIA-101> bitcoin: Con Kolivas * r2a19153d90db cgminer/Makefile.am: Add example.conf to makefile dist.
 25 2011-10-17 01:30:16 <CIA-101> bitcoin: Con Kolivas * r2ac11bd7d6c6 cgminer/NEWS: Update NEWS.
 26 2011-10-17 01:30:17 <CIA-101> bitcoin: Con Kolivas * r1acdad89ab6d cgminer/configure.ac: Bump version number to 2.0.7.
 27 2011-10-17 06:58:38 <ThomasV> who's running blockchain.info ? why is it down ?
 28 2011-10-17 07:00:15 <upb> because the coin is dieing
 29 2011-10-17 07:02:32 <ThomasV> dieing? like in a diet ?
 30 2011-10-17 07:06:36 <erus`> bitcoin is over :(
 31 2011-10-17 07:06:53 <Ycros> love is over
 32 2011-10-17 07:07:49 <erus`> pack it up guys, lets go home
 33 2011-10-17 07:15:53 <SomeoneWeird> lol
 34 2011-10-17 07:19:26 <noagendamarket> pretty much
 35 2011-10-17 07:19:38 <Blitzboom> hahah
 36 2011-10-17 07:38:51 <edcba> ;;bc,mtgox
 37 2011-10-17 07:38:52 <gribble> {"ticker":{"high":3.8975,"low":2.77002,"avg":3.418140592,"vwap":3.376363487,"vol":104634,"last_all":2.91,"last_local":2.91,"last":2.91,"buy":2.91,"sell":2.93358}}
 38 2011-10-17 07:45:54 <ThomasV> interactive graph of bitcoins moving over time: http://ecdsa.org/stats.html
 39 2011-10-17 07:46:42 <ThomasV> let me know it it's too slow to download
 40 2011-10-17 08:09:33 <Eliel> ThomasV: would be nice to have some explanation on what the different colors represent. I guess the green is the number of coins moved per block but I have no idea what the bluishgreen color represents.
 41 2011-10-17 08:10:16 <ThomasV> Eliel: see here http://www.youtube.com/watch?v=9q3UtmhR9G8
 42 2011-10-17 08:11:03 <UukGoblin> ThomasV, This Websense category is filtered: Potentially Damaging Content.
 43 2011-10-17 08:11:32 <ThomasV> UukGoblin: which one?
 44 2011-10-17 08:12:07 <UukGoblin> ThomasV, http://ecdsa.org/stats.html
 45 2011-10-17 08:12:19 <UukGoblin> ThomasV, don't worry, I think the bastards block all domains they haven't seen before
 46 2011-10-17 08:12:45 <ThomasV> UukGoblin: why are you using them?
 47 2011-10-17 08:12:51 <UukGoblin> ThomasV, I'm not, my company is
 48 2011-10-17 08:13:03 <UukGoblin> (I'm in an office)
 49 2011-10-17 08:14:03 <Eliel> ThomasV: ah, this is a great graph, just add some explanation to go with it on the page and it's perfect :)
 50 2011-10-17 08:14:23 <Eliel> interesting how the early adopter coins from 2009 have barely moved at all
 51 2011-10-17 08:15:06 <ThomasV> I am creating a blog entry with an explanation
 52 2011-10-17 08:16:03 <UukGoblin> satoshi wanting to stay anonymous?
 53 2011-10-17 08:42:29 <Eliel> ThomasV: one suggestion for the visualization. Could you modify the graph to be logarithmic? that way, even the big spikes would fit into the graph.
 54 2011-10-17 08:42:49 <ThomasV> Eliel: yes, I thought about that.
 55 2011-10-17 08:43:11 <ThomasV> otoh, we should expect it to become less and less spiky over time
 56 2011-10-17 08:43:28 <Eliel> another idea I had was to leak the too big spikes into the surrounding days but that might not work too well.
 57 2011-10-17 08:43:30 <ThomasV> (in a few years :-)
 58 2011-10-17 08:43:57 <ThomasV> well, it's possible to use larger time bins
 59 2011-10-17 08:44:07 <ThomasV> that's the case in the video
 60 2011-10-17 08:45:09 <Eliel> yes, daily accuracy is perhaps a bit too much :)
 61 2011-10-17 08:45:40 <Eliel> at least if you want to show the whole 2 and half years in the graph.
 62 2011-10-17 08:45:52 <ThomasV> but it would not remove all the spikes
 63 2011-10-17 08:46:10 <Eliel> just reducing them is good enough :)
 64 2011-10-17 08:48:44 <ThomasV> the problem with a logscale is that it does not make sense;  all coins are equal
 65 2011-10-17 08:49:24 <Eliel> maybe you could give users options on how to show the data? a few widgets on the side or below.
 66 2011-10-17 08:49:59 <ThomasV> heh, these are pngs, but sure we could do that
 67 2011-10-17 08:59:10 <Eliel> oh, pngs :D I didn't notice at all :)
 68 2011-10-17 11:13:38 <AlexWaters> anyone know of any updates on the binaries that Matt and Wladimir are working on?
 69 2011-10-17 11:19:40 <gavinandresen> AlexWaters: I haven't heard...
 70 2011-10-17 11:27:50 <_side> Hello people, can anyone help me? I've installed pushpool on my test server, and run client. And every time the client send getwork, server answer a different data-block. BC-wiki says that new block generates every 10-15 minutes, but even if I send "getwork" every second I got different data-block. Please explain me this situation..
 71 2011-10-17 11:28:44 <AlexWaters> Luke built some here: https://bitcointalk.org/index.php?topic=47586.msg573675#msg573675 - anyone know how he did it so that we can have Jenkins churn .exes out?
 72 2011-10-17 11:29:04 <mologie> _side: I am not very familiar with how bitcoin works, but what I can certainly tell you is that work is not related to the current block
 73 2011-10-17 11:29:29 <gavinandresen> _side: yes, every miner works on a slightly different variation of a block, so they don't duplicate work
 74 2011-10-17 11:29:31 <mologie> It is normal that you get a different result each time you query the server.
 75 2011-10-17 11:32:25 <_side> hmm, it looks total different, but first bytes is the same, looks like header
 76 2011-10-17 11:34:40 <_side> I thought that all clients got same block and hash it with different nonces..
 77 2011-10-17 11:40:01 <sipa> _side: the generation transaction (which pays to yourself) is different for everyone, and contains a nonce as well, and this influences the maerkle root (which forms a significant part of the block header)
 78 2011-10-17 11:49:17 <_side> don't completely understand) will try to find some documentation of this
 79 2011-10-17 11:51:23 <_side> "the generation transaction" is the block which pool's client got, or something more?
 80 2011-10-17 11:57:04 <sipa> a transaction is not a block
 81 2011-10-17 11:57:14 <sipa> a block is a collection of transactions
 82 2011-10-17 11:59:01 <_side> %)
 83 2011-10-17 11:59:57 <_side> when my minig clent got block from pool, it contain more than one transaction?
 84 2011-10-17 12:01:04 <UukGoblin> _side, each block has a generation transaction (the one that pays out the 50 BTC bounty) and user transactions.
 85 2011-10-17 12:01:34 <UukGoblin> _side, "normal" blocks pay this 50 BTC to a single address. With pools like eligius, they instead pay it out proportionally (per share, actually) to multiple addresses.
 86 2011-10-17 12:04:01 <_side> Understand this, thanks) But I can't understand why pool send any client different data and midstate, how this help to distribute work between clients ?
 87 2011-10-17 12:04:56 <sipa> _side: miners try to find blocks whose hash is low enough, but they all try different combinations
 88 2011-10-17 12:05:08 <sipa> if you request new work, you get a new combination
 89 2011-10-17 12:05:25 <sipa> such that you can give it to someone else, and be sure no one tries to search the same range
 90 2011-10-17 12:10:06 <_side> different combination of what? I thought basically pool has a block like 12345 0000 6789, and all clients try to hash 12345 0010 6789, to find hash that lower thaB target. But I see that client got totally different data, that  not begins from 12345
 91 2011-10-17 12:11:00 <_side> or nonce is greater than one dword ?
 92 2011-10-17 12:11:51 <UukGoblin> _side, only some of the block is dictated by the network. The rest (like the generation transaction) are up to the pool operator. There's some random data that can be modified
 93 2011-10-17 12:14:01 <sipa> the generation transaction can contain up to 800 bits of nonce data
 94 2011-10-17 12:15:30 <_side> this is the key)) When I write simple miner, I see that in pool nonce is 1 dword
 95 2011-10-17 12:15:59 <sipa> yes, the block hash is 32 bits
 96 2011-10-17 12:16:22 <sipa> but the most important field of the block hash is the 256-bit merkle root
 97 2011-10-17 12:16:36 <sipa> this merkle root is calculated from the transaction hashes
 98 2011-10-17 12:17:17 <sipa> and the first transaction is the generation (it's a special transaction, used to introduce new coins into circulation, and forms the basis for mining income - together with fees)
 99 2011-10-17 12:17:29 <sipa> this generation transaction itself has another nonce, of up to 800 bits
100 2011-10-17 12:23:06 <_side> thak you, and all other =)
101 2011-10-17 12:24:29 <_side> where does you know it all from ? from en.bitcoin.it/wiki/  or another full specification?)
102 2011-10-17 12:24:46 <UukGoblin> wiki, source code, IRC
103 2011-10-17 12:24:47 <UukGoblin> ;-]
104 2011-10-17 12:25:25 <_side> thanx =)
105 2011-10-17 12:51:45 <gavinandresen> Anybody know how small an ECDSA signature can possibly be?
106 2011-10-17 12:51:55 <gavinandresen> (number of bytes)
107 2011-10-17 12:53:14 <luke-jr> compact is 65
108 2011-10-17 13:02:21 <sipa> gavinandresen: 2*N bits for N-bit EC
109 2011-10-17 13:02:53 <sipa> so 64 bytes for secp256k1; the compact format adds one byte for the 2 hint bits
110 2011-10-17 13:03:13 <gavinandresen> sipa: thanks.
111 2011-10-17 13:28:36 <terrytibbs> sipa: What do you use to generate the graphs at bitcoin.sipa.be?
112 2011-10-17 13:28:56 <sipa> terrytibbs: gnuplot
113 2011-10-17 13:29:40 <terrytibbs> I knew they looked familiar!
114 2011-10-17 13:29:41 <terrytibbs> Thanks
115 2011-10-17 13:39:28 <ThomasV> is john tobey sometimes on this channel ?
116 2011-10-17 13:55:01 <Ukto> hmm
117 2011-10-17 13:55:10 <Ukto> wish there was a checkaddress command
118 2011-10-17 13:55:16 <Ukto> that would simply return if the address is local within the wallet
119 2011-10-17 13:55:17 <Ukto> or not
120 2011-10-17 13:55:33 <gavinandresen> Ukto: validateaddress
121 2011-10-17 13:55:41 <BlueMatt> ismine isnt it?
122 2011-10-17 13:55:45 <gavinandresen> yup
123 2011-10-17 13:56:03 <Ukto> hm
124 2011-10-17 13:56:09 <Ukto> really, did i really miss that
125 2011-10-17 13:56:12 <Ukto> i know i had seen that command before
126 2011-10-17 13:56:21 <Ukto> mebbe it just wasnt listing in the help command :/
127 2011-10-17 13:57:24 <BlueMatt> gavinandresen: doc/build-*.txt and doc/readme-qt.rst have become the same file with duplicate and (sometimes) conflicting info...
128 2011-10-17 13:57:59 <Ukto> wow, im bad.
129 2011-10-17 13:59:52 <gavinandresen> BlueMatt: yup... if I knew anything about building on windows or all the different flavors of linux I'd suggest fixes.
130 2011-10-17 14:00:26 <luke-jr> so who here is buying the BFLab thing?
131 2011-10-17 14:00:51 <gavinandresen> B-flab ?  is that some kind of "equal rights for fat people" organization?
132 2011-10-17 14:01:05 <luke-jr> &
133 2011-10-17 14:01:07 <BlueMatt> gavinandresen: heh, thats true...anyway someone needs to do it eventually...
134 2011-10-17 14:01:12 <luke-jr> no, they have 1 GH/s at 20 W for $500 preorder
135 2011-10-17 14:01:16 <BlueMatt> wtf?
136 2011-10-17 14:01:18 <luke-jr> FPGA
137 2011-10-17 14:01:21 <BlueMatt> mmm
138 2011-10-17 14:01:29 <sipa> nice
139 2011-10-17 14:01:43 <luke-jr> I figure it breaks even at least $2 under everyone else
140 2011-10-17 14:01:50 <luke-jr> so it should pay for itself in a year
141 2011-10-17 14:01:51 <gavinandresen> I still want an FPGA space heater for my office.
142 2011-10-17 14:01:58 <BlueMatt> link?
143 2011-10-17 14:02:03 <luke-jr> gavinandresen: well, apparently these are low-heat :p
144 2011-10-17 14:02:12 <copumpkin> that's why you need a thousand of them
145 2011-10-17 14:02:15 <copumpkin> then you get nice heat
146 2011-10-17 14:02:20 <luke-jr> lol
147 2011-10-17 14:02:24 <luke-jr> only 100 available for preorder ;)
148 2011-10-17 14:02:25 <copumpkin> most expensive space heater evar
149 2011-10-17 14:02:37 <luke-jr> copumpkin: or profitable, once it breaks evne
150 2011-10-17 14:02:40 <gavinandresen> not if it'll pay for itself...  I don't want to do no soldering, though
151 2011-10-17 14:02:44 <copumpkin> well, I guess the LHC could be considered a space heater
152 2011-10-17 14:02:54 <copumpkin> so maybe not the most expensive one
153 2011-10-17 14:03:04 <gmaxwell> Well, the butterfly laps people are claiming they plan to do a 54GH box... based on their numbers it should put out a kw of heat.. should keep your feet nice and tosty.
154 2011-10-17 14:04:49 <BlueMatt> heh, best space heater ever -> the LHC
155 2011-10-17 14:04:49 <gmaxwell> It would be nice if some well known bitcoiner unrelated to them would pay them a visit and tell everyone else if it looks real.
156 2011-10-17 14:04:49 <luke-jr> http://butterflylabs.com/product-details/
157 2011-10-17 14:04:49 <rjk2> hm that looks cool for 1 gh/s
158 2011-10-17 14:05:08 <rjk2> gmaxwell: yes
159 2011-10-17 14:05:38 <rjk2> but for $500, i can sell my existing rig and profit right away
160 2011-10-17 14:05:54 <rjk2> cause i only am making about 1.2 gh/s right now
161 2011-10-17 14:06:10 <luke-jr> also, they said they'll probably make an announcement this week
162 2011-10-17 14:06:24 <copumpkin> do they have a physical location?
163 2011-10-17 14:06:38 <luke-jr> copumpkin: they said I was welcome to pick up mine at their offices in Kansas City
164 2011-10-17 14:06:42 <copumpkin> hah
165 2011-10-17 14:06:46 <gavinandresen> http://butterflylabs.com/contact/
166 2011-10-17 14:06:47 <rjk2> Address: 25E 12th Street
167 2011-10-17 14:06:48 <copumpkin> okay, I won't be visiting anytime soon
168 2011-10-17 14:06:59 <copumpkin> unless a tornado comes along
169 2011-10-17 14:07:03 <copumpkin> and blows me to kansas
170 2011-10-17 14:07:09 <Ukto> theres no place like home...
171 2011-10-17 14:07:10 <luke-jr> I know lots of people in KC
172 2011-10-17 14:09:05 <gmaxwell> luke-jr: then send someone to check it out.
173 2011-10-17 14:09:05 <Ukto> wonder how well scrypt would run on it :P
174 2011-10-17 14:09:36 <gmaxwell> Ukto: depends on how much memory you're using with scrypt.
175 2011-10-17 14:09:48 <luke-jr> Ukto: I've decided scrypt is a bad idea
176 2011-10-17 14:10:10 <BlueMatt> luke-jr: werent you one of the people heavily pushing scrypt for bitcoin wallet encryption?
177 2011-10-17 14:10:20 <luke-jr> BlueMattno
178 2011-10-17 14:10:30 <luke-jr> I might also be confusing scrypt with something else
179 2011-10-17 14:10:44 <gmaxwell> BlueMatt: I was. And it would be a good thing for that.
180 2011-10-17 14:10:44 <luke-jr> isn't it the hash algo used for tenebrix and those other "CPU mining only" scamcoins?
181 2011-10-17 14:10:47 <BlueMatt> memory-heavy hashing algo...
182 2011-10-17 14:11:01 <gmaxwell> But yes, I've long said it was a bad idea for hash chain POW.
183 2011-10-17 14:11:17 <sipa> scrypt for POW = botnets
184 2011-10-17 14:11:21 <gmaxwell> 'cause it just means handing over the chain to the biggest botnet hardly an improvement.
185 2011-10-17 14:11:23 <copumpkin> prisoner of war proof of work
186 2011-10-17 14:11:36 <copumpkin> I prove my work by waging wars and acquiring prisoners
187 2011-10-17 14:11:38 <copumpkin> POW^2
188 2011-10-17 14:11:48 <luke-jr> sipa: botnets are inevitable
189 2011-10-17 14:11:59 <luke-jr> but CPU-only merely raises the cost to mass-mine
190 2011-10-17 14:12:02 <copumpkin> and on that deeply intellectual assertion, I shall retire from this conversation
191 2011-10-17 14:12:07 <luke-jr> right now, at least GPUs are commodity hardware
192 2011-10-17 14:12:07 <sipa> sure, but botnets on GPUs are harder than on CPUs
193 2011-10-17 14:12:22 <gmaxwell> Sure, but we don't need to hand 90% control to the botnets.
194 2011-10-17 14:12:41 <sipa> and it is an advantage (at least regarding keeping botnets at bay) to use a system were specialized hardware has a significant advantage over commodity hardware
195 2011-10-17 14:12:45 <BlueMatt> gmaxwell: well theres room in the implementation, go write it ;)
196 2011-10-17 14:12:49 <luke-jr> regular people can buy GPUs; not so much 256-core CPUs
197 2011-10-17 14:13:00 <BlueMatt> arg installing wine+mingw takes forever...
198 2011-10-17 14:13:03 <gmaxwell> BlueMatt: yea... fair enough, I was waiting for it to mature.
199 2011-10-17 14:13:23 <BlueMatt> as was I
200 2011-10-17 14:14:45 <BlueMatt> and since I dont pay attention to the latest news in the crypto world I will probably never realize it has matured enough...
201 2011-10-17 14:15:05 <gmaxwell> ha, well, what I was waiting for was _wallet crypto_ to mature. :)
202 2011-10-17 14:15:15 <BlueMatt> heh
203 2011-10-17 14:15:45 <BlueMatt> fair enough, as I said I dont pay enough attention to the world of cryto ;)
204 2011-10-17 14:15:54 <gmaxwell> luke-jr: someone trying to outpace bitcoin using asics still needs a fair number of them, because the GPUs are pretty good competition.  Scrypt is gate hungry, but the good miners don't have anything better than desktop cpus.
205 2011-10-17 14:16:32 <BlueMatt> anyway, intro to compsci lecture is done so Im getting off :)
206 2011-10-17 14:18:28 <luke-jr> gmaxwell: that's my point
207 2011-10-17 14:19:51 <gavinandresen> Looking for feedback on some op_eval/multisig stuff I've been experimenting with:  https://gist.github.com/e7041fe221b4e5f08c3f
208 2011-10-17 14:20:19 <gavinandresen> (thinking through RPC calls to support the use cases I/we care about)
209 2011-10-17 14:24:59 <gmaxwell> luke-jr: the funniest thing about Tenebrix is that their scrypt size is only 128k... which is large enough to blow up any currently available GPU but wouldn't actually be too horiffic for a asic implementation. Only something like 11x the gate count of a bitcoin miner.
210 2011-10-17 14:25:42 <sipa> gavinandresen: that implies an address-based keystore?
211 2011-10-17 14:26:09 <gavinandresen> sipa:  I implemented it as additional keystore methods:  AddCScript()/etc
212 2011-10-17 14:26:56 <gavinandresen> sipa: https://github.com/gavinandresen/bitcoin-git/blob/op_eval/src/keystore.h
213 2011-10-17 14:27:01 <sipa> and how do you store it in the wallet?
214 2011-10-17 14:27:20 <gavinandresen> sipa:  new "cscript"<hash> key
215 2011-10-17 14:27:46 <gavinandresen> (value is serialized CScript)
216 2011-10-17 14:28:25 <gavinandresen> ... which brings up the issue of wallet backups, since there is no multisig-address-keypool
217 2011-10-17 14:28:26 <sipa> i wonder if we can't move to a generic (xkey + version byte + address hash) -> (different types of data) entry in wallet
218 2011-10-17 14:28:55 <cuqa> hello
219 2011-10-17 14:28:59 <sipa> since all lookups are based on address, but they're stored by pubkey/cscript in the wallet
220 2011-10-17 14:29:02 <cuqa> can someone explain me how MM works?
221 2011-10-17 14:29:22 <gavinandresen> M&M's are little multicolored candies with chocolate inside
222 2011-10-17 14:29:22 <sipa> cuqa: https://bitcointalk.org/index.php?topic=7219.0
223 2011-10-17 14:29:38 <cuqa> nono I mean Merged Mining
224 2011-10-17 14:29:38 <gavinandresen> You eat them.
225 2011-10-17 14:29:44 <gavinandresen> Oh....
226 2011-10-17 14:29:54 <sipa> gavinandresen: such an address-based key would also allow "watched" addresses for example
227 2011-10-17 14:31:00 <gavinandresen> sipa: bigger changes than I want to tackle right now...
228 2011-10-17 14:31:16 <gavinandresen> (the whole op_eval and new bitcoin address are plenty big enough)
229 2011-10-17 14:31:47 <sipa> gavinandresen: it's quite independent from it
230 2011-10-17 14:31:59 <gavinandresen> sipa:  yes, that's why I don't want to think about it
231 2011-10-17 14:32:12 <gmaxwell> sipa: I don't know how you make a UI around watched addresses that isn't confusing.
232 2011-10-17 14:32:40 <sipa> gmaxwell: that's the eternal question :)
233 2011-10-17 14:44:23 <gmaxwell> gavinandresen: I sort of expected an eventual sign-gather interface to allow you to add inputs to a txn which is short inputs.
234 2011-10-17 14:44:51 <Diablo-D3> woah
235 2011-10-17 14:44:57 <Diablo-D3> fglrx supports hybrid switching now
236 2011-10-17 14:45:00 <gmaxwell> gavinandresen: e.g. if I give you a seralized txn which pays 5BTC to A but there is only 1 BTC of inputs you should be able to add more inputs to the txn.
237 2011-10-17 14:45:10 <Diablo-D3> aticonfig --px-dgpu | --px-igpu | --px-list
238 2011-10-17 14:45:56 <gmaxwell> gavinandresen: there are many uses for that, but the most pratical is that the last person to sign may need to add a fee in order to make the txn acceptable to the network.
239 2011-10-17 14:47:44 <cuqa> is there a way to create a wallet.dat with a private key?
240 2011-10-17 14:48:21 <cuqa> wallet seems to be corrupted. at least its crashing on start up
241 2011-10-17 14:48:41 <gmaxwell> cuqa: have you looked in the log?
242 2011-10-17 14:48:50 <gmaxwell> also did you build your own bitcoind
243 2011-10-17 14:54:45 <urstroyer> i messed up my wallet in using experimental bitcoin client on android
244 2011-10-17 14:55:06 <gmaxwell> urstroyer: messed up how?
245 2011-10-17 14:55:16 <gmaxwell> Do you have a backup?
246 2011-10-17 14:55:18 <urstroyer> i think its a software bug
247 2011-10-17 14:55:28 <urstroyer> i was able to extract the private key of my wallet
248 2011-10-17 14:55:50 <sipa> using?
249 2011-10-17 14:55:57 <urstroyer> so iam looking for a solution to get any client working with this private key
250 2011-10-17 14:56:30 <sipa> my dumpprivkey branch has code for importing private keys
251 2011-10-17 14:56:31 <gavinandresen> gmaxwell: want to propose an RPC interface for that?  I'm aiming for something pretty general...
252 2011-10-17 14:57:17 <gavinandresen> Doesn't mtGox allow 'sweeping' private keys now, too?
253 2011-10-17 14:58:24 <sipa> oh yes
254 2011-10-17 14:58:32 <gavinandresen> gmaxwell:   (and actually, wouldn't what I proposed support that?  if you submit a 'signtx' with a transaction with insufficient inputs, the signing process could add inputs)
255 2011-10-17 14:58:34 <urstroyer> oh well thats great
256 2011-10-17 14:58:40 <urstroyer> will try that out thank you so much
257 2011-10-17 14:58:59 <gmaxwell> gavinandresen: signtx <seralized> <amount_to_add>  would work I think, defaulting to zero, perhaps with a warning when the amount takes you over your pay_tx_fee in excess.   There also needs to be a "viewtx" that lets you see what you're signing before you do a signtx ...
258 2011-10-17 14:59:31 <gmaxwell> gavinandresen: and, submit would of course reject it if it wouldn't validate due to missing inputs.
259 2011-10-17 14:59:47 <sipa> TD suggested not using serialized transactions, but a somewhat higher level description
260 2011-10-17 14:59:48 <gavinandresen> gmaxwell: viewtx should be some python code outside of bitcoin, in my opinion.
261 2011-10-17 14:59:56 <sipa> so that clients don't need to parse them
262 2011-10-17 15:00:50 <gavinandresen> So TD wants all the bitcoin implementations to have two parsers....
263 2011-10-17 15:00:54 <gmaxwell> gavinandresen: I suppose. ... disadvantage there is that without access to your wallet, the python code couldn't tell what parts of the input scripts your signatures could satisify.
264 2011-10-17 15:01:10 <gavinandresen> that's what validateaddress is for
265 2011-10-17 15:01:42 <sipa> gavinandresen: to sign a transaction, you need to know how to sign it anyway
266 2011-10-17 15:01:56 <sipa> so it will have to be a type of transaction your client knows about
267 2011-10-17 15:01:59 <gavinandresen> The RPC interface isn't intended for everyday use, it is intended for web services
268 2011-10-17 15:02:32 <gavinandresen> viewtx would be very useful for debugging, though.... probably enough reason to have it
269 2011-10-17 15:02:36 <gmaxwell> But ... I like the rpc interface. :)
270 2011-10-17 15:02:55 <gavinandresen> (I like the rpc interface, too, but I know I'm an outlier)
271 2011-10-17 15:04:28 <gmaxwell> In any case... as I was saying, just an extra argument to add inputs to signtx is probably fine, perhaps with it possibly taking a special value of "fill" where it figures out whatever is requried (incl. fee) and adds that.
272 2011-10-17 15:05:01 <gmaxwell> since it doesn't actually submit you can decide after the fact if you like the fee... something you don't get with the current interface that annoys people.
273 2011-10-17 15:06:13 <gmaxwell> The annoying thing is that you can't change the outputs, so if you don't have an input of exactly the right size you need to make another txn to break up your input.
274 2011-10-17 15:07:57 <gavinandresen> Hmmm...  I'm not sure I see the use case.  Seems to me if I'm in the process of gathering signatures I already know all the transaction details
275 2011-10-17 15:10:20 <gavinandresen> Example:  user has a (bunch of) 2-sigs-required addresses with coins (they can provide 1 signature, they need to contact a wallet protection service for the second).  They create a "send 100 BTC to gmaxwell because he Rocks" transaction...
276 2011-10-17 15:11:03 <gavinandresen> ... before the wallet protection service is contacted the complete transaction is created, selecting N inputs, paying to gmaxwell's address, change to somewhere else in the wallet, and enough fees to make sure it all gets confirmed.
277 2011-10-17 15:13:16 <gavinandresen> Same thing for escrow:  there will be X bitcoins sent in possibly Y different transactions to a 2-of-3 escrow address.  User will ask their trading partner "pretty please sign this "pay me all X bitcoins" transaction"  (or maybe "pay me X-fee")
278 2011-10-17 15:13:30 <sipa> actually
279 2011-10-17 15:14:22 <sipa> the only thing you need to ask the WPS is "please give me a signature for using coin 4f37a972:0, with txsignhash A473B47A..."
280 2011-10-17 15:14:46 <sipa> although the WPS will probably want to see the transaction itself, in order to parse it and show it to the user
281 2011-10-17 15:14:59 <gavinandresen> yes, the WPS definitely needs to see where the coins are going
282 2011-10-17 15:15:01 <sipa> but what useful information can it give, even if it can completely parse all outputs
283 2011-10-17 15:15:21 <gavinandresen> ... so it can send you an SMS saying "confirm payment of X bitcoins to BLAH"
284 2011-10-17 15:15:25 <sipa> "you are sending X coins to address Y, and Z coins to address W, and Q coins to something we don't know"
285 2011-10-17 15:15:39 <sipa> the WPS does not know your address book, and does not know what those addresses mean
286 2011-10-17 15:15:41 <gmaxwell> There is also no advantage in hiding the data since they'd learn it when it hits the blockchain.
287 2011-10-17 15:16:12 <gmaxwell> sipa: you might have told the WPS in advance to only allow payouts to some finite set of addresses, for example. Or only allow payouts outside of the set with a phone call.
288 2011-10-17 15:17:15 <gmaxwell> gavinandresen: hm. I suppose you're right you'll know the size of the txn before you begin signing even though you don't have all the data yet.
289 2011-10-17 15:17:18 <sipa> your client could tell you a randomly generated secret, and include a check for HASH(secret)=X in the txout script
290 2011-10-17 15:17:45 <sipa> if you're then able to provide the correct secret to the WPS, it knows the tx can only have been generated by your client
291 2011-10-17 15:18:00 <gmaxwell> sipa: or the badguy who has 0wned up your client.
292 2011-10-17 15:18:04 <sipa> although that is not information that should end up in the block chain
293 2011-10-17 15:18:19 <gavinandresen> yes, whole reason for the WPS is to protect from a root-kitted machine
294 2011-10-17 15:18:19 <sipa> right of course
295 2011-10-17 15:18:34 <gmaxwell> gavinandresen: there is still the SIPA and I jointly wish to pay tcatm usecase, but I guess thats a seperate enough usecase that it can't be cleanly shoehorned onto this interface. Oh well.
296 2011-10-17 15:18:55 <gavinandresen> gmaxwell: what is that case?
297 2011-10-17 15:19:14 <gmaxwell> sipa: in any case, as I was saying there is no point in hiding it from them because they'd eventually see the signature in the blockchain. The only reason I can think to hide it would be to prevent them from announcing it to the network.
298 2011-10-17 15:19:33 <sipa> there is definitely no point in hiding
299 2011-10-17 15:19:43 <sipa> i'm just questioning what they can usefully do with it
300 2011-10-17 15:19:58 <sipa> the point is that the WPS has a way of contacting you - either through SMS, or some secure web interface, ...
301 2011-10-17 15:20:27 <sipa> if an attacker has stolen your device, yes, then he has the secrer, but they can't pass it to the WPS
302 2011-10-17 15:21:11 <gmaxwell> gavinandresen: e.g. tcatm says he'll release historical market data he's collected if someone pays him 10 btc. Sipa and I want to go in for 5btc each but don't trust each other to made good on the payment.  We could pay via a join transaction (each of us provides an input). If either renig, the txn just doesn't happen (not enough output).
303 2011-10-17 15:21:21 <sipa> if the attacker has both your client and the ability to fake being you in front of the WPS, there is nothing we can do
304 2011-10-17 15:21:41 <gmaxwell> gavinandresen: to get there from your API would require the ability to add inputs, but since you can't add change if its done like that it gets messy.
305 2011-10-17 15:21:51 <da2ce7> tcatm, I'll donate 5btc to that cause.
306 2011-10-17 15:21:52 <da2ce7> :)
307 2011-10-17 15:22:11 <da2ce7> but it must be released under something like CC-0
308 2011-10-17 15:22:31 <gavinandresen> gmaxwell: got it. I think you could do that by spending to two 5btc outputs, each verifying the other owns those outputs, then one generating a txn that they ask the other to sign....
309 2011-10-17 15:23:43 <gmaxwell> hm. yes, that would be a way to do it... just some call that lets you specify some random input you don't have the keys for as one of the inputs you're using.
310 2011-10-17 15:23:57 <gmaxwell> Then you just convince the far end to sign it.
311 2011-10-17 15:24:16 <gmaxwell> Thus all inputs are known up front and you don't have any fee confusion either.
312 2011-10-17 15:24:54 <gavinandresen> gmaxwell: yup.  Again, I'm thinking an outside-bitcoind-tool that lets you find transactions you don't own and generate transactions with arbitrary inputs might be the way to go.
313 2011-10-17 15:25:38 <gavinandresen> (instead of teaching bitcoind to be able to lookup any possible input by address)
314 2011-10-17 15:26:02 <luke-jr> &
315 2011-10-17 15:26:13 <gmaxwell> Also, it can happily write out the change too.. so e.g. I could just ask sipa to identify some random input of his, and then I write up a txn that sends the change back to it.
316 2011-10-17 15:26:41 <gmaxwell> It's up to him to look to make sure I didn't decide to write it to steal the whole input, of course. ;)
317 2011-10-17 15:26:42 <gavinandresen> gmaxwell: yup... you just need an input or inputs with more than 5btc unspent
318 2011-10-17 15:27:45 <gmaxwell> luke-jr: well, we don't want to have to constantly carry every possible index in every bitcoin node, thats not efficient.
319 2011-10-17 15:27:48 <gavinandresen> ... and you have to hope that he doesn't accidentally spend those...
320 2011-10-17 15:28:16 <gmaxwell> gavinandresen: yes, well, at some point we'll need to add some functionality to lock inputs.
321 2011-10-17 15:28:46 <gmaxwell> gavinandresen: but at least if he does we can just try again.. nothing is lost.
322 2011-10-17 15:29:39 <gavinandresen> gmaxwell: actually, easiest would be:  1. Create a multisig 2-of-2 address using public keys from you and tcatm.
323 2011-10-17 15:30:23 <gavinandresen> ....mmmm..... hang on.....  might accidently burn money....
324 2011-10-17 15:30:40 <gmaxwell> Also, don't want to bring tcatm into the agreement between sipa and I.
325 2011-10-17 15:30:56 <gavinandresen> (yeah, I meant sipa, not tcatm)
326 2011-10-17 15:31:15 <gmaxwell> oh yea, you get a randsom attack using the cross sigs.
327 2011-10-17 15:31:39 <gavinandresen> yeah, I'm going to stop thinking about that use case...
328 2011-10-17 15:32:33 <gmaxwell> Joint inputs are just generally good. And I think you figured out how to make it reasonable enough just some tool that lets you pick random inputs to use in a custom txn is enough.. that plus signtx should do it.
329 2011-10-17 15:33:19 <gavinandresen> ... plus maybe a way to tell your wallet "please don't use these inputs for the next X hours"
330 2011-10-17 15:34:40 <gmaxwell> gavinandresen: yes.. a 'lock input' that takes a timeout.  This also gets us pretty close to a system where web interfaces could have users approve the actual fees on txn.
331 2011-10-17 15:35:35 <gmaxwell> e.g. a getransaction <timeout> which forms a txn and locks the input for timeout seconds.  Then the webui inspects it and gets the user to approve the fees.  Then you signtx and submit.
332 2011-10-17 16:05:14 <gjs278> ;;bc,mtgox
333 2011-10-17 16:05:15 <gribble> {"ticker":{"high":3.75996,"low":2.26,"avg":3.068002102,"vwap":2.979368442,"vol":184776,"last_all":2.59001,"last_local":2.59001,"last":2.59001,"buy":2.59001,"sell":2.6}}
334 2011-10-17 16:05:17 <gjs278> lol
335 2011-10-17 16:05:21 <gjs278> ;;bc,stats
336 2011-10-17 16:05:25 <gribble> Current Blocks: 149655 | Current Difficulty: 1468195.4272208 | Next Difficulty At Block: 151199 | Next Difficulty In: 1544 blocks | Next Difficulty In About: 1 week, 5 days, 1 hour, 4 minutes, and 16 seconds | Next Difficulty Estimate: 1310154.61035974 | Estimated Percent Change: -10.7642902253
337 2011-10-17 16:24:15 <nhodges> http://news.ycombinator.com/item?id=3121669
338 2011-10-17 17:18:08 <CIA-101> bitcoinj: miron@google.com * r239 /trunk/src/com/google/bitcoin/core/Message.java: Disable assert in Message causing failing tests
339 2011-10-17 17:25:01 <midnightmagic> hey tcatm: any chance of getting you to release raw historical data for mtgox/etc again?
340 2011-10-17 17:25:29 <midnightmagic> oh, hey gavin.
341 2011-10-17 17:26:10 <gavinandresen> howdy
342 2011-10-17 17:26:52 <phantomcircuit> midnightmagic, we have that
343 2011-10-17 17:27:32 <phantomcircuit> http://bitcoinmarkets.com/bitcoinmarkets-2011-10.sql.bz2.torrent
344 2011-10-17 17:28:24 <phantomcircuit> actually screw that link i'll get mizery to give a better dump
345 2011-10-17 17:34:49 <midnightmagic> okay thanks phantom. is that 100% all-time history?
346 2011-10-17 17:35:32 <midnightmagic> tcatm was logging full market gets for a while I think, so even changes in the bid/ask tables was available.
347 2011-10-17 17:35:39 <midnightmagic> i don't know if he's still doing that.
348 2011-10-17 17:40:01 <wumpus> i'm also logging full market depths 10 seconds
349 2011-10-17 17:41:26 <sipa> i have mtgox bids/asks every minute
350 2011-10-17 17:42:31 <midnightmagic> how far back?
351 2011-10-17 17:42:37 <wumpus> first record is from Fri May 27 10:48:48 2011
352 2011-10-17 17:42:51 <midnightmagic> ah cool. willing to share?
353 2011-10-17 17:43:09 <wumpus> yeah no problem
354 2011-10-17 17:43:19 <midnightmagic> awesome, thanks!
355 2011-10-17 17:43:36 <sipa> i have data starting from march, but with gaps of a few days sometimes
356 2011-10-17 17:46:53 <midnightmagic> well, I'd be happy to re-release a stitched-together merge of any data I get, and I have lots of bandwidth on U.S. servers to re-release it.
357 2011-10-17 17:47:09 <wumpus> yes same for me, there's probably some holes where I had no internet connection or other trouble
358 2011-10-17 17:47:19 <wumpus> but it's mostly complete
359 2011-10-17 17:48:32 <wumpus> ok this thing is huge, I had no idea
360 2011-10-17 17:50:13 <wumpus> I'll need some more efficient representation (for example, diff/changes format) to make it realistic to upload this
361 2011-10-17 17:50:29 <midnightmagic> i can handle literally any archive format that exists. :) even the weird exotic ones like pak?j
362 2011-10-17 17:51:09 <sipa> i have them per month in a .tar.lzma
363 2011-10-17 17:52:09 <midnightmagic> cool! well anywhere you want to put them, i'm happy to d/l and pass it along to the next person. (Like so neither of you would have to re-transmit.)
364 2011-10-17 17:53:18 <sipa> i'm creating archives for the last few months
365 2011-10-17 17:53:38 <sipa> it takes a while... 40000 files per month
366 2011-10-17 17:55:56 <midnightmagic> okiedokie! hey thanks, man, I appreciate it. and if wumpus wants a copy, like I said, I'm happy to put the files up for additional downloads
367 2011-10-17 17:57:12 <wumpus> mine will be a ~30GB .gz file with a json record per line...
368 2011-10-17 17:57:59 <wumpus> I'll let you know when it finished exporting
369 2011-10-17 17:59:19 <BlueMatt> copumpkin: if you feel like debugging more: https://github.com/bitcoin/bitcoin/pull/587
370 2011-10-17 17:59:29 <sipa> 3 MB per month here
371 2011-10-17 18:00:03 <copumpkin> BlueMatt: I bookmarked it but am quite busy these days so might not get to it for a while :)
372 2011-10-17 18:00:17 <copumpkin> thanks
373 2011-10-17 18:00:19 <BlueMatt> copumpkin: fair enough, I know the feeling ;)
374 2011-10-17 18:00:20 <sipa> wumpus: hint: sorting the file by date before putting them in a tar file massively increases compression ratio
375 2011-10-17 18:00:31 <BlueMatt> if anyone else feels like debugging gcc oddities please feel free :)
376 2011-10-17 18:00:36 <copumpkin> oh shit, it's the wumpus
377 2011-10-17 18:01:36 <wumpus> it's more-or-less sorted by date, 30gb is the size of the uncompressed bson file, so it was just an estimate.. it could be the result is much smaller, but bson->json->gz is taking a while :-)
378 2011-10-17 18:08:17 <neofutur> same here I can retransmit and host it
379 2011-10-17 18:09:59 <neofutur> we could have a download page with rotating urls, pointing to different servers
380 2011-10-17 18:10:19 <midnightmagic> is that @ mtgox servers? don't you already have all this data? :)
381 2011-10-17 18:10:38 <midnightmagic> i didn't want to ask because i thought it would just be a pain to release it.
382 2011-10-17 18:11:38 <wumpus> or a torrent
383 2011-10-17 18:11:49 <Tuxavant> just going to leave this right here... (browser plugin idea) http://www.reddit.com/r/Bitcoin/comments/lezlk/bitcoin_value_is_dropping_dramatically_we_need_to/c2s5spi
384 2011-10-17 18:11:50 <rjk2> torrent ftw, i have a seedbox
385 2011-10-17 18:13:37 <BlueMatt> Tuxavant: have you seen https://en.bitcoin.it/wiki/URI_Scheme
386 2011-10-17 18:13:50 <Tuxavant> reading now
387 2011-10-17 18:14:51 <Tuxavant> BlueMatt, nice... is there any word on any kind of plugin that might do what was described?
388 2011-10-17 18:14:58 <BlueMatt> bitcoin 0.5
389 2011-10-17 18:15:03 <BlueMatt> no plugin needed
390 2011-10-17 18:15:19 <BlueMatt> Tuxavant: see my comment http://www.reddit.com/r/Bitcoin/comments/lezlk/bitcoin_value_is_dropping_dramatically_we_need_to/c2s8up2
391 2011-10-17 18:16:54 <BlueMatt> Tuxavant: and by that I meant http://www.reddit.com/r/Bitcoin/comments/lezlk/bitcoin_value_is_dropping_dramatically_we_need_to/c2s8va0
392 2011-10-17 18:16:58 <Tuxavant> gotcha... fantastic
393 2011-10-17 18:17:48 <Tuxavant> what's the eta for .5 anyway?
394 2011-10-17 18:17:51 <neofutur> midnightmagic: nop my own servers
395 2011-10-17 18:17:57 <midnightmagic> neofutur: k
396 2011-10-17 18:18:11 <neofutur> midnightmagic: i sell a service to mtgo as an independant worker, i m not an employee
397 2011-10-17 18:18:12 <BlueMatt> its already rc1, but building bitcoin-qt.exe is not looking good...
398 2011-10-17 18:18:23 <neofutur> my main work is https://bitcointalk.org/?topic=1687.0
399 2011-10-17 18:19:26 <neofutur> and afaik mtgox is not saving all the historic data, too many more urgent things to do
400 2011-10-17 18:19:26 <wumpus> you can build it succesfully on windows though
401 2011-10-17 18:19:53 <neofutur> and if they have . . . I dont :p
402 2011-10-17 18:19:58 <wumpus> cross-compile from linux still needs to be fixed :/
403 2011-10-17 18:20:01 <luke-jr> BlueMatt: &
404 2011-10-17 18:20:44 <luke-jr> Tuxavant: I have bitcoin-qt.exe for git just after rc1 ;)
405 2011-10-17 18:20:45 <Tuxavant> luke-jr, BlueMatt biting my nails in anticipation 8D
406 2011-10-17 18:21:10 <sipa> wumpus, BlueMatt: what is the problem exactly with the build?
407 2011-10-17 18:21:18 <luke-jr> Tuxavant: http://luke.dashjr.org/programs/bitcoin/files/bitcoin-qt-0.5.0rc1-755e2819.exe
408 2011-10-17 18:21:31 <luke-jr> Tuxavant: possibly throw these in the same dir http://luke.dashjr.org/programs/bitcoin/files/bitcoin-qt-libs.zip
409 2011-10-17 18:21:42 <wumpus> well it builds, the resulting exe is just borked, return arguments get corrupted in some calling convention issue
410 2011-10-17 18:22:03 <Tuxavant> luke-jr, sending a .exe to a guy named Tux... shame on you 8D
411 2011-10-17 18:22:20 <BlueMatt> sipa: its still the old mismatching calling conventions from earlier...
412 2011-10-17 18:22:38 <BlueMatt> luke-jr: yes, I know but its not reproduceable last time I checked...
413 2011-10-17 18:22:41 <wumpus> are you cross-compiling luke-jr?
414 2011-10-17 18:22:46 <luke-jr> maybe it's because you're trying to build for i386, and mingw is supposed to be i686 now
415 2011-10-17 18:22:57 <luke-jr> BlueMatt: what's not reproducable?
416 2011-10-17 18:23:01 <luke-jr> wumpus: yes
417 2011-10-17 18:23:03 <BlueMatt> your binary
418 2011-10-17 18:23:07 <luke-jr> BlueMatt: sure it is
419 2011-10-17 18:23:15 <BlueMatt> easily?
420 2011-10-17 18:23:20 <luke-jr> no, not easily
421 2011-10-17 18:23:22 <luke-jr> it was a pain
422 2011-10-17 18:23:27 <BlueMatt> exactly
423 2011-10-17 18:23:34 <BlueMatt> https://github.com/bitcoin/bitcoin/pull/587
424 2011-10-17 18:23:36 <luke-jr> I blame you guys for only making stuff work with Ubuntu :P
425 2011-10-17 18:23:42 <wumpus> luke-jr: yes it's probably something with the compiler flags, i386/686 COULD be the issue, one of the many
426 2011-10-17 18:24:07 <BlueMatt> of course it is, well more likely headers somewhere in qt...
427 2011-10-17 18:24:10 <luke-jr> http://paste.pocoo.org/show/494067/ <-- changes I made to bitcoin-qt
428 2011-10-17 18:24:23 <sipa> midnightmagic: http://bitcoin.sipa.be/mtgox.tgz
429 2011-10-17 18:24:25 <luke-jr> BlueMatt: nonsense, everyone can cross-compiles Qt apps fine
430 2011-10-17 18:24:34 <wumpus> it doesn't only work with ubuntu
431 2011-10-17 18:24:48 <wumpus> there's nothing ubuntu specific used at all
432 2011-10-17 18:24:57 <BlueMatt> luke-jr: I know, but it is something specific to qt+bitcoin
433 2011-10-17 18:25:01 <BlueMatt> not just bitcoin
434 2011-10-17 18:25:10 <luke-jr> BlueMatt: obviously not, since qt+bitcoin works fine here
435 2011-10-17 18:25:16 <wumpus> it would be interesting to try to compile the qt demos
436 2011-10-17 18:25:25 <midnightmagic> thanks sipa! you rule man, as usual.
437 2011-10-17 18:25:50 <midnightmagic> 3fee767aee598f364554a623196fba47cc9d640c07eaa4514ef5690f2bc8e486  mtgox.tgz   <-- that it?
438 2011-10-17 18:25:57 <luke-jr> oh, probably because of that "mingw32msvc" bit
439 2011-10-17 18:26:05 <BlueMatt> luke-jr: I made some of the same modifications...also, Im done discussing how your build works, thats great in fact, I have no doubt its not hard to build a working copy on win32/fedora/etc/etc/etc, but I couldnt care less
440 2011-10-17 18:26:07 <luke-jr> *normal* mingw is NOT compatible with msvc
441 2011-10-17 18:26:29 <wumpus> luke-jr: c is, c++ isn't
442 2011-10-17 18:26:41 <luke-jr> wumpus: Bitcoin-Qt is all C++ and Qt
443 2011-10-17 18:26:45 <wumpus> c++ calling convention is not standardized on windows
444 2011-10-17 18:26:50 <wumpus> luke-jr: right
445 2011-10-17 18:26:58 <wumpus> woo, seems we found our issue
446 2011-10-17 18:27:16 <luke-jr> so does Ubuntu have some hack of MingW that tries to be MSVC-compatible?
447 2011-10-17 18:27:43 <wumpus> I have no idea
448 2011-10-17 18:28:34 <luke-jr> maybe gitian should just switch to Gentoo which has saner crossdev tools
449 2011-10-17 18:29:19 <midnightmagic> if anyone wants sipa's file (assuming it's cool with sipa) let me know I have a URL here with a relatively fast US-based server.
450 2011-10-17 18:29:19 <wumpus> so we first spend 10 hours building our linux distribution before we can start building bitcoin? :P
451 2011-10-17 18:30:11 <gmaxwell> midnightmagic: hm? it's only 20mbytes.
452 2011-10-17 18:30:16 <wumpus> or isn't gentoo based on the emerge stuff anymore? been a long time for me
453 2011-10-17 18:30:27 <BlueMatt> still is
454 2011-10-17 18:30:38 <midnightmagic> oh! so it is. LOL. My eyes were off by a couple orders of magnitude. I thought it downloaded fast..
455 2011-10-17 18:31:14 <BlueMatt> oh, also qt build isnt deterministic yet, so theres that...
456 2011-10-17 18:33:30 <midnightmagic> woo! compression ratio of 308x, very nice.
457 2011-10-17 18:39:43 <luke-jr> wumpus: or you just untar the latest stage3&
458 2011-10-17 18:39:58 <luke-jr> wumpus: maybe crossdev binary packages if your PC is slow
459 2011-10-17 18:40:44 <luke-jr> BlueMatt: deterministicness should not be a blocker
460 2011-10-17 18:40:47 <luke-jr> that's a gitian issue/goal, not bitcoin-related
461 2011-10-17 18:41:00 <luke-jr> not a single other program is built deterministic
462 2011-10-17 18:41:04 <gmaxwell> Thats kind of frightening though and it makes it impossible for people to validate that the binary is what you claim it is.
463 2011-10-17 18:41:14 <luke-jr> gmaxwell: so? that's the status quo of all software
464 2011-10-17 18:41:22 <gmaxwell> Most programs don't do what bitcoin does.
465 2011-10-17 18:41:23 <luke-jr> even with gitian, it's still true
466 2011-10-17 18:41:38 <luke-jr> because Ubuntu's compiler could (and is) be messing with its generation
467 2011-10-17 18:41:39 <gmaxwell> Why isn't it determinstic?
468 2011-10-17 18:41:48 <luke-jr> you're just trusting Ubuntu
469 2011-10-17 18:41:54 <gmaxwell> My builds of my packages are determinstic.
470 2011-10-17 18:42:04 <gmaxwell> Yes, but better than trusting ubuntu plus something else
471 2011-10-17 18:42:17 <luke-jr> point is, it's a new unproven technology
472 2011-10-17 18:42:19 <luke-jr> not the norm
473 2011-10-17 18:42:31 <luke-jr> gitian's goals don't need to force bitcoin-qt to wait
474 2011-10-17 18:42:34 <gmaxwell> (and I often use checksums of the object files in order to make sure that syntax only changes really were syntax only changes)
475 2011-10-17 18:43:15 <gmaxwell> If the non-determinism is not understood it could be a result of a bug which is making bitcoin get miscompiled.
476 2011-10-17 18:43:53 <gmaxwell> Ideally someone should take the resulting binaries and use objdump -D and diff the output to see how they differ.
477 2011-10-17 18:44:48 <luke-jr> there are bigger problems than making Bitcoin-Qt deterministic
478 2011-10-17 18:44:58 <luke-jr> like making bitcoind not randomly corrupt wallets
479 2011-10-17 18:45:44 <gmaxwell> luke-jr: do we have any data on that yet?
480 2011-10-17 18:46:02 <luke-jr> nope
481 2011-10-17 18:55:32 <neofutur> sipa: mirrors : europe : http://xena.ww7.be/bitcoinmirror/ http://cahier2.ww7.be/bitcoinmirror/  US : http://archives.ww7.pe/bitcoinmirror/
482 2011-10-17 19:29:14 <Diablo-D3> gmaxwell: you know, I was thinking about that problem wrong
483 2011-10-17 19:29:44 <Diablo-D3> gmaxwell: I basically cant write FRAME_BUFFER_SIZE worth of samples
484 2011-10-17 19:36:04 <midnightmagic> wumpus: I don't suppose that archive of yours is up somewhere?
485 2011-10-17 19:36:09 <midnightmagic> :-)
486 2011-10-17 19:39:04 <kinlo> hmmmz
487 2011-10-17 19:39:25 <kinlo> anyone read that article about the person behind satoshi?
488 2011-10-17 19:39:36 <kinlo> anyone who can comment on that?
489 2011-10-17 19:39:51 <jeremias> what article
490 2011-10-17 19:39:54 <jeremias> there are numerous
491 2011-10-17 19:41:05 <jeremias> and all of them are bullshit
492 2011-10-17 19:41:18 <jeremias> afaik
493 2011-10-17 19:41:19 <kinlo> http://cryptome.org/0005/bitcoin-who.pdf
494 2011-10-17 19:41:37 <jeremias> well yeah definitely not true
495 2011-10-17 19:41:56 <jeremias> or that they haven't found the real satoshi
496 2011-10-17 22:35:02 <qscgy_> hi everyone
497 2011-10-17 22:41:37 <qscgy_> is it worth it to port bitcoin to C
498 2011-10-17 22:41:54 <Diablo-D3> =|