1 2012-03-01 00:00:05 <da2ce7> maybe put 10btc into the gavinandresen is great fund.
  2 2012-03-01 00:00:44 <BlueMatt> 10 btc is going to motivate gain to do what?
  3 2012-03-01 00:00:51 <BlueMatt> thats what, 50 usd?
  4 2012-03-01 00:01:06 <BlueMatt> and what do you think his time is actually worth? 50 usd will buy barely any time at all
  5 2012-03-01 00:01:26 <BlueMatt> realistically, bounties in oss will never have enough money in them to make it comparable to working
  6 2012-03-01 00:01:36 <BlueMatt> and at that point, you are just hiring programmers
  7 2012-03-01 00:03:59 <gmaxwell> BlueMatt: yea, thats the problem.. little amounts of money aren't very helpful to do something you need to to have income to replace someone's dayjob for long enough spans to make it worth doing. :(
  8 2012-03-01 00:04:19 <da2ce7> BlueMatt, it isn't 'to pay for time' rarther a nice way to say 'we care about ya'
  9 2012-03-01 00:04:27 <BlueMatt> yea, not that I dont think its cool to pay oss devs, but seriously
 10 2012-03-01 00:04:56 <BlueMatt> da2ce7: yea, but the easiest way to say we care about what you are doing is to help out, not give money ;)
 11 2012-03-01 00:05:23 <da2ce7> sure, but some people are too busy;
 12 2012-03-01 00:05:39 <BlueMatt> /nod
 13 2012-03-01 00:05:45 <Staatsfeind> s/busy/ignorant/
 14 2012-03-01 00:05:48 <BlueMatt> anyway, Ill be back later
 15 2012-03-01 00:06:24 <da2ce7> well if people think that it is rude... then I won't offer... just tryin to be a lil supportive.
 16 2012-03-01 00:08:27 <etotheipi_> da2ce7, you can always donate to *my* crowdfunding campaign:  https://bitcointalk.org/index.php?topic=64449;all
 17 2012-03-01 00:08:40 <etotheipi_> ya know... if you feel like supporting OSS developers ;)
 18 2012-03-01 00:08:56 <etotheipi_> you get rewards, and I will go part time at my real job to work on Armory more :)
 19 2012-03-01 00:09:19 <etotheipi_> plus guarantee that Armory stays open-source
 20 2012-03-01 00:09:21 <da2ce7> :)
 21 2012-03-01 00:09:44 <etotheipi_> it's worth donating $10 just to get my encryption seminar:  "Understanding Cryptography:  Using Boring Math for Something Useful"
 22 2012-03-01 00:09:55 <da2ce7> yep... I wanna copy of that.
 23 2012-03-01 00:10:04 <da2ce7> can you pm me a bitcoin address.
 24 2012-03-01 00:10:04 <etotheipi_> I've heard $10 is a good deal :)
 25 2012-03-01 00:10:22 <etotheipi_> it's on that link
 26 2012-03-01 00:10:39 <etotheipi_> although, as I look at that thread, I realize that the top post has gotten *really* long
 27 2012-03-01 00:10:48 <da2ce7> sure but I'm using an online wallet, so I cannot sign a message from the key it is comeing from.
 28 2012-03-01 00:11:10 <etotheipi_> oh, don't worry about signing
 29 2012-03-01 00:11:35 <etotheipi_> that was only if there was going to be confusion about who donated...
 30 2012-03-01 00:13:00 <etotheipi_> to anyone else here:  not only is the encryption seminar available for $10, but I have a very-heavy-math introduction to quantum computing
 31 2012-03-01 00:13:52 <etotheipi_> it's 85 slides, which I will similarly release for $10 (but I wouldn't do it unless you have a good math background)
 32 2012-03-01 00:14:05 <da2ce7> :)
 33 2012-03-01 00:14:08 <da2ce7> coool
 34 2012-03-01 00:14:53 <da2ce7> etotheipi_: how about you do a booklet on lattice based encyption.
 35 2012-03-01 00:15:19 <etotheipi_> do you have a math background?  the QC slides are solid... at the very least you should be able to understand Grover's algorithm when you're done
 36 2012-03-01 00:15:29 <sipa> etotheipi_: hmm, tempting :)
 37 2012-03-01 00:16:11 <etotheipi_> take a pure-guessing problem where there are N possible answers, all equally likely, but only one of them is correct, the only way to find out is to guess
 38 2012-03-01 00:16:31 <etotheipi_> so on average it will take you N/2 guesses to find the right answer, O(N)
 39 2012-03-01 00:16:43 <sipa> sure :)
 40 2012-03-01 00:16:52 <gmaxwell> http://www.scottaaronson.com/thesis.html  < has a pretty good breif QC tutorial in it.
 41 2012-03-01 00:17:04 <etotheipi_> but if you can convert the guessing problem into a quantum circuit, then use qubits, you can get the answer in O(sqrt(N))
 42 2012-03-01 00:17:21 <sipa> etotheipi_: i know the result; i don't know how or why it works
 43 2012-03-01 00:18:20 <etotheipi_> sipa, not everyone may know it... and I think it's a very intriguing result to suck others into donating for the seminar :)
 44 2012-03-01 00:18:41 <gmaxwell> (well technically the 'quantum circuit' has to be a special oracle which has no waste so the cancellation works, which is actually a @#$@# pita, one of those messy things that get waved away because someone proved that you can convert anything into such an oracle with only a polynomial factor gate increase)
 45 2012-03-01 00:19:44 <etotheipi_> any classical circuit has an equivalent quantum circuit representation
 46 2012-03-01 00:20:25 <etotheipi_> and it turns out that any quantum circuit can be created from (possibly massive) numbers of just two circuits ("basis vectors" of quantum computing)
 47 2012-03-01 00:24:15 <gmaxwell> etotheipi_: er. No. It's not sufficient for it to _just_ be a quantum circuit. For grover to work it has to have a special form: The output must be exactly the input plus one bit (the match) and it must destroy no information in the process.
 48 2012-03-01 00:25:00 <gmaxwell> I don't know how you could have explained grover without actually making this point clear, because it's important for understanding the cancellation.
 49 2012-03-01 00:25:21 <etotheipi_> gmaxwell, it's been 3 years since I studied this
 50 2012-03-01 00:25:26 <gmaxwell> ah.
 51 2012-03-01 00:25:38 <etotheipi_> I'm actually re-reading the seminar now
 52 2012-03-01 00:26:25 <gmaxwell> In the case of functions like SHA256 where a normal implementation destroys a lot of information, the reversable form has exponential interior state growth... ends up being a ginormous circuit. I was never able to figure out if there is a way to prevent the exponential state growth.
 53 2012-03-01 00:27:43 <gmaxwell> I suppose there must be because there is a proof that you can turn any circuit into such a form with only a polynomial increase in size but the proof wasn't the sort that was helpful (to me at least) in figuring out how.
 54 2012-03-01 00:28:11 <etotheipi_> gmaxwell, it sounds like you studied Quantum computer much more than I did
 55 2012-03-01 00:28:24 <etotheipi_> I simply did a (very long) final project on it for my computational-complexity class
 56 2012-03-01 00:29:20 <etotheipi_> spent most of the semester teaching myself on it to produce this presentation... but I never had a formal class on it
 57 2012-03-01 00:29:22 <gmaxwell> well, I sat down and wrote some programs for QCs just to understand it.
 58 2012-03-01 00:29:32 <etotheipi_> ahh, I should've done that...
 59 2012-03-01 00:29:49 <etotheipi_> at least it's been proven that you can simulate a QC on a CC :)
 60 2012-03-01 00:30:03 <gmaxwell> there is software to do so in fact!
 61 2012-03-01 00:42:18 <sipa> i believe there's a haskell package to do so :)
 62 2012-03-01 00:54:49 <gavinandresen> All righty, that Mac build was painful... but rc2 binaries are up at   https://sourceforge.net/projects/bitcoin/files/Bitcoin/bitcoin-0.6.0/test/?
 63 2012-03-01 00:55:26 <diki> lol, 3 fast blocks
 64 2012-03-01 00:55:51 <diki> 3 blocks in under 2 minutes
 65 2012-03-01 00:58:31 <sipa> gavinandresen: painful? how so?
 66 2012-03-01 00:58:54 <gavinandresen> macports python dependency hell.
 67 2012-03-01 00:59:02 <sipa> joy
 68 2012-03-01 00:59:15 <gavinandresen> mmm....  sigh.
 69 2012-03-01 00:59:56 <gavinandresen> My life would be easier if we stopped supporting 32-bit OSX 10.5 machines.
 70 2012-03-01 07:56:12 <topi`> my bitcoind produces huge amount of db logs (10 M per logfile, 300+ logs) ... how do I turn that db log off?
 71 2012-03-01 08:44:16 <topi`> I suspect -printtoconsole does not affect those db logs
 72 2012-03-01 10:04:03 <Eliel> topi`: most of the logs get cleaned out if you restart bitcoind.
 73 2012-03-01 10:11:54 <topi`> Eliel: the problem is that I get too many writes to my flash filesystem
 74 2012-03-01 10:12:26 <topi`> e.g. flushing even a single line of text writes the whole block (0.5 MB or more)
 75 2012-03-01 10:29:11 <topi`> hmm, these database log files seem to be binary files
 76 2012-03-01 10:29:38 <Diablo-D3> guess what! they are!
 77 2012-03-01 10:30:00 <Diablo-D3> topi`: btw, what logs are you talking about?
 78 2012-03-01 10:30:08 <Diablo-D3> because I suspect you're talking about the database itself
 79 2012-03-01 10:30:36 <Diablo-D3> the files in ~/.bitcoin/database?
 80 2012-03-01 10:31:30 <Diablo-D3> those are part of the operation of BDB, they're not logs in the way you mean.
 81 2012-03-01 10:33:06 <topi`> aha
 82 2012-03-01 10:33:13 <topi`> damn berkeleydb
 83 2012-03-01 10:33:34 <topi`> I guess I should put the .bitcoin dir on a real hard disk.
 84 2012-03-01 10:33:39 <Diablo-D3> hey man, fail-safe data transactional integrity is nice
 85 2012-03-01 10:34:02 <Diablo-D3> most likely an incomplete write can be recovered if your box dies
 86 2012-03-01 11:08:50 <phantomcircuit> topi`, mount the bitcoin directory as a tmpfs and then periodically copy your wallet to disk
 87 2012-03-01 11:09:00 <phantomcircuit> (using the rpc call)
 88 2012-03-01 11:41:12 <lianj> if you have enough ram left, yes
 89 2012-03-01 12:10:04 <topi`> only 1 GB ram :/
 90 2012-03-01 12:12:41 <phantomcircuit> topi`, dont worry about it tmpfs will swap out
 91 2012-03-01 12:12:55 <phantomcircuit> you're effectively just disabling fsync calls
 92 2012-03-01 12:13:08 <phantomcircuit> so at worst you'll end up downloading the blocks again
 93 2012-03-01 12:13:10 <phantomcircuit> big deal
 94 2012-03-01 12:26:08 <topi`> but downloading takes *days* at worst
 95 2012-03-01 12:26:39 <topi`> somehow I thought only the blk0001.dat was important
 96 2012-03-01 12:27:02 <topi`> even the index to that file is huge. and an index could be rebuilt at every startup
 97 2012-03-01 12:40:07 <topi`> fuck, journal commit I/O error
 98 2012-03-01 12:42:51 <gmaxwell> topi`: the building the index is basically what make it take time.
 99 2012-03-01 12:44:28 <AIEmpire> Hello, not sure if this is the right place to ask, but I downloaded the new btc Mac client, the wallet is automatically encrypted, but I can't change the password, because I have no idea what the default password is.
100 2012-03-01 12:44:57 <gmaxwell> The wallet is not automatically encrypted.
101 2012-03-01 12:45:31 <gmaxwell> It is only encrypted if you ask it to be, an it gives you omnious warnings about not forgetting the password and makes you confirm in order to encrypt it.
102 2012-03-01 12:46:15 <AIEmpire> Well it is. Maybe it's a bug?
103 2012-03-01 12:46:32 <AIEmpire> I literally just downloaded it, run it for the first time and didn't change any settings.
104 2012-03-01 12:47:10 <topi`> there it goes. funny how Ctrl-c'ing the bitcoind ends up in a massive spam of I/O errors. A reminder that one should not run from a USB stick.
105 2012-03-01 12:48:09 <AIEmpire> Should I just delete the wallet.dat and restart the client?
106 2012-03-01 12:49:24 <etotheipi_> gmaxwell, sipa , how do you feel about something like this:  http://pastebin.com/wMzsddmE  (it's already implemented in Armory, but could be modified to be interoperable with the Satoshi client if it uses somethign different)
107 2012-03-01 12:54:06 <gmaxwell> AIEmpire: move the wallet.dat to another location. If that fixes it for you, I'd like a copy of that wallet.dat
108 2012-03-01 12:54:26 <phantomcircuit> topi`, try downloading with the database on a tmpfs you'll find it no longer takes days
109 2012-03-01 12:54:27 <phantomcircuit> magic
110 2012-03-01 12:54:29 <AIEmpire> Ok, I'll try it now
111 2012-03-01 12:54:38 <gmaxwell> etotheipi_: ...
112 2012-03-01 12:55:02 <gmaxwell> etotheipi_: Why don't you even bother looking at what the Satoshi client does?
113 2012-03-01 12:56:47 <gmaxwell> We use key recovery for the public key, so there is no reason for the signature to include it. Also your use of the "standard double-sha256-hashed" means that I can tell you to sign some junk which is effectively a transaction to spend all your coins, and you won't know any better and will do it?
114 2012-03-01 12:58:02 <etotheipi_> gmaxwell, I don't look because it's not released and I have never compiled it
115 2012-03-01 12:58:24 <etotheipi_> plus, this only signs ASCII text... no binary
116 2012-03-01 12:58:32 <gmaxwell> It _is_ released.
117 2012-03-01 12:58:42 <gmaxwell> It's in 0.5.
118 2012-03-01 12:58:51 <phantomcircuit> topi`, 47k
119 2012-03-01 12:59:05 <etotheipi_> it is?
120 2012-03-01 12:59:11 <etotheipi_> I thought it was being added 0.6
121 2012-03-01 12:59:13 <gmaxwell> (the gui for it is some later addition, but the RPC has been there since 0.5)
122 2012-03-01 12:59:22 <etotheipi_> oh, I've actually never used the Satoshi client RPC
123 2012-03-01 12:59:41 <etotheipi_> or maybe once when I was helping you guys debug my invalid tx
124 2012-03-01 12:59:52 <etotheipi_> err.. when you were helping me
125 2012-03-01 13:00:10 <gmaxwell> etotheipi_: so you'd fail if my name was Gr??g?
126 2012-03-01 13:00:28 <etotheipi_> gmaxwell, good question...
127 2012-03-01 13:01:25 <phantomcircuit> topace, 86k
128 2012-03-01 13:01:53 <etotheipi_> gmaxwell, this is a first cut at implementing a message signing interface... if there are ways to improve it, I will
129 2012-03-01 13:02:24 <etotheipi_> and with regards to the public key:  I said it in the pastebin:  I don't have it implemented yet, but the format supports it
130 2012-03-01 13:02:48 <etotheipi_> you can just add your extra byte to the sig and leave the public key off (it's just that,at the moment, Armory can't support it)
131 2012-03-01 13:05:02 <phantomcircuit> topi`, 113k
132 2012-03-01 13:08:56 <AIEmpire> gmaxwell: Yes, that fixed it. Not sure why the first wallet.dat was encrypted by default, but here are the files (there was a wallet.dat.rewrite as well).
133 2012-03-01 13:09:00 <AIEmpire> http://www.mediafire.com/?nqbklt8b5f8cp1h,nh1b898p518hibt
134 2012-03-01 13:09:47 <etotheipi_> so gmaxwell, how can I find out what the Satoshi client does/creates?  What RPC commands do I use to sign a message?
135 2012-03-01 13:09:56 <gmaxwell> etotheipi_: signmessage
136 2012-03-01 13:11:31 <gmaxwell> etotheipi_: What bitcoin actually signs is H("Bitcoin Signed Message:\n"+message) to avoid that.
137 2012-03-01 13:11:58 <gmaxwell> It takes an address and message and returns a base64 encoded signature with the extra key recovery bits.
138 2012-03-01 13:12:08 <phantomcircuit> topi`, 125k
139 2012-03-01 13:12:59 <etotheipi_> ahh, that's a good idea to force extra characters into the message.  It can be made transparent, too
140 2012-03-01 13:13:27 <etotheipi_> I thought about your concern, and assumed that the lack of being able to use binary in the interface was good enough
141 2012-03-01 13:14:22 <gmaxwell> etotheipi_: er, unless I misunderstand you, I don't see why making it transparent is helpful. The prefix is there to basically to convert the hash function into a different hash function.
142 2012-03-01 13:14:24 <phantomcircuit> gmaxwell, in general you dont want to be signing random shit people give you
143 2012-03-01 13:14:48 <gmaxwell> phantomcircuit: in general, sure but it sure shouldn't be possible to spend your coins that way.
144 2012-03-01 13:14:48 <phantomcircuit> that should probably be a random prefix which then is passed as part of the signature... but whatever
145 2012-03-01 13:15:13 <gmaxwell> phantomcircuit: That was discussed and bytecoin was strongly opposed to it.
146 2012-03-01 13:15:28 <phantomcircuit> bytecoin?