1 2011-10-31 01:10:17 <CIA-101> libbitcoin: genjix * r505e1af70346 / (6 files in 4 dirs): handshake convenience wrapper.
  2 2011-10-31 01:20:40 <CIA-101> libbitcoin: genjix * reec55c6e2893 /tests/net.cpp: use std::array.fill(0) instead of {0, 0, 0, ...
  3 2011-10-31 01:25:48 <diki> tcatm:where can i have a look at your original 4way implementation?
  4 2011-10-31 01:26:04 <diki> the one in cgminer/minerd is not your original one
  5 2011-10-31 01:31:33 <theymos> I'm surprised at how many bugs there have been in Script. OP_CHECKMULTISIG is broken and now OP_0 has been determined to not work. You'd think Satoshi would have put a lot of time into perfecting this part, since it can't be easily changed.
  6 2011-10-31 01:32:13 <gmaxwell> theymos: eh, check multisig's breakage is kinda boring.
  7 2011-10-31 01:32:39 <gmaxwell> whats this with OP_0?
  8 2011-10-31 01:32:58 <luke-jr> how can OP_0 be broken? O.o
  9 2011-10-31 01:33:26 <theymos> OP_0 should push a numeric 0 value onto the stack, but it actually "pushes 0 bytes" -- a no-op. You have to subtract 1 from OP_1 if you want 0...
 10 2011-10-31 01:34:34 <luke-jr> O.o
 11 2011-10-31 01:44:45 <batouzo> theymos: funny that this wanst detected in practice yet
 12 2011-10-31 01:45:56 <theymos> It was discovered by genjix after someone used it in those recent broken transactions.
 13 2011-10-31 01:46:37 <Diablo-D3> what was?
 14 2011-10-31 01:46:51 <Diablo-D3> oh now I see
 15 2011-10-31 01:47:02 <Diablo-D3> theymos: satoshi is an okay software engineer, but hes a shitty coder
 16 2011-10-31 01:47:17 <Diablo-D3> hes an academic
 17 2011-10-31 01:47:29 <Diablo-D3> he can design the systems pretty well, but not implement them worth a damn
 18 2011-10-31 01:47:49 <Diablo-D3> very rarely do you get someone who does both good, so Im not going to blame him for it
 19 2011-10-31 02:10:36 <gmaxwell> Diablo-D3: Please. You have a very warped idea about the abilities of a "good coder"
 20 2011-10-31 02:10:56 <[Tycho]> Scripts are cool.
 21 2011-10-31 02:11:11 <gmaxwell> The best coder on earth can't create systems even simpler than bitcoin without bugs.  The software used in aerospace is only reliable due to insane testing requirements.
 22 2011-10-31 02:11:33 <Diablo-D3> good coders can cat /dev/urandom into a .pl file, have it actually run, and then understand what it does after 6 months of not reviewing the code
 23 2011-10-31 02:11:59 <Diablo-D3> a GREAT coder can do it with lisp.
 24 2011-10-31 02:12:10 <cocktopus> lol
 25 2011-10-31 02:12:12 <gmaxwell> (e.g. having a test setup which does full combinatoric satisfication of all branch conditions, and other insane things which are only pratical if you make the software very simple)
 26 2011-10-31 02:13:52 <copumpkin> academic == shit coder?
 27 2011-10-31 02:14:15 <copumpkin> I guess I should be glad I left academia then
 28 2011-10-31 02:17:46 <Diablo-D3> copumpkin: yes, but you stripped out the context
 29 2011-10-31 02:17:58 <Diablo-D3> academics simply dont have 10 years of experience in their chosen language
 30 2011-10-31 02:18:23 <Diablo-D3> they're too busy trying to satisfy phd requirements in college so they can pick up chicks with their phd
 31 2011-10-31 02:18:41 <cocktopus> piled higher and deeper
 32 2011-10-31 02:18:58 <copumpkin> Diablo-D3: oh, I must've missed that
 33 2011-10-31 02:19:25 <copumpkin> Diablo-D3: when I was working on my phd, I'd been programming since I was 10, as had a couple of my colleagues
 34 2011-10-31 02:19:33 <copumpkin> we also advocated release of code with papers, and blasphemy like that
 35 2011-10-31 02:19:47 <copumpkin> this is why I failed, I guess
 36 2011-10-31 02:20:01 <Diablo-D3> copumpkin: yeah no shit, Im surprise they didnt throw you out
 37 2011-10-31 02:20:21 <Diablo-D3> thats blasphemy to academics
 38 2011-10-31 02:20:22 <copumpkin> :)
 39 2011-10-31 02:20:52 <Diablo-D3> honestly, if you've been coding that long, traditional college can offer you nothing of value
 40 2011-10-31 02:20:56 <Diablo-D3> not at that cost, anyways
 41 2011-10-31 02:21:03 <gmaxwell> But seriously that kind of super enormous ego about "good coders don't write bugs" is a bunch of bullshit, and its why so much of the software is as buggy as it is.
 42 2011-10-31 02:21:21 <gmaxwell> People don't take precautions to find and deal with bugs because they know they are good coders.
 43 2011-10-31 02:21:24 <Diablo-D3> my code isnt buggy :<
 44 2011-10-31 02:21:28 <gmaxwell> Bullshit.
 45 2011-10-31 02:21:36 <Diablo-D3> fine, its not AS buggy
 46 2011-10-31 02:22:04 <gmaxwell> As what? bitcoin? bitcoin has had fairly few bugs.
 47 2011-10-31 02:22:32 <Diablo-D3> its written in c++ and uses boost
 48 2011-10-31 02:22:40 <Diablo-D3> thats not only two bugs, they're major design flaws
 49 2011-10-31 02:22:50 <[Tycho]> Java is much worse.
 50 2011-10-31 02:22:58 <[Tycho]> C is the best language ever.
 51 2011-10-31 02:23:12 <gmaxwell> [Tycho]: For writing reliable software? Nah. Java sucks in lots of ways but it promotes reliablity.
 52 2011-10-31 02:23:57 <gmaxwell> as far as C++ and boost well, bitcoin is a fraction of the size it would be if written in straight C (esp if using GTK). Thats fewer chances for things to to wrong.
 53 2011-10-31 02:24:16 <luke-jr> [23:19:26] <copumpkin> Diablo-D3: when I was working on my phd, I'd been programming since I was 10, as had a couple of my colleagues
 54 2011-10-31 02:24:17 <copumpkin> Diablo-D3: I learned lots of stuff unrelated to coding, and enjoyed the more theoretical CS I learned. In fact, I got sucked into CS theory and math a lot deeper after grad school
 55 2011-10-31 02:24:18 <gmaxwell> It's use of C++ is also fairly limited, it doesn't do anything super offensive like overloading shit in surprising ways.
 56 2011-10-31 02:24:22 <luke-jr> copumpkin: why waste your money?
 57 2011-10-31 02:24:28 <copumpkin> luke-jr: they pay me to do a phd
 58 2011-10-31 02:24:38 <luke-jr> oh
 59 2011-10-31 02:24:42 <luke-jr> silly people
 60 2011-10-31 02:24:44 <copumpkin> well
 61 2011-10-31 02:24:45 <gmaxwell> luke-jr: thats the norm, btw.
 62 2011-10-31 02:24:47 <copumpkin> it's basically a research job
 63 2011-10-31 02:24:51 <copumpkin> a badly paid one
 64 2011-10-31 02:24:54 <copumpkin> which is why I eventually quit
 65 2011-10-31 02:24:56 <gmaxwell> (well, its a norm in the sciences)
 66 2011-10-31 02:24:58 <luke-jr> lol
 67 2011-10-31 02:25:30 <luke-jr> gmaxwell: what is?
 68 2011-10-31 02:25:32 <copumpkin> </ninjacoder>
 69 2011-10-31 02:25:40 <copumpkin> luke-jr: stipends for phd students
 70 2011-10-31 02:25:46 <gmaxwell> luke-jr: In the sciences you get paid for getting your PHD.
 71 2011-10-31 02:25:52 <gmaxwell> ...You get paid not especially well to get your phd while doing bitch work for some professor.
 72 2011-10-31 02:26:02 <luke-jr> XD
 73 2011-10-31 02:26:15 <copumpkin> my advisor used to joke that getting a phd was an exercise in turning down highly-paid jobs in favor of a shit salary for four or five years
 74 2011-10-31 02:26:30 <Diablo-D3> [11:24:17] <copumpkin> Diablo-D3: I learned lots of stuff unrelated to coding, and enjoyed the more theoretical CS I learned. In fact, I got sucked into CS theory and math a lot deeper after grad school
 75 2011-10-31 02:26:40 <Diablo-D3> thats a different thing
 76 2011-10-31 02:26:46 <copumpkin> yep
 77 2011-10-31 02:26:53 <Diablo-D3> copumpkin: thats almost akin to the whole "dont drop out, drop in" theory of steve job's
 78 2011-10-31 02:26:57 <Diablo-D3> which I _do_ believe in
 79 2011-10-31 02:27:02 <copumpkin> what was that?
 80 2011-10-31 02:27:16 <Diablo-D3> he quit working on his degree, but started dropping in on actually interesting classes
 81 2011-10-31 02:27:30 <Diablo-D3> ended up being one of the most important things he ever did
 82 2011-10-31 02:27:36 <copumpkin> ah okay :)
 83 2011-10-31 02:28:00 <copumpkin> yeah, I was kind of blown away by some of the courses I took as an undergrad, in topics I'd never have thought to look otherwise
 84 2011-10-31 02:28:16 <Diablo-D3> too many college programs are assembly line shit
 85 2011-10-31 02:28:30 <Diablo-D3> raw student material goes in, piece of useless paper comes out
 86 2011-10-31 02:28:36 <copumpkin> hah
 87 2011-10-31 02:28:55 <gmaxwell> OMG diplomas are made of PEOPLE
 88 2011-10-31 02:29:10 <Diablo-D3> hah
 89 2011-10-31 02:29:14 <cocktopus> meatploma
 90 2011-10-31 02:29:14 <Diablo-D3> seriously though, its insane
 91 2011-10-31 02:29:22 <Diablo-D3> people pay all this goddamned money for what
 92 2011-10-31 02:29:34 <luke-jr> lol @ Alexander the Great
 93 2011-10-31 02:29:37 <copumpkin> certificates that they can put up with structured bullshit :)
 94 2011-10-31 02:29:40 <Diablo-D3> if you arent learning anything that truly means something to you, why the fuck are you there
 95 2011-10-31 02:29:53 <cocktopus> a chance to learn how to effectively use hookers and blow!
 96 2011-10-31 02:30:35 <copumpkin> also, in some sense they can be certificates that they're in the top percentiles of the population ranked by certain (possibly irrelevant for jobs, but not all employers think so) criteria
 97 2011-10-31 02:30:57 <copumpkin> depending on whether you went to a "good" school or not
 98 2011-10-31 02:31:13 <luke-jr> no school is good
 99 2011-10-31 02:31:15 <Diablo-D3> thats another thing
100 2011-10-31 02:31:25 <Diablo-D3> why the fuck are jobs that do NOT need deplomas requiring them now?
101 2011-10-31 02:31:56 <Diablo-D3> the government should be looking into why less people go to trade schools now adays
102 2011-10-31 02:32:11 <copumpkin> Diablo-D3: people want good/smart employees, and it's hard to gauge how good/smart someone is from a couple of interviews, so people try to quantify it with a piece of paper
103 2011-10-31 02:32:13 <Diablo-D3> we need more people who _do_ shit
104 2011-10-31 02:32:18 <copumpkin> it fails at it, but might be moderately correlated
105 2011-10-31 02:32:23 <Diablo-D3> copumpkin: its very easy in this industry
106 2011-10-31 02:32:29 <copumpkin> yeah, but in most it's harder
107 2011-10-31 02:32:39 <Diablo-D3> have github account? no? NEXT!
108 2011-10-31 02:32:44 <copumpkin> artists and CS people have portfolios
109 2011-10-31 02:32:49 <cocktopus> trades are important, but kids these days are too lazy and afraid of monsters and shit
110 2011-10-31 02:53:31 <luke-jr> Diablo-D3: GitHub sucks
111 2011-10-31 02:54:50 <cocktopus> luke-jr; but everyone is doing it! :P
112 2011-10-31 02:55:09 <luke-jr> ;;bc,stats
113 2011-10-31 02:55:13 <gribble> Current Blocks: 151201 | Current Difficulty: 1203461.92638 | Next Difficulty At Block: 153215 | Next Difficulty In: 2014 blocks | Next Difficulty In About: 2 weeks, 5 days, 10 hours, 34 minutes, and 36 seconds | Next Difficulty Estimate: 1731602.48482576 | Estimated Percent Change: 43.8851073614
114 2011-10-31 03:12:19 <batouzo> JFK911: re our bubble talk, there are some rumors of upcoming next one (in usa) - studen loans
115 2011-10-31 03:15:34 <batouzo> http://news.slashdot.org/story/11/10/30/235248/student-loans-in-america-the-next-big-credit-bubble
116 2011-10-31 03:17:11 <RAM2012> I hope there is some reform in that area. tuition is too high because the govt lends money too easily
117 2011-10-31 03:25:12 <phungus> I personally know of people who are refusing to pay their student loans back but yet are continuing to go to school because they can and they think the government owes it to them
118 2011-10-31 03:25:33 <phungus> that bubble is going to explode
119 2011-10-31 03:25:43 <phungus> err,
120 2011-10-31 03:25:45 <phungus> implode
121 2011-10-31 03:25:46 <phungus> maybe
122 2011-10-31 03:27:34 <batouzo> "suddenly German debt decreased by 55,000,000,000 EUR because the debt was by mistake calculated as too high" <--- seriously?
123 2011-10-31 03:28:09 <nanotube> luke-jr: well, tbh if someone says "no but i do have gitorious acct" that will be acceptable :)
124 2011-10-31 03:28:33 <nanotube> or even no, but i do know how to use git and host my own repos ...
125 2011-10-31 03:29:17 <copumpkin> NO
126 2011-10-31 03:29:19 <copumpkin> IT MUST BE SOCIAL
127 2011-10-31 03:29:25 <nanotube> haha
128 2011-10-31 03:29:34 <copumpkin> hosting your own repos is the opposite of team player
129 2011-10-31 03:29:38 <copumpkin> and we only hire team players
130 2011-10-31 03:29:38 <nanotube> lol
131 2011-10-31 03:29:59 <copumpkin> also, rockstar ninja awesome team players
132 2011-10-31 03:31:36 <nanotube> joke rocket losing propulsion, estimated crash into ocean in t+15 :P
133 2011-10-31 03:31:54 <copumpkin> :)
134 2011-10-31 03:32:21 <copumpkin> here, enjoy this instead: http://www.youtube.com/watch?v=QtJ77qsLrpw
135 2011-10-31 03:34:20 <copumpkin> it's good in a nerdy sense
136 2011-10-31 03:34:22 <copumpkin> it's the game of life
137 2011-10-31 03:35:30 <gmaxwell> git /window 19
138 2011-10-31 03:35:32 <gmaxwell> oops
139 2011-10-31 03:35:43 <Diablo-D3> er, what?
140 2011-10-31 03:36:28 <Diablo-D3> [12:27:34] <batouzo> "suddenly German debt decreased by 55,000,000,000 EUR
141 2011-10-31 03:36:28 <nanotube> copumpkin: oh heh nice
142 2011-10-31 03:36:34 <Diablo-D3> 55 billion?
143 2011-10-31 03:36:52 <batouzo> Diablo-D3: that is what the article said
144 2011-10-31 03:36:54 <Diablo-D3> yeah but
145 2011-10-31 03:36:56 <Diablo-D3> that snothing
146 2011-10-31 03:37:00 <Diablo-D3> thats within the margin of error
147 2011-10-31 03:37:08 <Diablo-D3> the German debt is measured in trillions
148 2011-10-31 03:37:13 <batouzo> http://www.moneycontrol.com/news/wire-news/germany-finds-55-bln-euros-after-accountancy-error_607515.html
149 2011-10-31 03:38:52 <batouzo> Diablo-D3: then it is not directly a big change for economy, except, someone is really bad at managing/governing there
150 2011-10-31 03:39:03 <nanotube> yea it's like you finding out that something costs not 3000 usd, but only 2945. that's pretty nice, but nothing to write home about
151 2011-10-31 03:39:49 <batouzo> you think such mistakes are common? or it shows problems in administration
152 2011-10-31 03:40:14 <nanotube> no idea. but running any large bureaucracy just has to be full of mistakes, i'd bet
153 2011-10-31 03:40:19 <nanotube> not that i have experience in running one
154 2011-10-31 03:40:28 <batouzo> oh well bitcoin also has such mistakes like 2000+ ? burned in OP 0 transactions yesterday
155 2011-10-31 03:40:43 <nanotube> hehe yes
156 2011-10-31 03:40:59 <nanotube> "suddenly mtgox balance decreased by 2000 btc"
157 2011-10-31 03:41:00 <batouzo> but at least nothing can change total amount of coins
158 2011-10-31 03:41:33 <gmaxwell> As I said when it happend that wasn't even a particularly expensive mistake. Crap happens.
159 2011-10-31 03:41:41 <copumpkin> I'll take the 2000 lost btc, thanks
160 2011-10-31 03:41:52 <nanotube> yes, everyone's bitcoins just became 2e3/7e6 more valuable.
161 2011-10-31 03:42:06 <nanotube> copumpkin: be sure to share with your good friend nanotube !
162 2011-10-31 03:42:11 <copumpkin> certainly!
163 2011-10-31 03:42:16 <copumpkin> if they magically appear in my wallet
164 2011-10-31 03:42:17 <nanotube> hehe
165 2011-10-31 03:43:07 <nanotube> gmaxwell: indeed. exactly my point about german 55bln as well. :)
166 2011-10-31 03:46:23 <MagicalTux> is there any mining software using midstate ?
167 2011-10-31 03:46:41 <graingert> !later tell BlueMatt ppa works wonderfully
168 2011-10-31 03:46:41 <gribble> The operation succeeded.
169 2011-10-31 03:48:16 <jgarzik> MagicalTux: requiring, yes.  using, no.
170 2011-10-31 03:51:33 <shadders> jgarzik: just the dude who might be able to unmuddle me... 12 hrs of debugging later I've just noticed that bitcoinj reverses tx hash when building merkle root.  But from what I can see of the bitcoind code this doesn't happen.
171 2011-10-31 03:51:53 <shadders> building merkle tree*
172 2011-10-31 03:52:50 <shadders> any idea what's going on?  does bitcoind maybe store the tx bytes internally in LE?
173 2011-10-31 03:54:07 <MagicalTux> jgarzik: they require it but don't use it? why don't we drop it altogether then?
174 2011-10-31 03:54:27 <MagicalTux> (in fact I'm working on an implementation of getwork right now, and the documentation about that one is kind of lacking)
175 2011-10-31 03:56:07 <shadders> MagicalTux: you can calculate it yrself but a lot of crypto libs don't give you access to internal state.  For poolserverj I had to hack bouncycastle lib to be able to extract the midstate
176 2011-10-31 03:56:34 <nanotube> there's been some talk of just getting rid of the midstate, iirc.
177 2011-10-31 03:58:26 <shadders> maybe that's why it was included originally... I don't think it really hurts... if your going to modify getwork you might as well do it properly and implement a differential system where first work contains all the extra info like midstate, target, full header etc... and subsequent ones only include merkleroot+time which is all that changes.  And a full one whenever any of the static data items changes
178 2011-10-31 03:59:51 <MagicalTux> heh
179 2011-10-31 04:00:07 <MagicalTux> so as of today, no (or almost none) mining software rely on midstate ?
180 2011-10-31 04:01:16 <[Tycho]> Only ~80 bytes should be enough for getwork :)
181 2011-10-31 04:01:35 <shadders> MagicalTux: I think cgminer is the only one
182 2011-10-31 04:02:31 <shadders> [Tycho]: I'm trying to put together a spec for binary differential getwork protocol (~40 byte if binary)
183 2011-10-31 04:02:49 <shadders> but not getting much interest from miner devs
184 2011-10-31 04:03:02 <Ukto> MagicalTux: gonna start your own pool?
185 2011-10-31 04:03:20 <shadders> Diablo-D3 is a lazy prick and permanently cranky
186 2011-10-31 04:03:23 <[Tycho]> shadders, I have an idea about light-RPC with selectable fields to request.
187 2011-10-31 04:04:27 <MagicalTux> Ukto: may
188 2011-10-31 04:04:59 <Diablo-D3> Im a lazy prick whos always doing work.
189 2011-10-31 04:05:06 <Diablo-D3> hows that working for you, shadders?
190 2011-10-31 04:05:44 <shadders> works ok... still curious why you've been in bad mood since 1986 though :p
191 2011-10-31 04:06:03 <Diablo-D3> s/1986/1983/
192 2011-10-31 04:08:19 <RAM2012> tux if you start a pool it should be for merged mining
193 2011-10-31 04:09:33 <shadders> anyone know where I can get a raw block (bytes or hex) easily?  I'm going to have to test this backwards tx thing myself I guess
194 2011-10-31 04:10:13 <RAM2012> There should be an exchange supported by a network of bitcoin / namecoin ATMS
195 2011-10-31 04:11:27 <jgarzik> MagicalTux: cpuminer and cgminer will fail if any of the four fields are not present
196 2011-10-31 04:11:36 <jgarzik> MagicalTux: even though not all fields are actually used
197 2011-10-31 04:11:57 <jgarzik> MagicalTux: it's a question that's been asked before...  :)
198 2011-10-31 04:12:04 <MagicalTux> jgarzik: what about providing a null midstate ?
199 2011-10-31 04:12:43 <Diablo-D3> heh
200 2011-10-31 04:12:50 <Diablo-D3> jgarzik: which fields?
201 2011-10-31 04:12:52 <Diablo-D3> in what?
202 2011-10-31 04:13:14 <Diablo-D3> because DiabloMiner needs the entire 80 byte header to verify solution
203 2011-10-31 04:13:31 <jgarzik> MagicalTux: I take that back.  midstate is used directly by 4way and sse2_64 algorithms
204 2011-10-31 04:13:38 <MagicalTux> ok
205 2011-10-31 04:13:41 <MagicalTux> then I'll compute it
206 2011-10-31 04:13:48 <Graet> latest cgminer doesnt need midstate according to changelog
207 2011-10-31 04:16:03 <Nesetalis> anyone know where a list of bitcoin nodes is?
208 2011-10-31 04:18:01 <Diablo-D3> jgarzik: diablominer uses midstate
209 2011-10-31 04:18:07 <MagicalTux> Nesetalis: got one
210 2011-10-31 04:18:20 <Diablo-D3> although I do have sha256 round shit in the java, so someone could easily submit a patch to remove that
211 2011-10-31 04:18:40 <MagicalTux> Diablo-D3: do you know if there's some documentation somewhere about getwork ?
212 2011-10-31 04:18:53 <Diablo-D3> nope
213 2011-10-31 04:19:02 <Diablo-D3> not that Im aware of anyways
214 2011-10-31 04:19:07 <Diablo-D3> I just do what the json says
215 2011-10-31 04:25:52 <Nesetalis> -sighs- IPv6 needs to move forward and NAT needs to go the way of the dinosaur
216 2011-10-31 04:28:57 <batouzo> would jgarzik's pool (python) work with litecoin?
217 2011-10-31 04:29:25 <gmaxwell> Why the heck would you pool litecoin?
218 2011-10-31 04:29:50 <shadders> batouzo: poolserverj has scrypt support now so it should work with litecoin
219 2011-10-31 04:30:01 <Diablo-D3> yeah but who the fuck has a scrypt miner
220 2011-10-31 04:30:04 <gmaxwell> grep -i genera ~/.litecoin/debug.log  | wc -l
221 2011-10-31 04:30:09 <gmaxwell> 17
222 2011-10-31 04:30:10 <batouzo> gmaxwell: why not?
223 2011-10-31 04:30:10 <shadders> don't think anyone's tried it though
224 2011-10-31 04:30:13 <gmaxwell> ^ thats in a bit under 48 hours.
225 2011-10-31 04:30:30 <batouzo> is there cpu miner for litecoin?
226 2011-10-31 04:30:50 <gmaxwell> No. Goat based mining only...
227 2011-10-31 04:30:52 <cjdelisle> hehe litecoin pool
228 2011-10-31 04:30:55 <gmaxwell> What do you think? :)
229 2011-10-31 04:31:39 <cjdelisle> I have 300,000 cpus available to me, what should I do with them... hmmm. Litecoin Pool!
230 2011-10-31 04:31:40 <batouzo> do we have standalone cpu miner for it or just the code in litecoind itself
231 2011-10-31 04:32:17 <batouzo> cjdelisle: yes something like that
232 2011-10-31 04:32:17 <gmaxwell> cjdelisle: if you have 300k cpus, ... care to help me do some fun things to litecoin? ;)
233 2011-10-31 04:32:36 <batouzo> to a lesser scale :)
234 2011-10-31 04:34:00 <gmaxwell> aww.
235 2011-10-31 04:34:08 <batouzo> all this poor CPUs that must suffer so we can test our new perfumes
236 2011-10-31 04:34:54 <cjdelisle> oh messing with litecoin.. meh, it's the botnet part that I'm not in to
237 2011-10-31 04:35:11 <batouzo> if only we could in future move this to something without hurting them, like use animals instead?
238 2011-10-31 04:35:36 <batouzo> cjdelisle: who said botnet?
239 2011-10-31 04:36:29 <cjdelisle> I'll entertain the idea that you can get 300k cpus other ways for a few minutes
240 2011-10-31 04:36:40 <gmaxwell> litecoin seems like a fantastic platform for exploring bloat attacks.
241 2011-10-31 04:37:08 <batouzo> cjdelisle: I did said 300 k ?
242 2011-10-31 04:37:15 <gmaxwell> 4x the peak chain growth rate of bitcoin but a fraction of the cost to mine blocks.
243 2011-10-31 04:39:01 <batouzo> cjdelisle: I don't have a botnet lol, and you come up with some huge number of CPUs. I will have access to less then that. Anyway what this has to do with pool software
244 2011-10-31 04:39:36 <gmaxwell> batouzo: there isn't much reason to do pooling on litecoin.. just a couple computers will get you several blocks per day solo mining it.
245 2011-10-31 04:40:12 <batouzo> gmaxwell: so what hashing power are you getting on it?
246 2011-10-31 04:41:02 <batouzo> "hashespersec" : 1255,    "networkhashps" : 5435515  ... do I read it correct that this is 0.0002 part of network power
247 2011-10-31 04:41:06 <cjdelisle> hehe
248 2011-10-31 04:41:13 <gmaxwell> wow, thats slow.
249 2011-10-31 04:41:14 <cjdelisle> 95.20.54.178.in-addr.arpa domain name pointer unallocated.sta.synapse.net.ua. <-- discovered bittorrent node
250 2011-10-31 04:41:40 <batouzo> gmaxwell: that is default compilation from git source running on around i5 quadcore or something
251 2011-10-31 04:42:16 <batouzo> what power do you get?
252 2011-10-31 04:42:24 <cjdelisle> What will be interesting is litecoin on an APU
253 2011-10-31 04:43:30 <gmaxwell> batouzo: thats busted. I think I'm getting about 2k/core.
254 2011-10-31 04:44:55 <gmaxwell> And I only bothered starting it on two 32 way systems and I've got about a percent or two of the network.
255 2011-10-31 04:45:19 <gmaxwell> (hard to measure, it's only running when they're otherwise idle and I normally keep them cranking)
256 2011-10-31 05:00:50 <phantomcircuit> <Element {http://schemas.datacontract.org/2004/07/Dwolla.Lib.Services.DataContracts}Amount at 0x17aa500>
257 2011-10-31 05:00:57 <phantomcircuit> facepalm
258 2011-10-31 05:18:52 <Diablo-D3> gmaxwell: does oggpack have a maximum buffer limit for a write context?
259 2011-10-31 05:19:54 <gmaxwell> Diablo-D3: do you mean "is there a maximum packet size" or "is there a maximum time between calls to page out"?
260 2011-10-31 05:20:25 <Diablo-D3> no, I mean, Im using oggpack, and my app is segfaulting during a call to realloc inside of oggpack_write
261 2011-10-31 05:22:13 <gmaxwell> Oh you're actually using oggpack? don't do that... but I don't recall any limit other than only being able to write 32bits at a time.
262 2011-10-31 05:22:30 <Diablo-D3> well, I just wanna molest individual bits
263 2011-10-31 05:22:46 <Diablo-D3> and I dont want to write my own bit stream handler
264 2011-10-31 05:23:01 <Diablo-D3> I wonder what endbyte in oggpack_buffer is
265 2011-10-31 05:23:10 <Diablo-D3> or storage
266 2011-10-31 05:23:29 <Diablo-D3> because its 1532 and 1536 respectively and if thats in bytes, wtf
267 2011-10-31 05:23:38 <gmaxwell> The only thing that uses the oggpack stuff is vorbis, and we generally regard putting that bitpacker in libogg as a mistake.
268 2011-10-31 05:23:45 <Diablo-D3> gmaxwell: =/
269 2011-10-31 05:23:53 <Diablo-D3> but I dont want to write my own
270 2011-10-31 05:25:06 <gmaxwell> (and it's also not particularly good, or well documented and I've long swapped out how it works.  If you go into #vorbis and ask Monty will probably help you when he stops by to visit earth)
271 2011-10-31 05:25:39 <Diablo-D3> it looks well documented to me =/
272 2011-10-31 05:26:11 <Diablo-D3> https://www.xiph.org/ogg/doc/libogg/bitpacking.html
273 2011-10-31 07:03:08 <shurdeek> anyone knows what happened in block 124725? It looks like 0.01000001 Bitcoins where lost.
274 2011-10-31 07:10:51 <[Tycho]> shurdeek, which TX ?
275 2011-10-31 07:11:17 <shurdeek> lemme check
276 2011-10-31 07:18:36 <CIA-101> bitcoinjs/node-bitcoin-p2p: Stefan Thomas master * rdeee35e / lib/scriptinterpreter.js : Removed debug statements. - http://git.io/dNxz6g
277 2011-10-31 07:28:34 <shurdeek> [Tycho], actually it's 124724, tx 5d80a29be1609db91658b401f85921a86ab4755969729b65257651bb9fd2c10d
278 2011-10-31 07:32:41 <phantomcircuit> shurdeek, they paid a fee of 0.00000001
279 2011-10-31 07:33:41 <shurdeek> phantomcircuit, who paid the fee? when I look at it in abe explorer, it says generation 50, fee 0.01, and result is 49.99999999
280 2011-10-31 07:34:27 <phantomcircuit> someone moved 50 generated bitcoins and paid a fee of 0.00000001 to do so
281 2011-10-31 07:34:33 <shurdeek> sorry, it says Generation: 49.98999999 + 0.01 total fees
282 2011-10-31 07:36:51 <shurdeek> it makes no sense to pay the fee on generation, since the one who generates it reaps the fee anyway.
283 2011-10-31 07:37:45 <shurdeek> could that bee a coding error or something like that?
284 2011-10-31 07:38:08 <cjdelisle> I'm pretty sure that was a mistake or someone being silly, I remember some discussion about it before
285 2011-10-31 07:38:51 <shurdeek> ok. There are some other examples like that, tiny amounts going missing.
286 2011-10-31 07:45:38 <shurdeek> tx phantomcircuit, [Tycho], cjdelisle
287 2011-10-31 07:48:03 <[Tycho]> Hmm, nice TX :)
288 2011-10-31 07:50:17 <[Tycho]> I wonder why it says 0.00000001 total fees when it should be 0.01000001
289 2011-10-31 07:50:29 <[Tycho]> Or 0.01
290 2011-10-31 07:52:20 <[Tycho]> Who gets the fee from http://blockexplorer.com/tx/dc72b9fe114788ec367ddbce2a47884afd147e2d02e1cb91b9c518f348103673 then ?
291 2011-10-31 07:52:45 <kinlo> I guess it's lost?
292 2011-10-31 07:52:49 <kinlo> if nobody has it...
293 2011-10-31 07:53:03 <[Tycho]> Seems so, but strange anyway.
294 2011-10-31 07:53:33 <[Tycho]> May be it's someone's custom client. Who mined that block &
295 2011-10-31 07:53:35 <[Tycho]> ?
296 2011-10-31 07:55:22 <[Tycho]> So that's a new way to "nullroute" coins :)
297 2011-10-31 07:55:51 <DrHaribo> no generate tx - someone threw away 50 BTC?
298 2011-10-31 07:56:06 <[Tycho]> Looks like it.
299 2011-10-31 07:56:42 <edcba> isn't that mandatory ?
300 2011-10-31 07:56:52 <[Tycho]> 50 BTC ? No.
301 2011-10-31 07:57:02 <kinlo> there is a generate tx, just not for the full 50 BTC+fee's
302 2011-10-31 07:57:33 <[Tycho]> I mean generate tx without outputs.
303 2011-10-31 07:57:58 <kinlo> which block does that?
304 2011-10-31 07:58:07 <[Tycho]> http://blockexplorer.com/block/0000000000004c78956f8643262f3622acf22486b120421f893c0553702ba7b5
305 2011-10-31 07:58:22 <[Tycho]> There is one output, but less than 50
306 2011-10-31 07:58:47 <[Tycho]> May be he tried to test it - that's why only 0.00000001 is missing.
307 2011-10-31 07:58:51 <DrHaribo> looks like rounding error in someone's software ;)
308 2011-10-31 07:59:10 <[Tycho]> No, more like controlled experiment.
309 2011-10-31 07:59:18 <edcba> i doubt that
310 2011-10-31 07:59:19 <DrHaribo> hmm or not.. there's only 1 output. Yes probably an experiment.
311 2011-10-31 07:59:34 <kinlo> yeah, but still one output
312 2011-10-31 07:59:34 <[Tycho]> Otherwise it should be 50.00999999
313 2011-10-31 07:59:37 <edcba> you need not to experiment on real network
314 2011-10-31 07:59:54 <kinlo> there is a test network for such things
315 2011-10-31 08:00:07 <kinlo> I think it's just a bug
316 2011-10-31 08:00:29 <edcba> and why would you test that anyway ?
317 2011-10-31 08:00:29 <kinlo> there are people experimenting with other software generating blocks instead of the regular bitcoin client
318 2011-10-31 08:00:53 <[Tycho]> Well, at least there is a bug in BBE
319 2011-10-31 08:01:54 <DrHaribo> What's BBE?
320 2011-10-31 08:02:03 <[Tycho]> blockexplorer.com
321 2011-10-31 08:03:05 <DrHaribo> maybe the bitcoin program should not accept blocks where inputs and outputs don't add up
322 2011-10-31 08:03:16 <[Tycho]> It says "Generation: 50 + 0.00000001 total fees" instead of "Generation: 49.98999999 + 0.01 total fees"
323 2011-10-31 08:03:20 <DrHaribo> or is throwing away coins a feature?
324 2011-10-31 08:03:22 <[Tycho]> DrHaribo, why ?
325 2011-10-31 08:03:34 <[Tycho]> I think that this should be accepted.
326 2011-10-31 08:04:24 <shurdeek> i use abe explorer instead of bbe, and it lists it correctly
327 2011-10-31 08:04:31 <DrHaribo> I don't see the purpose with having coins disappear this way.
328 2011-10-31 08:04:34 <[Tycho]> Do you have a link ?
329 2011-10-31 08:04:52 <edcba> hmm maybe a new format for tx would be handy
330 2011-10-31 08:05:01 <[Tycho]> DrHaribo, that's not a reason for wasting the entire block.
331 2011-10-31 08:05:11 <edcba> ie specify inputs as full number, specify outputs as fraction
332 2011-10-31 08:05:24 <edcba> would allow to gain space
333 2011-10-31 08:05:45 <[Tycho]> shurdeek, do you have a link ?
334 2011-10-31 08:06:01 <shurdeek> just a sec
335 2011-10-31 08:07:28 <shurdeek> http://abe.john-edwin-tobey.org/block/0000000000004c78956f8643262f3622acf22486b120421f893c0553702ba7b5
336 2011-10-31 08:08:03 <[Tycho]> Heh, correct numbers :)
337 2011-10-31 08:08:13 <[Tycho]> But site design is worse.
338 2011-10-31 08:08:57 <DrHaribo> Funny how blockexplorer shows the correct fee in the fee column for the block.
339 2011-10-31 08:09:23 <shurdeek> well you can pull the code for bitcoin-abe and run it yourself, and it also calculates bitcoin days destroyed.
340 2011-10-31 08:13:03 <edcba> only font is worse imo
341 2011-10-31 08:14:20 <edcba> i think transactions should not be shown as addresses but as "Address 1", "Address 2" and have address as tooltips
342 2011-10-31 08:14:26 <edcba> should be easier to read then
343 2011-10-31 08:14:45 <edcba> you could easily see which one is reused etc
344 2011-10-31 08:15:14 <edcba> hashes as such are quite meaningless
345 2011-10-31 08:21:01 <DrHaribo> Is it ok to use a hash of pubkey for generate outputs, like other transaction outputs, instead of the actual pubkey like generate outputs normally use?
346 2011-10-31 08:21:29 <Diablo-D3> no.
347 2011-10-31 08:23:42 <DrHaribo> Why not? :)
348 2011-10-31 08:35:39 <edcba> // First transaction must be coinbase, the rest must not be
349 2011-10-31 08:35:44 <edcba> i knew it was mandatory...
350 2011-10-31 08:43:55 <gmaxwell> 01:58 < [Tycho]> May be he tried to test it - that's why only 0.00000001 is missing.
351 2011-10-31 08:44:04 <gmaxwell> That was midnightmagic being cute.
352 2011-10-31 08:44:33 <gmaxwell> He was trying to make 1 satoshi not come into existance to symbolize the absense of satoshi from the project.
353 2011-10-31 08:44:39 <gmaxwell> But he screwed up and lost the fee too.
354 2011-10-31 08:44:51 <gmaxwell> Thats not the only coinbase lost forever.
355 2011-10-31 08:45:04 <Diablo-D3> yes that
356 2011-10-31 08:45:17 <gmaxwell> edcba: and you can happily take zero of the coinbase output ... or mine duplicated coinbases.
357 2011-10-31 08:47:55 <edcba> yes but i mean the transaction is still mandatory
358 2011-10-31 09:05:34 <GMP> i got an idea, i wonder if its impossible for some reason before i investigate it.
359 2011-10-31 09:05:40 <GMP> Lets say i have alternate blockchain DATACOIN, where coins represent bytes/bandwidth/upload/download like torrent trackers.. or some other strange blockchain (not money)
360 2011-10-31 09:05:51 <GMP> The idea: to make DTC<>BTC exchange P2P i use special type of bitcoin addresses:
361 2011-10-31 09:06:02 <GMP> Pubkey of normal address is 65 bytes is encoded (0x04,X,Y) - both coordinates of point on elliptic curve
362 2011-10-31 09:06:20 <GMP> special address will be 65 bytes too but (0x00-0x02,X,hash(transaction in DTC chain)) where 0-2 tells which root of cubic equation used to get Y from X (thats openssl feature)
363 2011-10-31 09:06:40 <GMP> exchange will be essentially one atomic operation (one party receives BTC, other receives DTC)
364 2011-10-31 09:06:54 <GMP> DTC transaction considered valid only when corresponding BTC transaction made it to blockchain
365 2011-10-31 09:08:28 <gmaxwell> There is no need to introduce point compression in bitcoin. The transaction format is not so constrained, and even if it were what you're proposing would change it anyways.
366 2011-10-31 09:08:51 <gmaxwell> Also, transactions aren't normally directed to public keys.
367 2011-10-31 09:09:54 <edcba> ok it's still the piggybacking on bitcoin chain problem anyway
368 2011-10-31 09:10:05 <edcba> i think we have already some litterature about it :)
369 2011-10-31 09:11:02 <edcba> i'd like have some timestamping service separated from bitcoin but i don't think there is some nice way to do it
370 2011-10-31 09:11:21 <gmaxwell> GMP: what you can also do is secure the transactions with additional secret keys, e.g. I make a BTC transaction to you secured with H(A) and H(B) (and a pubkey), I know A and you know B and gave me H(B)  and you do the opposite on your other chain.
371 2011-10-31 09:11:37 <gmaxwell> Then the txn are both stuck until A,B have been made public.
372 2011-10-31 09:11:54 <gmaxwell> edcba: thats what merged mining is for.
373 2011-10-31 09:13:41 <edcba> i don't remember what merged mining was but i doubt it convinced me :)
374 2011-10-31 09:13:55 <gmaxwell> oy.
375 2011-10-31 09:15:04 <gmaxwell> edcba: it allows other block-POW things to share hashing effort with bitcoin mining, without adding a bunch of junk to the bitcoin blockchain.
376 2011-10-31 09:15:05 <GMP> gmaxwell: transaction will be directed to hash160 as normal, the recipient will have to use spetial pubkey late, when spending
377 2011-10-31 09:15:23 <edcba> i'd like some chain where you have many (serviceid,hash) with bitcoin having some fixed serviceid and hash its merkleroot
378 2011-10-31 09:15:23 <GMP> s/late/later/
379 2011-10-31 09:15:59 <edcba> ie a chain shared by many programs
380 2011-10-31 09:16:16 <edcba> like programs share tcp on one computer
381 2011-10-31 09:17:05 <gmaxwell> edcba: but then those things would have to share difficulty, time between blocks, etc. A lot of degrees of freedom removed, plus everyone mining root would need to work all the systems.
382 2011-10-31 09:17:21 <gmaxwell> Merged mining doesn't have those limits.
383 2011-10-31 09:17:38 <edcba> share difficulty is not a problem :)
384 2011-10-31 09:17:58 <edcba> of course time will have to be reduced eventually
385 2011-10-31 09:18:14 <gmaxwell> It sure is if the result is that everyone mining the common system must participate in all networks and if they don't less popular ones get starved of blocks.
386 2011-10-31 09:18:17 <edcba> and i don't understand your point about mining
387 2011-10-31 09:18:30 <gmaxwell> Reduced?? Thats crazy.
388 2011-10-31 09:18:39 <edcba> you don't have to support all networks
389 2011-10-31 09:19:05 <edcba> programs report what they want to be included to miner, miner includes all of them
390 2011-10-31 09:19:32 <edcba> now each program may or may not have some merkleroot on one block
391 2011-10-31 09:19:38 <gmaxwell> edcba: and when they 'report' an rules violating block on chain X what happens?
392 2011-10-31 09:19:58 <edcba> the block is ignored for that program
393 2011-10-31 09:20:03 <edcba> but the chain is still good
394 2011-10-31 09:20:15 <GMP> gmaxwell: "introduce point compression", isnt it already possible? now! or is there something that prevents it?
395 2011-10-31 09:20:55 <gmaxwell> edcba: so what happens when 99.999% of the reports for service X to the miners are garbage?
396 2011-10-31 09:22:05 <gmaxwell> edcba: in any case, don't waste my time discussing this general subject until you've bothered to go read up on merged mining. I think it does most of what you want, and I suspect it does it better than whatever you're thinking of.
397 2011-10-31 09:22:14 <edcba> then majority of cpu voted that service X is crap :)
398 2011-10-31 09:22:43 <edcba> gmaxwell: it introduces dependancy to whole bitcoin
399 2011-10-31 09:22:45 <gmaxwell> GMP: What do you mean? No. There is no ability on the bitcoin network today to do the operation you're describing.
400 2011-10-31 09:22:53 <gmaxwell> edcba: What does?
401 2011-10-31 09:22:59 <edcba> merged mining
402 2011-10-31 09:23:04 <gmaxwell> No sir, it doesn't.
403 2011-10-31 09:23:22 <gmaxwell> bitcoin can die dead as a rock and chains which are merged mining with bitcoin would keep going fine.
404 2011-10-31 09:23:34 <edcba> a bitcoin chain must validate transaction in its blocks
405 2011-10-31 09:23:49 <edcba> that means a client sharing the chain must share validating transaction code
406 2011-10-31 09:23:53 <gmaxwell> Yep. Good thing merged mining keeps the chains completly seperate.
407 2011-10-31 09:24:01 <gmaxwell> Only if it cares to make blocks in that chain.
408 2011-10-31 09:24:29 <edcba> if you don't you don't share the cpu power used in the other chain
409 2011-10-31 09:24:35 <edcba> so it would be useless
410 2011-10-31 09:24:47 <GMP> gmaxwell: openssl automatically accepts both encodings, and i dont see anything that checks for 0x04 in sources
411 2011-10-31 09:24:54 <edcba> i think i begin to remember merged mining
412 2011-10-31 09:25:03 <gmaxwell> edcba: it sounds like you don't understand it at all.
413 2011-10-31 09:26:30 <gmaxwell> GMP: Go try it out on testnet. If its accepted we have a bug we need to eliminate.
414 2011-10-31 09:28:06 <GMP> why do you think its a bug? more like harmless but very useful feature
415 2011-10-31 09:28:07 <gmaxwell> edcba: in any case, the consequence of merged mining is that you can have N merged chains. You can mine on any M of them, and you only need to do rules validation for the M you're interested in.  You do one unit of work, and get a candidate hash that works on all M. Any N-M remainders your hashes wouldn't be good on.
416 2011-10-31 09:28:58 <gmaxwell> GMP: because certicom goes around _agressively_ claiming to have patents on transmission of public keys in point compressed form. This is why redhat strips all ECC code from their openssl distribution.
417 2011-10-31 09:29:31 <gmaxwell> And it's also not an useful feature because its better to not transmit the public key at all. It can be recovered from the signature.
418 2011-10-31 09:29:46 <edcba> merged mining add some hash in bitcoin transactions ?
419 2011-10-31 09:30:19 <gmaxwell> edcba: it adds a single sha256 merkle root in the coinbase, yes.
420 2011-10-31 09:32:19 <edcba> so you need bitcoin transactions code in your other program...
421 2011-10-31 09:32:25 <gmaxwell> No.
422 2011-10-31 09:32:52 <edcba> how do you add it then ?
423 2011-10-31 09:33:36 <Diablo-D3> by dipping a unicorn horn into the fruit juice.
424 2011-10-31 09:33:47 <gmaxwell> If you think of the normal bitcoin setup as having a header... H=[timestamp,prev,h(root),nonce].. for the merged chains, merged mining changes them to accept H0=[garbage,h(H1||garbageA),garbage],H1=[garbage;MM=H(H)] and H=normal header.
425 2011-10-31 09:34:20 <gmaxwell> The miner provides the H0,H1 values which are mostly garbage and ignored to the merged chain, though it checks the hash of H0 to measure the difficulty
426 2011-10-31 09:34:34 <gmaxwell> Plus it provides the concatinated garbage in the hashes.
427 2011-10-31 09:35:16 <gmaxwell> If the miner is also mining bitcoin then H0 is the bitcoin header, H1 is the coinbase txn.  But they don't have to be mining bitcoin too.. if they aren't then they can just use the rest as extra nonce space or set it to zeros or whatever.
428 2011-10-31 09:35:29 <gmaxwell> Nothing in the merged chain parses anything but the cascade of hashes.
429 2011-10-31 09:35:33 <Diablo-D3> gmaxwell: you know, audio codecs suck dick.
430 2011-10-31 09:35:39 <Diablo-D3> just throwing that out there.
431 2011-10-31 09:35:45 <edcba> ok i see the explanation i read is not good and what you call coinbase isn't coinbase either i think
432 2011-10-31 09:36:07 <edcba> so you just hash bitcoin block chain and add it to your chain
433 2011-10-31 09:36:14 <gmaxwell> No.
434 2011-10-31 09:36:18 <edcba> so you don't care about coinbase but merkle root ?
435 2011-10-31 09:37:34 <gmaxwell> edcba: You care about the H0 and H1. From which you extract hashes which are useful to you. (and on H0 you measure difficulty by H(H0))
436 2011-10-31 09:37:59 <gmaxwell> But you don't care about them except as blobs of garbage that happen to contain useful hash values.
437 2011-10-31 09:38:12 <gmaxwell> Which you use to connect down the the actual header for your chain.
438 2011-10-31 09:39:46 <edcba> ok i think i see i won't enter in details now assuming it work for namecoin anyway
439 2011-10-31 09:40:25 <edcba> so all is good you don't need to have deep understanding of bitcoin to share the cpu power but...
440 2011-10-31 09:40:46 <edcba> now we want to add some 3rd program
441 2011-10-31 09:41:05 <edcba> how do you chain that to take advantage of namecoin users and bitcoin users ?
442 2011-10-31 09:41:26 <gmaxwell> sure. There are already three things using merged mining, IIRC. it's simple.
443 2011-10-31 09:42:06 <gmaxwell> the value written into the bitcoinbase is just a hash root for a tree of connected merged things.
444 2011-10-31 09:42:52 <gmaxwell> So the miner gives the namecoin network the some opaque blobs (the bitcoin coinbase txn and header, but they don't know that or care) plus a tree fragment to connect their real header back through to the top blob.
445 2011-10-31 09:43:21 <gmaxwell> For some other service they'd give them the same blobs, plus a different tree fragment (and of course the proper header for that service and the matching block)
446 2011-10-31 09:44:34 <edcba> ok so to take advantage of all the services you need to form a chain of services
447 2011-10-31 09:44:58 <gmaxwell> There isn't a chain of services, just a hash tree.. same way transactions get connected to the root of the blocks in bitcoins.
448 2011-10-31 09:45:43 <gmaxwell> So for N services you'll have to tell each service about log2(N) hashes they'd need to walk through to connect their service header to the POW header.
449 2011-10-31 09:45:59 <edcba> ok bitcoin don't earn cpu power from namecoin users so ?
450 2011-10-31 09:46:35 <gmaxwell> Bitcoin doesn't earn power from namecoin only miners.
451 2011-10-31 09:46:42 <gmaxwell> Namecoin doesn't earm power from bitcoin only miners.
452 2011-10-31 09:46:52 <edcba> that's the problem
453 2011-10-31 09:46:55 <gmaxwell> But someone can mine both and do only one hash.
454 2011-10-31 09:47:02 <edcba> my solution would solve that
455 2011-10-31 09:47:09 <gmaxwell> No, your solution would just fail.
456 2011-10-31 09:47:23 <edcba> lol
457 2011-10-31 09:47:24 <gmaxwell> Because you can't 'gain power' from someone who isn't doing validation. Thats meaningless.
458 2011-10-31 09:47:48 <edcba> i don't understand
459 2011-10-31 09:47:54 <edcba> miners accept X services
460 2011-10-31 09:48:02 <gmaxwell> Thats like saying that I'm sexy because sean connery is sexy.  I have nothing to do with him!
461 2011-10-31 09:48:28 <edcba> they may require 1 BTC for including your hash for some service
462 2011-10-31 09:48:31 <edcba> or whatever
463 2011-10-31 09:48:34 <gmaxwell> edcba: if the miner isn't validating it then the hash power is worthless. If they are, then it's good and thats what you have with merged mining.
464 2011-10-31 09:48:46 <edcba> why it would be worthless ?
465 2011-10-31 09:49:11 <gmaxwell> Because it isn't actually a commitment to a distributed decision.
466 2011-10-31 09:49:13 <edcba> what is the problem with not validating ?
467 2011-10-31 09:49:32 <edcba> you have timestamped information
468 2011-10-31 09:49:51 <edcba> then bitcoin network may validate that information or not
469 2011-10-31 09:50:03 <edcba> i just decouple the timestamping part
470 2011-10-31 09:50:16 <gmaxwell> Why have bitcoin miners validate?
471 2011-10-31 09:50:22 <edcba> now you can put a price on that timestamping service if you want
472 2011-10-31 09:50:32 <edcba> no miners won't validate
473 2011-10-31 09:50:36 <edcba> just regular users
474 2011-10-31 09:50:45 <edcba> miners just hash :)
475 2011-10-31 09:50:50 <edcba> they hash data
476 2011-10-31 09:51:03 <edcba> each service validate their data then
477 2011-10-31 09:51:08 <gmaxwell> There is nothing stopping you from just putting whatever you want in the MM hash tree for commitments to things you're not validating.
478 2011-10-31 09:51:31 <edcba> so then they will be invalidated
479 2011-10-31 09:51:47 <edcba> free for miners to decide what service is trash
480 2011-10-31 09:52:12 <edcba> that's how bitcoin works too
481 2011-10-31 09:52:31 <edcba> free for miners to gen blocks without any useful transactions
482 2011-10-31 09:53:10 <gmaxwell> Well, not necessarily, other miners could choose to refuse to extend blocks that don't contain enough common memory pool txn.
483 2011-10-31 09:53:16 <gmaxwell> Gavin was proposing that for testnet recently.
484 2011-10-31 09:53:54 <edcba> you could have some feedback from services to tell which data is valid or not anyway
485 2011-10-31 09:54:12 <gmaxwell> yes, but running the service. Tada.
486 2011-10-31 09:54:28 <edcba> but that would be indeed useless
487 2011-10-31 09:54:40 <gmaxwell> Why is that useless?
488 2011-10-31 09:55:05 <edcba> anyway enough common memory pool txn is not valid
489 2011-10-31 09:55:12 <edcba> you can't tell identities on bitcoin
490 2011-10-31 09:55:21 <edcba> so i could gen blocks only useful to me
491 2011-10-31 09:55:39 <gmaxwell> edcba: But I know what transactions I have in my memorypool.
492 2011-10-31 09:56:00 <gmaxwell> I may even know the identity of many of their authors due to agreements I have in place.
493 2011-10-31 09:56:33 <gmaxwell> I could reconize anti-social blocks onces which inexplicably exclude too many of the available transactions and replace them with others that I wouldn't have included.
494 2011-10-31 09:57:23 <gmaxwell> You can't force me to extend your block and if a majority makes a common decision to ignore it then it will be wasted effort.
495 2011-10-31 09:58:16 <edcba> i should propose my idea to miners
496 2011-10-31 09:58:39 <edcba> pools i mean :)
497 2011-10-31 09:58:56 <gmaxwell> edcba: I think you should take the time to understand other things completely before proposing new things.
498 2011-10-31 09:59:26 <edcba> i think you don't see enough globally
499 2011-10-31 10:00:17 <gmaxwell> Because I think proposals that require every other service to know and validate the bitcoin blockchain are stupid on their face?
500 2011-10-31 10:01:00 <edcba> my proposal don't require other service to know anything about other services
501 2011-10-31 10:01:18 <gmaxwell> edcba: how do they know the timestamp is valid?
502 2011-10-31 10:01:52 <edcba> the only thing common to services are the block chain loosely defined as bitcoin chain
503 2011-10-31 10:02:07 <edcba> it just doesn't go down to the merkleroots
504 2011-10-31 10:02:14 <edcba> it's just data for that
505 2011-10-31 10:02:31 <edcba> you only have some timestamp proof of work and data
506 2011-10-31 10:03:27 <gmaxwell> So when I mine in invalid bitcoin block which won't survive it will trick your service unless your service waits several blocks, no?
507 2011-10-31 10:03:57 <gmaxwell> And the service will still need to know the complete history of bitcoin in order to apply the difficulty and time rules
508 2011-10-31 10:04:10 <edcba> you may mine some invalid bitcoin block put the hash in the timestamping chain, it will just make bitcoin network wait for another valid block later
509 2011-10-31 10:04:33 <gmaxwell> And if someone gives miners invalid values for your service, which they can't identify because they're not running it, you'll be trivially DOSed until the next block comes... and the next and the next.
510 2011-10-31 10:05:00 <gmaxwell> edcba: I can put a valid $service hash in an invalid bitcoin block. It will look valid to the service but the bitcoin block will not survive.
511 2011-10-31 10:06:33 <edcba> bitcoin will necessarily be advantaged on the timestamping service since ppl are greedy
512 2011-10-31 10:06:33 <gmaxwell> (soon enough I'll have you thinking you were proposing merged mining all along)
513 2011-10-31 10:07:01 <edcba> so they won't waste cpu power for nothing
514 2011-10-31 10:07:02 <gmaxwell> Hm? who says bitcoin are worth anything at all?
515 2011-10-31 10:07:14 <edcba> miners
516 2011-10-31 10:07:20 <edcba> since they use cpu for it :)
517 2011-10-31 10:07:48 <gmaxwell> no, you're missing the point you're making so that if bitcoin loses popularity all those dependant services must die just because no one cares about bitcoin anymore.
518 2011-10-31 10:08:14 <edcba> if ppl have interest on other non bitcoin program they will mine too
519 2011-10-31 10:08:37 <edcba> so the DoSing problem is really a popularity problem
520 2011-10-31 10:08:42 <edcba> voting with cpu
521 2011-10-31 10:08:42 <gmaxwell> But they can't not without mining bitcoin, because they'll produce invalid blocks that don't survive otherwise.
522 2011-10-31 10:08:58 <gmaxwell> And people can DoS by doing so intentionally.
523 2011-10-31 10:09:50 <edcba> hmm i'm not sure if you understand that timestamping service would 'run' parallel programs at same time
524 2011-10-31 10:10:06 <edcba> ie timestamped data = (bitcoin, hash) + (namecoin, hash) + ...
525 2011-10-31 10:23:01 <diki> why exactly are we skipping from 0.4.0 to directly 0.5?
526 2011-10-31 10:24:52 <cjdelisle> 1.4.1 is a bugfix version
527 2011-10-31 10:25:56 <diki> 1.4 seems quite far in the future
528 2011-10-31 10:27:20 <Diablo-D3> mrb_: wait hold up
529 2011-10-31 10:27:34 <Diablo-D3> that italian actually managed to make it work and be net positive?
530 2011-10-31 10:28:37 <tcatm> diki: because 0.4 -> 0.5 is a rather big change (new GUI)
531 2011-10-31 10:29:31 <gmaxwell> 04:09 < edcba> hmm i'm not sure if you understand that timestamping service would 'run' parallel programs at same time