1 2013-02-20 00:00:51 <gjs278> geez... 86% cpu on 5 threads for bitcoind right now
2 2013-02-20 00:01:34 <sipa> gjs278: if you want less CPU usage, use less threads
3 2013-02-20 00:02:14 <gjs278> I see it now
4 2013-02-20 00:02:27 <sipa> it's intended to be fast, so it will use whatever you give it
5 2013-02-20 00:02:58 <gjs278> should I adjust dbcache, I have ram to blow if it will help
6 2013-02-20 00:31:08 <richardus> hi guys, is there an interface to the bitcoin dev mailing list that isn't the awful sourceforge one?
7 2013-02-20 00:31:14 <richardus> web interface, that is
8 2013-02-20 00:32:19 <sipa> your mail client? :)
9 2013-02-20 00:32:44 <richardus> you know, i added in the second line thinking that some smart ass would say that
10 2013-02-20 00:33:02 <sipa> it's mirrored on several places, but the SF one is the main one i think
11 2013-02-20 00:33:42 <richardus> oh it's on gmane, great
12 2013-02-20 00:34:13 <sipa> http://blog.gmane.org/gmane.comp.bitcoin.devel
13 2013-02-20 01:14:55 <muhoo> heh, satoshi dice runs on bitcoinj?
14 2013-02-20 01:29:09 <swhitt> the desktop client...
15 2013-02-20 01:29:22 <swhitt> which is why it has issues spending unconfirmed coins
16 2013-02-20 03:24:42 <MobGod> sipa when i install the new qt do i replace files or keep both?
17 2013-02-20 03:25:05 <MobGod> i would think keep both but just checking
18 2013-02-20 03:49:35 <jgarzik> Call for bittorrent testers... http://gtf.org/garzik/bitcoin/bootstrap.dat.torrent
19 2013-02-20 03:49:59 <Haifisch> jgarzik call received
20 2013-02-20 03:50:25 <Haifisch> ACTION picks up phone
21 2013-02-20 03:55:45 <jgarzik> firewall problem on this side
22 2013-02-20 03:55:48 <jgarzik> wait
23 2013-02-20 04:02:45 <randy-waterhouse> will 0.8 clean up extra copies of blockchain when it has re-indexed?
24 2013-02-20 04:03:11 <jgarzik> no
25 2013-02-20 04:03:25 <randy-waterhouse> which ones can be deleted then?
26 2013-02-20 04:19:20 <randy-waterhouse> so do we need to keep two copies of blockchain now? i.e. one in ~/.bitcoin and another in ~/.bitcoin/blocks/
27 2013-02-20 04:30:14 <gmaxwell> randy-waterhouse: no.
28 2013-02-20 04:30:25 <gmaxwell> randy-waterhouse: it creates a hardlink, you only have one copy.
29 2013-02-20 04:31:18 <gmaxwell> randy-waterhouse: and the filename in ~/.bitcoin is only needed so you can switch back to an old version.
30 2013-02-20 04:31:18 <randy-waterhouse> gmaxwell: thnx, hadn't noticed that
31 2013-02-20 04:31:40 <randy-waterhouse> ah, makes sense
32 2013-02-20 04:36:51 <randy-waterhouse> might be useful info in Gavin's notes on release page ... ?
33 2013-02-20 04:41:05 <jgarzik> Does this work? magnet:?xt=urn:btih:6fe493ba606847eac163baf35aae9db319735482&dn=bootstrap.dat&tr=udp://tracker.openbittorrent.com:80&tr=udp://tracker.publicbt.com:80&tr=udp://tracker.ccc.de:80&tr=udp://tracker.istole.it:80
34 2013-02-20 04:41:56 <randy-waterhouse> what does "Misbehaving : 173.242.112.53:8333 (0 -> 0) message indicate?
35 2013-02-20 04:46:13 <Haifisch> any java coders here?
36 2013-02-20 05:02:39 <ZenBalance> Does anyone know how to trouble shoot a curl: (7) couldn't connect to host error when trying to remotely connect to a bitcoind server via rpc?
37 2013-02-20 05:03:14 <Luke-Jr> don't do bitcoind remotely.
38 2013-02-20 05:06:02 <Luke-Jr> .. what is with all these BSD users suddenly?
39 2013-02-20 05:07:37 <gmaxwell> grarpamp has been around forever, and with the same not terribly useful bug reports about incorrect uninitilzied warnings from really old gccs.
40 2013-02-20 05:09:19 <ne0futur> ZenBalance: check your firewalls on the 2 sides ( client and server ), netstat could help too on the server side
41 2013-02-20 05:10:10 <ZenBalance> ah! Do I have to have a specific port on the computer I am curling from open?
42 2013-02-20 05:10:20 <ZenBalance> I am a little bit of a novice, so I am not too familiar with netstat
43 2013-02-20 05:10:50 <Luke-Jr> ne0futur: he's probably not using -rpcallowip <.<
44 2013-02-20 05:11:03 <Luke-Jr> ZenBalance: but seriously, it's a bad idea to remotely connect to bitcoind RPC
45 2013-02-20 05:11:15 <ZenBalance> why?
46 2013-02-20 05:11:25 <ZenBalance> I am using rpcallowip
47 2013-02-20 05:11:34 <ZenBalance> I can do the curl locally, just not remotely
48 2013-02-20 05:11:57 <ZenBalance> I was planning on having a front facing site that made remote calls to a bitcoind server on another machine. Is that poor design?
49 2013-02-20 05:12:08 <Luke-Jr> yes
50 2013-02-20 05:12:33 <ZenBalance> what would be a better one?
51 2013-02-20 05:12:41 <ne0futur> Luke-Jr: if the fireewall allows only one ip to connect . . . is it really bad ?
52 2013-02-20 05:12:47 <Luke-Jr> put the real wallet machine offline. encrypt its wallet. when you boot it, use walletpassphrase with some insanely huge timeout (so it never locks)
53 2013-02-20 05:13:09 <Luke-Jr> now generate 10000 addresses and put those in a database on the webserver
54 2013-02-20 05:13:25 <Luke-Jr> copy wallet.dat to the webserver now, and run bitcoind there too - but DON'T unlock it ever
55 2013-02-20 05:13:48 <Luke-Jr> ne0futur: IP-based security is pretty bad; and you're still transmitting your password clear
56 2013-02-20 05:14:09 <ne0futur> true
57 2013-02-20 05:14:46 <ne0futur> (07:59) < ZenBalanc> I am a little bit of a novice, so I am not too familiar with netstat
58 2013-02-20 05:14:52 <Luke-Jr> ZenBalance: if you do it that way, no matter how much your webserver gets compromised, your funds are safe on the offline machine
59 2013-02-20 05:14:56 <gmaxwell> ne0futur: e.g. someone who's compromised an upstream router from you??? or even perhaps another host on the same subnet??? can sniff then spoof and take all your coins.
60 2013-02-20 05:15:16 <ZenBalance> ah, okay. I think I understand most of that
61 2013-02-20 05:15:29 <ne0futur> ZenBalance: if I were you I wouldnt try to host a bitcoin service without having a good linux sysadmin first ;)
62 2013-02-20 05:15:39 <Luke-Jr> ZenBalance: when you want to spend them, you can decide whether to risk connecting the wallet machine to the net, or learn to use the raw transaction API to copy transactions over USB sticks or such
63 2013-02-20 05:15:54 <Luke-Jr> what ne0futur just said too ;)
64 2013-02-20 05:16:17 <Luke-Jr> ZenBalance: you might also take a serious look at BitPay and similar services - they do all this stuff for you
65 2013-02-20 05:16:22 <ZenBalance> hmm?where would be the best place to better familiarize myself with this?
66 2013-02-20 05:16:48 <Luke-Jr> ZenBalance: sadly, I think we're very lacking in good docs on all this :<
67 2013-02-20 05:16:50 <randy-waterhouse> first consultation free?
68 2013-02-20 05:16:57 <ZenBalance> I am doing this for a school project (so I don't plan on moving more than the smallest amount of BTC that I can) the point is to familiarize myself more with sysadmin stuff and encryption
69 2013-02-20 05:18:12 <Luke-Jr> gmaxwell: actually, has anyone we know audited BitPay's setup at all? Tony seems like a reasonably competent person, but Bitcoin stuff can be difficult to get right
70 2013-02-20 05:18:22 <ne0futur> ZenBalance: after 10 years familiarizing with sysadmin , I still dont feel strong enough to host a bitcoin service ;)
71 2013-02-20 05:18:33 <Luke-Jr> ZenBalance: I highly recommend using testnet then :P
72 2013-02-20 05:20:52 <ZenBalance> okay, cool. I'll look into that then. Still remains though, where would you guys recommend I start reading to learn more about the security involved?
73 2013-02-20 05:21:36 <andytoshi> ZenBalance: keep an eye on this channel, basically.. as luke says, we have a lack of docs
74 2013-02-20 05:21:42 <Luke-Jr> ZenBalance: sadly, to be competent in Bitcoin security, I think one needs to familiarize themselves with how it works at a very low level
75 2013-02-20 05:22:03 <Luke-Jr> and also be competent in ordinary security, which is the part I lack :/
76 2013-02-20 05:22:20 <Luke-Jr> (well, not that I'm incompetent in ordinary security, but I'm no security expert)
77 2013-02-20 05:23:30 <ZenBalance> hmm?then another question: Do you have any suggestions for some interesting projects that might get me more familiar with bitcoin without having to worry about too much risk?
78 2013-02-20 05:23:44 <Luke-Jr> ZenBalance: testnet!
79 2013-02-20 05:23:54 <Luke-Jr> ZenBalance: just run bitcoind with the -testnet option and it's all monopoly money
80 2013-02-20 05:24:26 <ZenBalance> Sorry, yes. Definitely using testnet, but what I meant is what simple projects might be a good way to start?
81 2013-02-20 05:24:33 <andytoshi> ZenBalance: it might be cool to implement probablistic transactions
82 2013-02-20 05:24:38 <ZenBalance> for instance, would a simple version of satoshi dice be hard to implement?
83 2013-02-20 05:24:40 <andytoshi> that'd force you to learn the low level stuff
84 2013-02-20 05:24:51 <andytoshi> ZenBalance: if you did it with probablistic transactions ??? ByronJohnson
85 2013-02-20 05:24:53 <Luke-Jr> ZenBalance: read bitcoin.pdf for sure
86 2013-02-20 05:24:57 <andytoshi> (sorry, copy/paste error)
87 2013-02-20 05:25:01 <ZenBalance> what do you mean by probabilistic transactions?
88 2013-02-20 05:25:08 <Luke-Jr> ZenBalance: read the source code until you understand what's actually going on under the hood
89 2013-02-20 05:25:09 <Luke-Jr> etc
90 2013-02-20 05:25:12 <andytoshi> one moment, i have a couple links..
91 2013-02-20 05:25:22 <ZenBalance> have read the pdf, only understand around 10-20% of it
92 2013-02-20 05:25:32 <andytoshi> here is my doc on probablistic transactions: http://download.wpsoftware.net/bitcoin/bitcoin-probablistic-payments.pdf
93 2013-02-20 05:25:53 <Luke-Jr> ZenBalance: you probably want to understand 100% of it
94 2013-02-20 05:26:18 <andytoshi> there is also a wiki page with links to the forum thread where the idea was first posted..i badly failed to cite on that PDF
95 2013-02-20 05:27:05 <wumpus> Luke-Jr: no, naming it satoshi client was not my idea, but like Sipa I prefer one name for the whole project
96 2013-02-20 05:27:34 <andytoshi> ZenBalance https://en.bitcoin.it/wiki/Nanopayments
97 2013-02-20 05:27:44 <wumpus> and I really don't like bitcoin-qt, it was a temporary working title while I was working on the "new gui", I expected it to go away when merged
98 2013-02-20 05:28:35 <wumpus> certainly not for people to start calling the entire project that...
99 2013-02-20 05:29:27 <ne0futur> ZenBalance: google is a good place to start searching about iptables -L and netstat
100 2013-02-20 05:29:30 <andytoshi> wumpus: haha
101 2013-02-20 05:30:29 <wumpus> no one gives a shit what gui rendering toolkit is used, it certainly shouldn't be part of the name of a program
102 2013-02-20 05:30:55 <andytoshi> wumpus: people gave a shit when it was "wx, but not the released wx"
103 2013-02-20 05:31:05 <wumpus> well developers give a shit but they have other way to find out like lookin at the source
104 2013-02-20 05:31:13 <wumpus> or the about menu for that sake
105 2013-02-20 05:31:38 <wumpus> really those two letter abbreviations like wx or qt don't tell users anything
106 2013-02-20 05:31:40 <ZenBalance> hmm?andytoshi, so basically probabilistic transactions are a way to implement really small transactions in a feasible manner
107 2013-02-20 05:31:48 <ZenBalance> this is really low level stuff, no?
108 2013-02-20 05:31:50 <ZenBalance> Pretty cool
109 2013-02-20 05:32:01 <Luke-Jr> wumpus: meh, do brands need to contain useful info?
110 2013-02-20 05:32:11 <ZenBalance> I'm not sure if I have the cryptography experience to implement something like that though
111 2013-02-20 05:32:44 <wumpus> Luke-Jr: maybe they don't, but adding a name of a ui library distracts from the underlying, financial part that is the focus imo
112 2013-02-20 05:33:19 <Luke-Jr> wumpus: yeah, probably want a name that sounds good for individuals but not big business or something
113 2013-02-20 05:33:30 <Luke-Jr> I still think a contest would be a good idea :p
114 2013-02-20 05:34:37 <wumpus> hehe, maybe
115 2013-02-20 05:35:20 <andytoshi> ZenBalance: yes, very low level
116 2013-02-20 05:35:31 <andytoshi> but the crypto stuff is wrapped up in the OpenSSL library for you
117 2013-02-20 05:35:50 <andytoshi> so you need to understand it on a high level, but only stuff you'd need to understand bitcoin anyway
118 2013-02-20 05:35:57 <andytoshi> maybe, this gives you something to shoot for
119 2013-02-20 05:37:30 <ZenBalance> okay. I'll look into it. How does one actually go about integrating probabilistic payments into the actual network (or testnetwork) though?
120 2013-02-20 05:39:49 <andytoshi> ZenBalance: you can coax bitcoind into doing it with raw transactions, i believe
121 2013-02-20 05:40:16 <andytoshi> basically, on the server side you generate a transaction and look at its hash (it's easy to do this with bitcoind)
122 2013-02-20 05:40:38 <andytoshi> on the client side, you have to create a transaction with a guessed hash as an input (not sure if bitcoind will do this for you)
123 2013-02-20 05:41:59 <ZenBalance> so I have a bitcoind server running on both machines? Can those servers talk directly (prompted by command line calls) or do I build a specific software layer in-between that passes the messages?
124 2013-02-20 05:42:26 <andytoshi> ZenBalance: they talk directly, using the bitcoin protocol
125 2013-02-20 05:42:49 <andytoshi> you create transactions and broadcast them to the network using bitcoind as a p2p client
126 2013-02-20 05:42:57 <andytoshi> for transactions, see: https://en.bitcoin.it/wiki/Transactions
127 2013-02-20 05:42:59 <andytoshi> https://en.bitcoin.it/wiki/Script
128 2013-02-20 05:47:57 <andytoshi> ZenBalance: what course is this project for?
129 2013-02-20 05:50:19 <ZenBalance> a general CS engineering course
130 2013-02-20 05:50:47 <ZenBalance> I did policy research on bit coin for a CS ethics class last year and go really interested in the technical aspects
131 2013-02-20 05:51:20 <ZenBalance> I am majoring in CS systems (but definitely still a novice) so I thought I would push myself for an open ended class project to build a "useful" web service
132 2013-02-20 05:52:44 <andytoshi> cool, i think you'll be okay then..
133 2013-02-20 05:52:59 <andytoshi> i had a similar idea for a graduate-level physics course i am taking
134 2013-02-20 05:53:13 <andytoshi> and now i'm studying 8-10 hours every day for this one course
135 2013-02-20 05:53:31 <andytoshi> so, it's possible to bite off more than you can chew
136 2013-02-20 05:54:01 <ZenBalance> lol, I am feeling that already, but am trying to keep it open ended so I have backup
137 2013-02-20 05:54:08 <andytoshi> (not "similar" in that my project has anything to do with bitcoin, just similar in that i went way outside of the course to explore something i had no idea about)
138 2013-02-20 05:54:44 <ZenBalance> hmm? you're physics, huh? Makes sense, some of the best CS guys I know come from strong physics backgrounds. Legit stuff
139 2013-02-20 05:54:58 <andytoshi> ZenBalance: well, even talking about bitcoin or translating the satoshi paper into more accessible/precise language would be a good project
140 2013-02-20 05:55:29 <ZenBalance> Never thought about that. Couldn't do it for this class, but would be a good project for a humanities req.
141 2013-02-20 05:55:42 <ZenBalance> I was also thinking of building off the coin base API to do something...
142 2013-02-20 05:56:54 <andytoshi> i'm very unfamiliar with what coinbase is or what its API does :P
143 2013-02-20 06:08:29 <ZenBalance> It basically lets you easily spin up wallets and take money in and out
144 2013-02-20 06:08:35 <ZenBalance> (sorry for the delay in response)
145 2013-02-20 06:08:42 <ZenBalance> they take a 1% cut
146 2013-02-20 06:08:53 <ZenBalance> which is pretty significant for the number of transactions they are doing
147 2013-02-20 06:09:04 <ZenBalance> building ontop of coin base, kind of defeats the purpose.
148 2013-02-20 06:09:13 <ZenBalance> also http://coinapult.com is pretty cool
149 2013-02-20 06:09:27 <ZenBalance> I thought it would be fun to build a replica of that, but the security limits might be too high
150 2013-02-20 06:10:19 <randy-waterhouse> ok, really happy with 0.8 ... gonna go back to running a full node again.
151 2013-02-20 06:10:30 <randy-waterhouse> good work guys, thanks
152 2013-02-20 06:18:36 <ZenBalance> andytoshi, signing off for now, but thanks for the tips!
153 2013-02-20 09:36:02 <_dr> is there any comprehensive forum post of the state of secp256k1 optimization?
154 2013-02-20 09:37:40 <sipa> _dr: hal finney made some posts about it
155 2013-02-20 09:38:17 <sipa> _dr: one optimization he suggested/implemented is ready as pull request (#2061) for the reference client
156 2013-02-20 09:38:33 <_dr> i've read various posts from him. but all i can find are algorithmic optimizations
157 2013-02-20 09:38:48 <_dr> i was wondering if it was worth the trouble to replace the whole openssl stuff with some assembly code
158 2013-02-20 09:38:56 <sipa> perhaps it is
159 2013-02-20 09:39:15 <sipa> but openssl already uses quite advanced algorithms, and assembly-optimized lowlevel routines
160 2013-02-20 09:39:30 <_dr> all the more incentive to try to beat them :)
161 2013-02-20 09:39:58 <sipa> i've recently been trying to beat openssl, actually
162 2013-02-20 09:40:03 <sipa> but so far, no luck :)
163 2013-02-20 09:41:02 <_dr> well, as far as i can see from the posts openssl doesn't seem to do batch verification
164 2013-02-20 09:41:21 <sipa> batch verification is very hard in our setting, actually
165 2013-02-20 09:41:32 <_dr> why is that?
166 2013-02-20 09:41:50 <sipa> first it requires the full R point to be known, not just its x coordinate, so you'd need to calculate y from that first, and this can't be batched
167 2013-02-20 09:42:17 <sipa> secondly, you'd need some pretty smart script evaluation to deal with it
168 2013-02-20 09:42:51 <sipa> as some nasty person could write a script that tries an ecdsa verification, and requires it to fail for his script to succeed
169 2013-02-20 09:45:32 <sipa> but if you can do it (perhaps just for standard scripts, and do others separately), it may be worthwhile
170 2013-02-20 09:45:53 <sipa> the reconstruction of Y doesn't seem too expensive to me
171 2013-02-20 09:46:28 <_dr> i'll try intel's crypto library first, it will set a good lower bar for what you can possibly achieve
172 2013-02-20 09:46:38 <_dr> i think i read they have support for bitcoin's curve in their lib
173 2013-02-20 09:48:21 <sipa> specific speedups can come from: the fact that our field modulus is 2^256-p with p only 33 bits, the fact that the curve a parameter is 0, and the split-ec-mult in 2 trick that hal suggested
174 2013-02-20 09:48:58 <_dr> i was hoping that vectorization and crypto instructions might be of use
175 2013-02-20 09:49:20 <_dr> haswell's AVX2 will finally introduce 8-way simd for integers, and i don't know what aes-ni etc. really offer
176 2013-02-20 09:49:20 <sipa> hmm, i know too little about those
177 2013-02-20 09:49:53 <sipa> for doing multiple sigchecks in parallel you may gain something through simd i guess
178 2013-02-20 10:06:08 <sipa> but high-speed implementations depend on being able to do a 64bit * 64bit multiplication (with 128 bit result) in hardware
179 2013-02-20 10:06:24 <sipa> i doubt there are simd versions of those available?
180 2013-02-20 10:06:41 <sipa> (specifically x86_64 arch)
181 2013-02-20 10:08:41 <_dr> yes there are
182 2013-02-20 10:08:57 <_dr> even sse has them i think, let me check
183 2013-02-20 10:13:58 <_dr> unfortunately not vectorized ones. you can only do two 32bit * 32bit and store the two 64bit results
184 2013-02-20 10:15:16 <sipa> iirc on the same hardware, openssl versions of ecdsa with assembly implementation, the 32-bit based versions are about 2 times slower than the 64-bit ones
185 2013-02-20 10:15:40 <sipa> so being able to do 2 in parallel wouldn't gain you anything
186 2013-02-20 10:17:05 <_dr> if you can somehow emulate a 64bit mult using two 32bit mults (which you probably can) that makes sense
187 2013-02-20 10:17:27 <_dr> in avx2, you should be able to do four in parallel, but maybe there's something hidden in avx-ni; i'll check
188 2013-02-20 10:17:44 <_dr> aes-ni
189 2013-02-20 10:21:52 <_dr> hm, they don't preserve the 128bit result. too bad, the 64bit*64bit = 128bit seems to be some special case here
190 2013-02-20 11:05:25 <fimp> if I replace my wallet with another wallet in 0.8, will the client then have to do all the block indexing again?
191 2013-02-20 11:25:55 <sipa> fimp: no
192 2013-02-20 11:26:01 <sipa> wallets and blockchains are independent
193 2013-02-20 11:49:49 <HM2> bah
194 2013-02-20 11:50:05 <HM2> dead wifi which wouldn't reconnect
195 2013-02-20 11:50:05 <HM2> woke up to a 600 MB resident X running at 100% cpu
196 2013-02-20 11:50:21 <HM2> something is really fucked
197 2013-02-20 11:50:59 <SomeoneWeird> 0.o
198 2013-02-20 11:51:03 <SomeoneWeird> that aint good
199 2013-02-20 13:09:51 <shwooop> hola, I remember having a link to a guide of the satoshi client source. think it was a blog post. can someone help me out?
200 2013-02-20 13:10:35 <shwooop> s/blog/forum
201 2013-02-20 13:31:19 <MC1984> hey
202 2013-02-20 13:31:31 <MC1984> is the bitcoin dev mailing list open to all
203 2013-02-20 13:31:42 <sipa> yes
204 2013-02-20 13:32:36 <MC1984> how do i subscribe? I promise i will never myself reply, and only observe
205 2013-02-20 13:34:11 <tcatm> MC1984: List-Subscribe: <https://lists.sourceforge.net/lists/listinfo/bitcoin-development> <mailto:bitcoin-development-request@lists.sourceforge.net?subject=subscribe>
206 2013-02-20 13:35:28 <MC1984> thanks
207 2013-02-20 13:53:03 <HM2> i've come to the conclusion the only reason i use Geany as my primary text editor is it has a teapot as a logo
208 2013-02-20 13:53:38 <HM2> then i realised it was an Aladdinesque oil lamp
209 2013-02-20 14:03:03 <helo> 272MB up to 302MB in 21 hours
210 2013-02-20 14:03:30 <helo> just to present a data point
211 2013-02-20 14:06:53 <HM2> memory?
212 2013-02-20 14:06:54 <epscy> helo: what do those numbers relate to?
213 2013-02-20 14:07:06 <helo> yes
214 2013-02-20 14:07:19 <epscy> are you limiting the connections at all?
215 2013-02-20 14:07:27 <helo> no upnp at router, so 8
216 2013-02-20 14:07:39 <epscy> i found that helped a lot when my bitcoind kept running out of memory