1 2012-12-04 00:21:30 <etotheipi_> sipa: btw, what are you using for benchmarking?
2 2012-12-04 00:21:45 <sipa> gmaxwell
3 2012-12-04 00:21:54 <etotheipi_> haha
4 2012-12-04 00:22:06 <sipa> wall clock time, really
5 2012-12-04 00:22:19 <etotheipi_> I meant, what code constructs?
6 2012-12-04 00:22:27 <etotheipi_> C++, CLOCKS_PER_SEC?
7 2012-12-04 00:23:27 <sipa> boost::posix_time::microsec_clock::universal_time()
8 2012-12-04 00:23:41 <etotheipi_> ooh, right... boost has useful stuff for that
9 2012-12-04 00:23:58 <etotheipi_> I have a cool construct I came up with, which might be useful
10 2012-12-04 00:24:05 <etotheipi_> I now just automatically time all my functions
11 2012-12-04 00:25:00 <etotheipi_> it's a singleton class that keeps a map of timers (indexed by name/string).... and you create a TimerToken at the top of the function to start the timer, and it stops the timer when it destructs
12 2012-12-04 00:25:34 <etotheipi_> and then accumulates numCalls and avgTime into a .csv
13 2012-12-04 00:25:42 <etotheipi_> for each neam
14 2012-12-04 00:25:47 <etotheipi_> *name
15 2012-12-04 00:26:16 <sipa> you may want to look into compiling with profiling enabled
16 2012-12-04 00:26:24 <sipa> that does that automatically :)
17 2012-12-04 00:26:48 <etotheipi_> well, it can of course time arbitrary code blocks, too
18 2012-12-04 00:27:32 <etotheipi_> I'm just throwing it out there, that it's a pretty compact timing class that is easy to use, if you do a lot of benchmarking
19 2012-12-04 00:28:22 <etotheipi_> but I probably should've known that compilers have profiling options...
20 2012-12-04 00:29:28 <yellowhat> i'll just randomly throw out that i love my java profiling tools. they are really mature and partly integrated in the ide/testcases. of course that does not help you right now :) good night!
21 2012-12-04 00:30:18 <etotheipi_> you can also put tokens inside of loops and it will time every cycle of the loop
22 2012-12-04 00:44:16 <jgarzik> D34TH: http://tech.slashdot.org/story/12/06/17/1351251/spdy-not-as-speedy-as-hyped
23 2012-12-04 00:45:52 <jgarzik> now that TX signing is implemented, time to turn attention back to the SPV client
24 2012-12-04 00:59:34 <gmaxwell> jgarzik: where are you getting the random data for signing?
25 2012-12-04 01:05:55 <D34TH> alright time to pull the latest picocoin and see if i can compile it
26 2012-12-04 01:06:42 <D34TH> damn
27 2012-12-04 01:07:09 <D34TH> glib is a PITA
28 2012-12-04 01:09:19 <D34TH> static declaration of g_list_free_full
29 2012-12-04 01:09:29 <D34TH> jgarzik, naughty?
30 2012-12-04 01:10:42 <D34TH> removed it, and it got back up to pfd?
31 2012-12-04 01:20:35 <D34TH> jgarzik i'm totally going to pullreq some code
32 2012-12-04 01:39:33 <BlueMatt> jgarzik: yes, half the point of spdy is that you can marginally speed up over http so that you can enable https on all connections (which the internet desperately needs...)
33 2012-12-04 01:41:10 <Luke-Jr> but https depends on the broken CA model and doesn't work for people who won't shell out $$$ for a cert
34 2012-12-04 01:42:26 <BlueMatt> yes, pki is broken, but having marginal security that depends on pki is 1000x better than no encryption at all
35 2012-12-04 01:42:30 <Luke-Jr> but that article is bogus, since he was "testing" SPDY on non-SPDY sites!
36 2012-12-04 01:42:31 <Luke-Jr> but that article is bogus, since he was "testing" SPDY on non-SPDY sites!
37 2012-12-04 01:43:15 <gmaxwell> 12/04/12 02:01:07 SetBestChain: new best=00000000839a8e6886ab5951d76f411475428afc90947ee320161bbf18eb6048 height=1 work=8590065666 tx=2 date=01/09/09 02:54:25
38 2012-12-04 01:43:18 <gmaxwell> 12/04/12 02:41:20 SetBestChain: new best=000000000000048b95347e83192f69cf0366076336c639f9b7228e9ba171342e height=210000 work=628963747775700992096 tx=9344662 date=11/28/12 15:24:38
39 2012-12-04 01:43:27 <gmaxwell> sipa: your latest update seems slower; I ran it stice to be sure.
40 2012-12-04 01:43:31 <gmaxwell> twice
41 2012-12-04 01:44:55 <BlueMatt> Luke-Jr: ok, I didnt actually read the article, but, still...https everywhere...
42 2012-12-04 01:49:58 <D34TH> jgarzik, :D pull req submitted
43 2012-12-04 02:00:22 <gmaxwell> Luke-Jr: well, see startssl... there are free certs now.
44 2012-12-04 02:01:28 <Luke-Jr> gmaxwell: for my IP address? :p
45 2012-12-04 02:01:41 <gmaxwell> hahah
46 2012-12-04 02:01:49 <gmaxwell> technically you can have a cert with an IP address.
47 2012-12-04 02:01:56 <gmaxwell> I bet some moron CA would issue it too.
48 2012-12-04 02:03:32 <Luke-Jr> <.<
49 2012-12-04 02:09:56 <etotheipi_> gmaxwell: well do you think about... if a hard-fork-update were required anyway, adding 8 bytes to TxIn serializations to include the value of the TxOut being spent?
50 2012-12-04 02:11:11 <etotheipi_> or is that too much of a hard change?
51 2012-12-04 02:12:20 <weex> etotheipi_: < hasha> is there a btc Armory channel?
52 2012-12-04 02:12:30 <weex> from #bitcoin
53 2012-12-04 02:12:51 <etotheipi_> weex: hmm... maybe I should start one...
54 2012-12-04 02:13:05 <etotheipi_> but I don't spend that much time on IRC
55 2012-12-04 02:13:24 <etotheipi_> I guess that wouldn't stop other people from chit-chatting about it without me, though :)
56 2012-12-04 02:13:24 <weex> yeah, i know how that goes
57 2012-12-04 02:13:51 <weex> started a channel #CoinDL and just put a bot in it today
58 2012-12-04 02:14:17 <hasha> yes you should start one :)
59 2012-12-04 02:14:24 <weex> make a bitcoin wiki lookup command for said bot
60 2012-12-04 02:14:38 <etotheipi_> hmm... I don't know anything about it
61 2012-12-04 02:14:38 <weex> so many development distractions :)
62 2012-12-04 02:15:31 <hasha> irc is one huge distraction
63 2012-12-04 02:15:41 <etotheipi_> hasha: that's exactly why I don't spend much time on it
64 2012-12-04 02:16:02 <etotheipi_> my productivity went up quite a bit after I closed the IRC window (instead of minimizing it)
65 2012-12-04 02:16:37 <hasha> the &
66 2012-12-04 02:17:35 <etotheipi_> so is it easy to setup an IRC bot?
67 2012-12-04 02:17:44 <weex> yeah about 30 mins tops
68 2012-12-04 02:17:47 <weex> i used phenny
69 2012-12-04 02:17:52 <weex> python
70 2012-12-04 02:18:06 <etotheipi_> ...I like python :)
71 2012-12-04 02:18:07 <weex> then moddd the wikipedia module to point at en.bitcoin.it
72 2012-12-04 02:19:01 <weex> i'm thinking of a module to query coindl or try to buy something
73 2012-12-04 02:19:21 <weex> probably better off working on the web interface or an api though
74 2012-12-04 02:19:43 <hasha> i have a Amory question.. i just pulled down .85-beta and tried to import a backup of my wallet made with bitcoin-qt 0.7.1...
75 2012-12-04 02:20:00 <hasha> after scanning it for some time, it didnt actually open it
76 2012-12-04 02:20:05 <etotheipi_> hasha: Armory doesn't support importing wallets from bitcoin-qt
77 2012-12-04 02:20:12 <etotheipi_> it used to, before they switched to compressed public keys
78 2012-12-04 02:20:16 <hasha> ah, that explains it
79 2012-12-04 02:20:25 <hasha> thanks
80 2012-12-04 02:20:28 <etotheipi_> I never got around to upgrading it (but will soon)
81 2012-12-04 02:20:52 <etotheipi_> (I'm making a new wallet format with that in mind... I want users to be able to do it, I just didn't want to touch my rock-solid, thoroughly-tested wallet code)
82 2012-12-04 02:21:16 <hasha> i might have missed that on the site :( sorry if im asking a idiot question
83 2012-12-04 02:21:25 <etotheipi_> Nope, not an idiot question
84 2012-12-04 02:21:32 <etotheipi_> I *should* be more explicit about that
85 2012-12-04 02:21:34 <weex> etotheipi_: are you doing another funding round?
86 2012-12-04 02:21:35 <weex> etotheipi_: are you doing another funding round?
87 2012-12-04 02:21:42 <etotheipi_> weex: I wasn't planning on it
88 2012-12-04 02:22:26 <weex> cool, it's a good model though if it works
89 2012-12-04 02:22:33 <etotheipi_> it felt kind of beggar-y the first time around, and even though it was successful, I think I even stated that I wasn't going to come back for funding later
90 2012-12-04 02:23:09 <weex> ahh i see
91 2012-12-04 02:23:30 <etotheipi_> I'm pretty happy working 30 hrs per week at my normal job, and comfortable with the income stream
92 2012-12-04 02:23:53 <hasha> like anyone who goes to see venture capitalists... first they make you feel bad, then take 80% of your company :)
93 2012-12-04 02:23:54 <etotheipi_> and maybe I'll find a way to make some money from Armory itself, without being to intrusive
94 2012-12-04 02:23:55 <etotheipi_> and maybe I'll find a way to make some money from Armory itself, without being to intrusive
95 2012-12-04 02:24:29 <etotheipi_> or maybe some VC will come find me and give me an offer I can't refuse :)
96 2012-12-04 02:24:42 <hasha> an angel investor is always nice too
97 2012-12-04 02:24:56 <etotheipi_> but I have every intention on keeping Armory 90%+ free
98 2012-12-04 02:24:57 <etotheipi_> but I have every intention on keeping Armory 90%+ free
99 2012-12-04 02:25:10 <etotheipi_> if anything, only charge for the really advanced features
100 2012-12-04 02:25:20 <etotheipi_> (most of which haven't been written yet)
101 2012-12-04 02:25:35 <weex> enterprise features
102 2012-12-04 02:26:13 <weex> integrate with active directory
103 2012-12-04 02:26:39 <weex> IIS + Bitcoin = pure magic :P
104 2012-12-04 02:27:08 <etotheipi_> interesting...
105 2012-12-04 02:27:09 <etotheipi_> interesting...
106 2012-12-04 02:27:39 <etotheipi_> I don't know for sure which way to go with it... I kinda just want to implement everything, and have people donate enough regularly so I don't have to work... but I doubt that's going to happen :)
107 2012-12-04 02:29:23 <hasha> could happen. if it turns into the default go-to app
108 2012-12-04 02:30:28 <etotheipi_> I'm very hesitant to go towards the paid-product model... then I have "obligations" to my "customers"
109 2012-12-04 02:31:41 <etotheipi_> even if it's only for a fraction of the users
110 2012-12-04 02:35:27 <weex> yeah keeping it free is cool. someone can always develop on it for some commercial needs
111 2012-12-04 02:37:33 <etotheipi_> I thought of a fun way to charge for features if I needed to though: "This feature will be enabled when there are 1 BTC worth of transfers to the Armory address in your combined wallet history"
112 2012-12-04 02:38:07 <gmaxwell> etotheipi_: import other people's pubkeys. :P
113 2012-12-04 02:38:18 <etotheipi_> and then a relatively easy way to bypass it
114 2012-12-04 02:38:36 <gmaxwell> sipa: thoughts on wallet cache information??? stuff that we might want to have on disk about the wallet but doesn't really need to persist??? and how that should interact with append only wallet stuff in the future?
115 2012-12-04 02:39:06 <etotheipi_> I dont' want to put in some kind of draconian DRM, and I know people will be selfish and bypass it anyway, I can at least be good-humored about it (which gains me fans), and simply make the payment option the path of least resistance
116 2012-12-04 02:39:45 <gmaxwell> sipa: e.g. I'd be hesitant to keep a union structure for taint tracking in memory because holding all addresses in memory might be problematic. But I don't like the idea of putting the relationships in the wallet because it'll generate a bunch of writes that would be kinda ugly on an append only wallet.
117 2012-12-04 02:40:01 <etotheipi_> "Send 1 BTC to enable this feature, or change the "IAmASelfishJerk" variable from False to True in the ~/.armory/settings.txt file"
118 2012-12-04 02:40:40 <gmaxwell> etotheipi_: the thin line you walk there is that people should ask themselves what other 'antifeatures' are in the software.
119 2012-12-04 02:41:16 <gmaxwell> (of course they should always ask that??? but they don't :P)
120 2012-12-04 02:45:50 <gavinandresen> etotheipi_: I bet a "send 0.01 BTC to support Armory development" checkbox on your send-coins dialogs (not checked by default) would get used.
121 2012-12-04 02:47:44 <gmaxwell> gavinandresen: the miners authors seem to have not had great results with the 'direct $x% of your mining to support development'
122 2012-12-04 02:48:19 <gmaxwell> (poor enough that e.g. cgminer has dropped it)
123 2012-12-04 02:48:42 <gavinandresen> gmaxwell: you might be right, everybody might be selfish...
124 2012-12-04 02:49:20 <gavinandresen> gmaxwell: I wonder if people feel differently about a service like a mining pool versus a product they download and run.
125 2012-12-04 02:50:39 <etotheipi_> gavinandresen: there's already a button on the send-dialog that adds a 1 BTC pre-filled donation to their current transactions
126 2012-12-04 02:51:07 <etotheipi_> I've had users tell me they use it
127 2012-12-04 02:51:37 <gavinandresen> etotheipi_: don't you know if they use it? You get the donations, right?
128 2012-12-04 02:51:54 <Luke-Jr> gavinandresen: well, that *was* a program they download and run. it switched to the author's pool/account for N% of the time
129 2012-12-04 02:54:28 <gmaxwell> Well. there are other factors??? the switching code was not bug free at all times, and armory users aren't miner users.
130 2012-12-04 02:54:40 <gavinandresen> OK. I never liked donations as a sustainable business model....
131 2012-12-04 02:56:23 <gavinandresen> (I was pleasantly surprised to see how many BTC the Foundation has received in donations, though)
132 2012-12-04 02:58:49 <etotheipi_> gavinandresen: I get a lot of donations, but I don't know if those donations were made because of the button
133 2012-12-04 02:59:10 <etotheipi_> or rather, I don't know how many of those donations would've been made anyway, if the button wasn't there
134 2012-12-04 02:59:44 <etotheipi_> I could always put the "donate" button as a bigger button around/behind the send button, so a slight misclick is a win for me :)
135 2012-12-04 03:00:03 <gavinandresen> A marketing person would have told you to use a separate donation address for every donation "channel" (I probably just got that marketingspeak wrong)
136 2012-12-04 03:02:02 <gavinandresen> They'd also want you to experiment with making the suggested donation amount different (at random), to see if asking for 0.09 BTC got you 20 times the donations as asking for 1.0 ...
137 2012-12-04 03:14:07 <weex> have the actual send button bounce around in a field of donate
138 2012-12-04 03:44:37 <Luke-Jr> someone just tried to send 0.03 BTC to anyone-can-spend? O.o
139 2012-12-04 03:44:38 <jgarzik> gmaxwell: where am I getting the random data? hah! That would imply I tested it!
140 2012-12-04 03:44:55 <jgarzik> gmaxwell: TX signing is at Linus Stage #3: it builds
141 2012-12-04 03:45:43 <jgarzik> ACTION was thinking about generating some keys, then creating some test data for test_bitcoin in bitcoin.git: [private key, serialized TX unsigned, serialized TX signed]
142 2012-12-04 03:46:07 <jgarzik> then import that into picocoin.git, once it's working in bitcoin.git
143 2012-12-04 03:46:19 <jgarzik> there are a few signing tests, but they are all procedural
144 2012-12-04 03:46:27 <gmaxwell> jgarzik: I only asked because the source of random data is on my mental checklist for very easy to get wrong and terrible wrt signing. (esp because a failure is invisible until someone steals all your money)
145 2012-12-04 03:47:47 <jgarzik> gmaxwell: does the aforementioned triple sound ok to you?
146 2012-12-04 03:57:35 <gmaxwell> signing is not deterministic with plain ECDSA, which might frustrate your tests. :P
147 2012-12-04 04:06:20 <ByronJohnson> .u'e ba'o toseikrefu se'u tolmo'i lonu barda fa lo ve zukrai zmadu be fo la jbobau
148 2012-12-04 04:06:33 <ByronJohnson> Sorry, wrong channel.
149 2012-12-04 04:06:34 <ByronJohnson> Sorry, wrong channel.
150 2012-12-04 04:09:30 <jgarzik> gmaxwell: really? not even with the same private key?
151 2012-12-04 04:09:38 <jgarzik> and input hash
152 2012-12-04 04:13:34 <gmaxwell> Correct!
153 2012-12-04 04:14:30 <gmaxwell> jgarzik: ecdsa needs a random numbers. And woe to you if you reuse a random number or choice it too non-uniformly.
154 2012-12-04 04:15:43 <gmaxwell> Result: total private key recovery. This can be fixed by generating the random number as H(message || secret data) ... as ed25519 specifies. but you need a secret for that...
155 2012-12-04 04:20:54 <jgarzik> gmaxwell: hmmmm
156 2012-12-04 04:21:06 <jgarzik> ACTION starts thinking about PRNGs and seeds
157 2012-12-04 04:25:46 <gmaxwell> you need to use a CSPRNG initilized from /dev/(u)random or the like. Openssl has randomhandling magic for you. (unless you're on debian :P)
158 2012-12-04 04:25:57 <gmaxwell> randomness*
159 2012-12-04 04:29:29 <jgarzik> gmaxwell: I'm talking about an artificial environment that produces repeatable tests
160 2012-12-04 04:29:34 <jgarzik> gmaxwell: not normal signing
161 2012-12-04 04:30:51 <jgarzik> gmaxwell: e.g. maybe the tests can provide [prng seed, private key (full privkey, not secret), unsigned TX, signed TX]
162 2012-12-04 04:31:21 <jgarzik> and test that [seed, privkey, unsigned TX] yields [signed TX]
163 2012-12-04 04:31:52 <jgarzik> a bit PRNG-specific, but should be repeatable
164 2012-12-04 04:31:53 <jgarzik> a bit PRNG-specific, but should be repeatable
165 2012-12-04 08:10:09 <sipa> gmaxwell: how do you mean slower... i didn't change anything!
166 2012-12-04 08:10:42 <sipa> just cosmetical... oh and 128 as batch limit
167 2012-12-04 08:12:08 <sipa> gmaxwell: about data in memory for wallets... i wouldn't bother for now! as we keep everything in memory whatsoever already
168 2012-12-04 08:52:11 <doublec> ByronJohnson: xu do tavla bau la lojban
169 2012-12-04 11:15:28 <Luke-Jr> jgarzik: new 3 TB drive "Device does not support SMART"
170 2012-12-04 11:16:05 <sipa> Luke-Jr: connected how?
171 2012-12-04 11:16:18 <Luke-Jr> sipa: SATA of some sort
172 2012-12-04 11:18:30 <Luke-Jr> looks like passing -d sat fixes it
173 2012-12-04 11:43:30 <Jouke> jgarzik: I get an error while trying to create a new wallet in picocoin http://pastebin.com/dbdz8YeY
174 2012-12-04 11:56:12 <Luke-Jr> 190 Airflow_Temperature_Cel 0x0022 037 031 045 Old_age Always FAILING_NOW 63 (0 87 69 24 0)
175 2012-12-04 12:06:05 <Luke-Jr> scary :/
176 2012-12-04 12:37:59 <BitDev> hi all, i gave a problem... my Ecdsa private key is 279 bytes long and public key is 65 bytes long... but in wiki i found that bitcoin private key length is 32 bytes... what i am doing wrong?
177 2012-12-04 12:38:40 <sipa> nothing
178 2012-12-04 12:39:02 <sipa> OpenSSL's serialized private keys are 279 bytes indeed, but they contain a lot of redundant information
179 2012-12-04 12:39:21 <sipa> including the field parameters, the curve information, the public key, and the private parameter itself
180 2012-12-04 12:39:22 <sipa> including the field parameters, the curve information, the public key, and the private parameter itself
181 2012-12-04 12:39:50 <sipa> in most cases, we're only interested in the private parameter as all the rest is either fixed, or derivable from the private parameter
182 2012-12-04 12:40:12 <BitDev> how to get only private key?
183 2012-12-04 12:40:13 <BitDev> how to get only private key?
184 2012-12-04 12:40:29 <BitDev> and public key length must be 65 bytes long?
185 2012-12-04 12:40:48 <sipa> uncompressed public keys are 65 bytes
186 2012-12-04 12:42:15 <BitDev> to generate bitcoin address i must use uncompressed public key? (here: Version concatenated with RIPEMD-160(SHA-256(public key)))
187 2012-12-04 12:42:17 <BitDev> &
188 2012-12-04 12:42:18 <BitDev> ?
189 2012-12-04 12:44:06 <kinlo> do not think of compressed keys like you think of zip files. Compressed keys are just a different way of writing it so it takes less space but they are directly useable
190 2012-12-04 12:44:08 <BitDev> and, Version = 1 byte of 0 (zero); on the test network, this is 1 byte of 111, - for testnet this value will be one byte with value 3 and for real network 1 byte with value 0?
191 2012-12-04 12:44:29 <kinlo> afaik you can use either an uncompresser or compressed public key to generate the bitcoin address
192 2012-12-04 12:44:49 <BitDev> to calculate bitcoin address what kind of public key i must to use?
193 2012-12-04 12:44:55 <BitDev> ow
194 2012-12-04 12:44:58 <kinlo> the public key
195 2012-12-04 12:44:59 <BitDev> great
196 2012-12-04 12:45:11 <BitDev> and how to compress public key?
197 2012-12-04 12:45:51 <sipa> if you're using OpenSSL, by setting a flag on the EC_KEY structure
198 2012-12-04 12:46:37 <sipa> from bitcoind's key.cpp:
199 2012-12-04 12:46:39 <sipa> EC_KEY_set_conv_form(pkey, fCompressed ? POINT_CONVERSION_COMPRESSED : POINT_CONVERSION_UNCOMPRESSED);
200 2012-12-04 12:47:41 <BitDev> thnx and one more question... for testing purpose i want to export public key in format that other client like Armory - how i can do this?
201 2012-12-04 12:49:35 <sipa> check how Armory exports public keys; i have no clue
202 2012-12-04 12:53:51 <BitDev> and how Bitcoin-Qt can import private key?
203 2012-12-04 12:54:06 <BitDev> in compressed form or uncompressed?
204 2012-12-04 12:54:29 <sipa> both
205 2012-12-04 12:54:54 <sipa> see the importprivkey RPC
206 2012-12-04 12:55:33 <BitDev> thnx i will
207 2012-12-04 13:41:45 <helo> ACTION notes that "show qrcode" isn't working in head
208 2012-12-04 13:45:38 <jaromil> Jouke, jgarzik: looks like picocoin would benefit from a valgrind shave as much as I'll do from a SPA next weekend
209 2012-12-04 13:45:39 <jaromil> Jouke, jgarzik: looks like picocoin would benefit from a valgrind shave as much as I'll do from a SPA next weekend
210 2012-12-04 14:27:48 <sipa> ;;bc,blocks
211 2012-12-04 14:27:54 <gribble> timed out
212 2012-12-04 14:28:02 <sipa> ;;bc,blocks
213 2012-12-04 14:28:08 <gribble> timed out
214 2012-12-04 14:30:49 <BlueMatt> 210875
215 2012-12-04 14:32:42 <phantomcircuit> BlueMatt, 210876
216 2012-12-04 14:33:03 <BlueMatt> hmm, still not here
217 2012-12-04 14:33:30 <phantomcircuit> i thought we were counting
218 2012-12-04 14:33:34 <phantomcircuit> ACTION lols
219 2012-12-04 14:33:35 <BlueMatt> ...
220 2012-12-04 14:33:36 <BlueMatt> ...
221 2012-12-04 14:33:40 <phantomcircuit> ahah
222 2012-12-04 14:33:49 <sipa> well there *are* 210876 blocks
223 2012-12-04 14:33:50 <phantomcircuit> sorry i couldn't help myself
224 2012-12-04 14:34:11 <BlueMatt> sipa: you lie
225 2012-12-04 14:34:21 <phantomcircuit> he's technically correct
226 2012-12-04 14:34:28 <phantomcircuit> genesis block is block 0 iirc
227 2012-12-04 14:34:32 <sipa> correct
228 2012-12-04 14:34:33 <BlueMatt> ...
229 2012-12-04 14:34:46 <phantomcircuit> and as we all know
230 2012-12-04 14:34:53 <phantomcircuit> technically correct is the only kind of correct
231 2012-12-04 14:35:10 <sipa> technically, technically correct is the only technical kind of correct
232 2012-12-04 14:35:30 <epscy> correctification
233 2012-12-04 14:35:31 <epscy> correctification
234 2012-12-04 14:35:51 <BlueMatt> correction: technically cs people should count from 0, so...
235 2012-12-04 14:36:22 <sipa> count from 0 when identifying elements, but use the total number when counting
236 2012-12-04 14:36:40 <sipa> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
237 2012-12-04 14:36:41 <BlueMatt> yes, and ,,bc,blocks doesnt really specify which ;)
238 2012-12-04 14:36:42 <BlueMatt> yes, and ,,bc,blocks doesnt really specify which ;)
239 2012-12-04 14:36:44 <gribble> 210686
240 2012-12-04 14:36:44 <sipa> 6987 pieterw 20 0 1555m 581m 109m R 988 1.8 15:47.52 bitcoind
241 2012-12-04 14:36:52 <BlueMatt> damnit gribble
242 2012-12-04 14:37:55 <phantomcircuit> ahah
243 2012-12-04 14:54:35 <MrTiggr> Quote of the Week: "All programmers are playwrights and all computers are lousy actors." --Unknown
244 2012-12-04 14:57:10 <edcba_> playwrights ?
245 2012-12-04 14:58:21 <edcba_> would have s/playwrights/playrihts/ & s/Unknown/A programmer/
246 2012-12-04 14:58:43 <edcba_> fuck i made more errors than i wanted
247 2012-12-04 14:58:45 <epscy> playrihts?
248 2012-12-04 14:59:01 <epscy> it is playwright, i'm fairly sure
249 2012-12-04 14:59:20 <edcba_> yes but would have been funnier with a fault
250 2012-12-04 15:23:29 <jgarzik> Jouke, jaromil: a bug introduced late last night. Just fixed. picocoin actually gets valgrind treatment on a somewhat regular basis.
251 2012-12-04 15:30:13 <ciscoftw> how is it that satoshidice is able to send 0.00000001 without a transaction fee?
252 2012-12-04 15:30:43 <Eliel> jgarzik: do you think libccoin/picocoin would be suitable for mobile devices?
253 2012-12-04 15:31:04 <jgarzik> Eliel: when complete, sure
254 2012-12-04 15:31:27 <kinlo> ciscoftw: they do pay fee's
255 2012-12-04 15:31:50 <kinlo> ciscoftw: in fact, they account for most fee's in current blocks I think
256 2012-12-04 15:31:59 <ciscoftw> how would i be able to see that via bitcoind
257 2012-12-04 15:32:07 <kinlo> see what?
258 2012-12-04 15:32:43 <ciscoftw> that a transaction fee was included into a transaction
259 2012-12-04 15:33:00 <Eliel> jgarzik: if the target public release date is May 2013, do you think it'd make sense to start building an app on top of libccoin?
260 2012-12-04 15:33:32 <jgarzik> Eliel: sure
261 2012-12-04 15:33:48 <ciscoftw> ./bitcoind getrawtransaction, returns the raw transaction, but i cant determine if a fee was included or not?
262 2012-12-04 15:33:54 <jgarzik> Eliel: just make sure your mobile platform supports C development
263 2012-12-04 15:33:59 <Eliel> We're thinking of getting started building a bitcoin wallet app for Jolla (as well as Meego/Maemo)
264 2012-12-04 15:34:08 <kinlo> ciscoftw: that's not so easy
265 2012-12-04 15:34:13 <sipa> ciscoftw: transactions don't know that
266 2012-12-04 15:34:15 <kinlo> ciscoftw: first get raw transaction
267 2012-12-04 15:34:21 <kinlo> ciscoftw: then decode that transaction
268 2012-12-04 15:34:22 <Eliel> We might have a shot at getting it included in the official Jolla.
269 2012-12-04 15:34:36 <kinlo> ciscoftw: then add all output's to see how much coins have been reassigned
270 2012-12-04 15:34:54 <kinlo> ciscoftw: then do getrawtransaction and decode transaction on all input transactions
271 2012-12-04 15:35:00 <kinlo> count how much the input was total
272 2012-12-04 15:35:05 <kinlo> substract those 2 values
273 2012-12-04 15:35:09 <kinlo> and you have the fee payed
274 2012-12-04 15:35:36 <ciscoftw> so the output im seeing is hex format? looks like its asci...
275 2012-12-04 15:35:58 <kinlo> try the decodetransaction function from the bitcoin client
276 2012-12-04 15:36:04 <kinlo> and feed it the raw data
277 2012-12-04 15:37:43 <ciscoftw> kinlo: fucking awesome!!!
278 2012-12-04 15:38:00 <kinlo> yes bitcoin is well designed and awesome :p
279 2012-12-04 15:38:08 <ciscoftw> many thax for your insight
280 2012-12-04 15:38:12 <kinlo> np
281 2012-12-04 16:13:04 <Jouke> jgarzik: now "picocoin help" segfaults
282 2012-12-04 16:13:27 <Jouke> I was able to create a wallet
283 2012-12-04 16:24:32 <jgarzik> Jouke: reproduced
284 2012-12-04 16:24:52 <jgarzik> This is my punishment for encouraging third party submissions ;p
285 2012-12-04 16:30:09 <jgarzik> Jouke: fix pushed
286 2012-12-04 16:35:22 <TD> hello
287 2012-12-04 16:39:39 <BlueMatt> hey TD
288 2012-12-04 16:42:31 <sipa> TD: seen #2060 and #2061? :)
289 2012-12-04 16:44:20 <TD> cool
290 2012-12-04 16:44:26 <TD> i saw 2060 last night but didn't look at it yet
291 2012-12-04 16:45:00 <TD> shouldn't finishing ultraprune be the priority though?
292 2012-12-04 16:45:10 <TD> it still isn't shippable due to lack of upgrade support, i thought?
293 2012-12-04 16:45:46 <sipa> indeed, and other things, but i got tired of waiting hours for synchronization
294 2012-12-04 16:46:13 <BlueMatt> so add a [temporary] checkpoint on 210000?
295 2012-12-04 16:46:22 <sipa> just benchmarked a full import from network in 37 minutes
296 2012-12-04 16:46:37 <TD> ok
297 2012-12-04 16:46:54 <sipa> BlueMatt: this helps more :)
298 2012-12-04 16:49:06 <TD> so for bloom filtering, my understanding is that the C++ branch has your new merkle block encodings already merged, but it isn't reflected in the BIP draft or in the java impl yet
299 2012-12-04 16:49:11 <TD> just trying to ensure i have full state
300 2012-12-04 16:50:20 <BlueMatt> TD: another important thing to make pimping spv clients on bitcoin.org a better idea is to add upgradeable support to upgrade to full clients
301 2012-12-04 16:50:48 <BlueMatt> TD: correct, well, I have a java impl, just not well tested or pushed yet
302 2012-12-04 16:51:32 <BlueMatt> TD: on that note, I have a slightly different db format for the h2fullprunedblockstore thinggy that supports future upgrading that I'd like to have in before there are any releases (even if none of the upgrade stuff is there)
303 2012-12-04 16:52:01 <TD> ok
304 2012-12-04 16:52:06 <TD> sure
305 2012-12-04 16:52:09 <TD> there's no hurry to release
306 2012-12-04 16:52:24 <BlueMatt> just kept forgetting to mention it, so I figured I might as well while I was thinking of it
307 2012-12-04 16:52:55 <TD> alright
308 2012-12-04 16:53:14 <TD> i'm not sure upgrading support matters that much. if you're going to run a full node, it should be a conscious decision you made to make a little bit of effort
309 2012-12-04 16:53:43 <TD> going from SPV to full client means fully downloading and rebuilding the chain anyway
310 2012-12-04 16:53:52 <TD> so switching to a different bit of software is hardly a big deal compared to that
311 2012-12-04 16:54:12 <sipa> i'm not convinced it should be a conscious decision
312 2012-12-04 16:54:15 <BlueMatt> TD: yes, but support for it in-client via a nice little checkbox (and ample description) is really nice, and imho, really allows us to say everyone can use one of these clients...
313 2012-12-04 16:54:37 <BlueMatt> TD: not if your wallet is already somewhere and having to transfer it is a pain
314 2012-12-04 16:54:43 <sipa> it certainly shouldn't happen invisible to the user, or without choice
315 2012-12-04 16:54:45 <TD> well, i think having laptop users who constantly connect and disconnect isn't really helpful
316 2012-12-04 16:54:50 <BlueMatt> TD: making it painless is important, whether conscious or not
317 2012-12-04 16:55:01 <TD> BlueMatt: why would you need to stop using the SPV client? just run full node as an independent thing with no wallet
318 2012-12-04 16:55:29 <TD> i mean, if you run a full node, you should ideally at least know what you're doing, be willing to upgrade occasionally, have a non-firewalled connection, etc
319 2012-12-04 16:55:33 <TD> and ideally be running it on a server
320 2012-12-04 16:55:44 <BlueMatt> because people are going to be more willing to run a full node if they dont have to do anything (like run a second program)
321 2012-12-04 16:55:55 <TD> think tor relay. having tor relays that appear and disappear randomly isn't that great. you want a bit more dedication than that
322 2012-12-04 16:56:10 <sipa> i wonder whether that isn't what we should aim for in the first place: make all wallet software SPV and SPV-only, and run full nodes as separate and perhaps system-wide daemons without wallet
323 2012-12-04 16:56:24 <BlueMatt> meh, having random disconnections/connections (as long as its not every second) isnt a big deal for us
324 2012-12-04 16:56:26 <sipa> which doesn't mean they can't use the same codebase
325 2012-12-04 16:56:29 <BlueMatt> at least, not as big as it is for tor
326 2012-12-04 16:56:45 <sipa> or that the client software could have a "spawn full node daemon" feature built-in
327 2012-12-04 16:57:26 <BlueMatt> sipa: as long as the spv client then just connects to the local full daemon, though...now we're just back at the wallet/block chain/etc process separation
328 2012-12-04 16:57:41 <sipa> BlueMatt: yes, of course
329 2012-12-04 16:57:53 <sipa> if you run your own full node, you use that as a trusted single peer
330 2012-12-04 16:57:54 <sipa> if you run your own full node, you use that as a trusted single peer
331 2012-12-04 16:58:09 <TD> BlueMatt: it doesn't help much either though.
332 2012-12-04 16:58:44 <BlueMatt> TD: sure it does, take any given peer that was running an spv node and upgrade it, even if its not online 24/7, much better for the network
333 2012-12-04 16:58:45 <BlueMatt> TD: sure it does, take any given peer that was running an spv node and upgrade it, even if its not online 24/7, much better for the network
334 2012-12-04 16:59:03 <BlueMatt> (and connection turnover isnt bad anyway, sipa was looking at doing it randomly just for overall network health anyway)
335 2012-12-04 17:03:36 <BitDev> can some one help me, my public key is 33 bytes long and i want calculate my address... first of all i do is RIPEMD-160(SHA-256(public key)) then i get check sum like this sha256(sha256([0, hash160])) first 4 bytes and then i make base58encode( [hash160, 4 byte checksum]) and its not working - what i am doing wrong?
336 2012-12-04 17:06:45 <TD> BlueMatt: only a little bit better, and it degrades the users experience if they aren't ready for it
337 2012-12-04 17:07:02 <TD> i mean if you're using your computer why would you want it churning away 24/7 using cpu and disk iops
338 2012-12-04 17:07:22 <TD> encouraging users to run full nodes on their crappy low-specced macbook airs or whatever, just limits our scalability in future
339 2012-12-04 17:07:50 <BlueMatt> thats an argument for it being a conscious decision, not to avoid encouraging it in general
340 2012-12-04 17:08:25 <BlueMatt> even a low-speced modernish laptop can run a full node once its caught up reasonably well
341 2012-12-04 17:08:31 <TD> yes, now
342 2012-12-04 17:08:36 <TD> what if tx volumes double again?
343 2012-12-04 17:08:49 <sipa> well, but now is important
344 2012-12-04 17:08:50 <BlueMatt> I I dont forsee us outgrowing moore's law any time soon?
345 2012-12-04 17:08:51 <BlueMatt> I I dont forsee us outgrowing moore's law any time soon?
346 2012-12-04 17:09:17 <sipa> we can't just look at the future we want and say: let's just dobwhat we think will be necessary then
347 2012-12-04 17:09:26 <BlueMatt> sd did it once (sort of) but our average is still pretty low across the growth period (ie excluding the early days)
348 2012-12-04 17:09:50 <sipa> if the usrr experience is bad right now, this may have an impact on the future in general
349 2012-12-04 17:13:06 <TD> yeah, it's hard to know what future growth looks like (if nay)
350 2012-12-04 17:13:08 <TD> any
351 2012-12-04 17:13:39 <TD> but, if tx volume does increase and suddenly half our nodes vanish because they were run by people who weren't really interested in doing so, that'd just compound the issue
352 2012-12-04 17:13:45 <BlueMatt> (part of the upgrade logic could, ofc, also include a downgrade logic)
353 2012-12-04 17:13:51 <TD> i think it should be quite easy to get people to run nodes on spare machines, workstations, servers etc
354 2012-12-04 17:13:55 <TD> tor doesn't seem to have an issue with it
355 2012-12-04 17:14:41 <BlueMatt> yes, we should continue to encourage that, but there is no harm in looking for more from people who run upgrade-able nodes
356 2012-12-04 17:15:26 <BlueMatt> in any case, this discussion seems premature, I think we all agree upgradeablity is a nice feature, but lets wait to discuss the finer points until there is an implementation we can frame the discussion around :)
357 2012-12-04 17:16:48 <TD> ACTION -> home
358 2012-12-04 17:16:57 <BlueMatt> see ya
359 2012-12-04 17:32:01 <jgarzik> heh
360 2012-12-04 17:32:09 <jgarzik> a little matter of a missing comma, and boom, crash
361 2012-12-04 17:32:38 <jgarzik> just proves the one should _always_ put a trailing comma in C lists
362 2012-12-04 17:32:45 <jgarzik> even if it's the last item in an array
363 2012-12-04 17:37:01 <gmaxwell> 09:45 < TD> it still isn't shippable due to lack of upgrade support, i thought?
364 2012-12-04 17:37:14 <gmaxwell> It isn't shippable because its still too risky to the network.
365 2012-12-04 17:38:39 <gmaxwell> I just hit coins corruption a few days that we're not sure of the cause, though there is one fix that might have resolved it. But since I can't reproduce it we're can be sure. It's simply going to take time.
366 2012-12-04 17:39:31 <gmaxwell> We also have the reports of leveldb simply failing on win32 after unclean shutdowns which I don't think anyone has really looked into yet.
367 2012-12-04 17:40:13 <gmaxwell> (considering that BDB now seems to have the same problem ... well, its still a major issue even if similar problems now exist in 0.7.1 :( )
368 2012-12-04 17:41:53 <sipa> gmaxwell: + reports of high CPU/memory usage, resolved by restarting
369 2012-12-04 17:41:57 <sipa> and crashing
370 2012-12-04 17:42:15 <sipa> i really don't feel confortable with the state of the software on windows now
371 2012-12-04 17:44:08 <jgarzik> does the block index receive a lot of _updates_ (as opposed to new records)?
372 2012-12-04 17:44:29 <helo> ACTION makes a note to install next-test on his w8 box at home
373 2012-12-04 17:46:04 <jgarzik> I know the UTXO index receives a lot of updates. Block index seems like it would mainly append new records, lending itself to a flat file.
374 2012-12-04 18:12:37 <sipa> jgarzik: it does get updates (location of undo data, which checks have been performed, ...)
375 2012-12-04 18:12:43 <sipa> jgarzik: but not many
376 2012-12-04 18:13:12 <sipa> if you'd want an append-only format, probably have records with initial header data, and records of type 'update state of block Y to X"
377 2012-12-04 18:16:29 <gmaxwell> You could make it almost never get updates by deferring writing the best block.
378 2012-12-04 18:17:14 <sipa> gmaxwell: sure
379 2012-12-04 18:17:49 <sipa> but you don't want to write just the best blocks
380 2012-12-04 18:18:53 <gmaxwell> I mean that second best blocks are reorgs out much less often than the best ones.
381 2012-12-04 18:19:38 <sipa> sure, you could avoid it most of the time, but you'd still need a format that supports changing states
382 2012-12-04 18:29:17 <jgarzik> nod
383 2012-12-04 18:29:34 <jgarzik> an append-only format with "update X" records would work
384 2012-12-04 18:35:28 <midnightmagic> does anyone have any experience with hudson?
385 2012-12-04 18:37:37 <sipa> jgarzik: if you use things like delayed writing, you can even gain significant amount of space (not really relevant compared to block and coindb size, but maybe useful for SPV nodes): if you have multi-block records, you can drop the hashPrev, just store timediff (with a varint, 1-2 bytes is enough), and drop bits
386 2012-12-04 18:38:36 <jgarzik> sipa: true, but I doubt such compression gains much additional value over a more simplistic design
387 2012-12-04 18:39:19 <jgarzik> might gain more if used for CCoins index
388 2012-12-04 18:39:36 <sipa> ?
389 2012-12-04 18:41:47 <midnightmagic> er.. nevermind the hudson question, just realised it "is" jenkins.
390 2012-12-04 18:42:34 <BCBot2`> owowo: Error: "mtgo" is not a valid command.
391 2012-12-04 18:42:34 <owowo> !mtgo woops
392 2012-12-04 18:42:35 <gribble> Error: "mtgo" is not a valid command.
393 2012-12-04 18:42:41 <sipa> jgarzik: do you have any idea how much effort it took to squeeze the bits out of CCoins already? :p
394 2012-12-04 18:42:58 <owowo> sry,.. wrong channel...
395 2012-12-04 19:26:21 <jaromil> ACK for the valgrinding picocoin already, off my TODO list (which anyway looks more like a collection of ex-voto on a Dias del Muertos night)
396 2012-12-04 19:37:22 <Diapolo> what's the plan for 0.7.2?
397 2012-12-04 19:38:31 <sipa> there are only 1.5 gitian builds for 0.7 .2rc2
398 2012-12-04 19:39:37 <sipa> (one by me, and half of onr by luke)
399 2012-12-04 19:40:25 <Diapolo> was wondering as there was no more discussion in the Github thread ...
400 2012-12-04 19:40:54 <Diapolo> seems some of the other core devs are a little rare currently?
401 2012-12-04 19:42:01 <sipa> yes, an endangered species we are
402 2012-12-04 19:43:29 <Diapolo> :-P at least you are pushing
403 2012-12-04 19:44:09 <D34TH> needs more 1549
404 2012-12-04 19:46:46 <D34TH> oh yea, add -benchmark
405 2012-12-04 19:46:47 <D34TH> oh yea, add -benchmark
406 2012-12-04 19:47:40 <Diapolo> od1n makes me want to scream (Github issue) ^^ argh
407 2012-12-04 19:48:04 <Diapolo> agreed, add -benchmark can't hurt
408 2012-12-04 19:56:15 <Luke-Jr> Diapolo: set up gitian so you can build!
409 2012-12-04 19:57:22 <Diapolo> I tried that once and found out the setup process was not that straight-forward
410 2012-12-04 19:57:29 <sipa> Luke-Jr: did you do a linux build yet?
411 2012-12-04 19:57:43 <Luke-Jr> sipa: my gitian VM is only 32-bit
412 2012-12-04 19:57:48 <sipa> Diapolo: that's an understatement, unfortunately
413 2012-12-04 19:57:54 <sipa> Luke-Jr: why?
414 2012-12-04 19:58:21 <maaku> gmaxwell: if that extra gig of space is too valuable, then don't run a utxo-tree node for your miners
415 2012-12-04 19:58:45 <Luke-Jr> sipa: because I didn't expect it to demand 64-bit when I built it
416 2012-12-04 19:58:46 <Luke-Jr> sipa: because I didn't expect it to demand 64-bit when I built it
417 2012-12-04 19:58:47 <maaku> the beauty of a meta-chain is its entirely optional
418 2012-12-04 19:59:02 <gmaxwell> maaku: ... then the utxo can't be normative and validated. And its not just an extra gig, its a ~doubling of the space a full node needs.
419 2012-12-04 19:59:10 <gmaxwell> Nope. If it's optional it's pointless.
420 2012-12-04 19:59:34 <D34TH> hmm
421 2012-12-04 19:59:35 <gmaxwell> If its correctness isn't enforced then there is no reason to trust it. Might as well just skip the hash tree structure entirely.
422 2012-12-04 19:59:37 <maaku> gmaxwell: by my calculations, it would be about a gig of storage
423 2012-12-04 19:59:44 <Diapolo> Luke-Jr: If someone is willing to write or extend the current how-to I'm fine with participating, but I can't even count the hours it took me to only setup a Qt / MinGW Build chain with all libs...
424 2012-12-04 19:59:46 <D34TH> i think im ready to propose another pullreq
425 2012-12-04 20:00:25 <Luke-Jr> Diapolo: 1. checkout gitian git code; 2. do something; 3. ????; 4. profit!
426 2012-12-04 20:00:32 <maaku> gmaxwell: i think you misunderstand--it's optionally either used or not, but if it is used its correctness is always enforced
427 2012-12-04 20:00:37 <gmaxwell> maaku: a gig of storage for the current chain? congrats you've increased the required space by 9x instead of the 2x I was assuming!
428 2012-12-04 20:00:52 <gmaxwell> maaku: if its correctness is enforced then you can't mine without it.
429 2012-12-04 20:01:06 <D34TH> pullreq 2073
430 2012-12-04 20:01:07 <Diapolo> Luke-jr: sorry? don't get it...
431 2012-12-04 20:01:10 <gmaxwell> (nor can you fully validate)
432 2012-12-04 20:01:12 <Luke-Jr> me either
433 2012-12-04 20:01:36 <maaku> gmaxwell: it's a *meta-chain* it's not part of the block chain!
434 2012-12-04 20:01:49 <maaku> anyone can mine a bitcoin block without updating the meta chain
435 2012-12-04 20:02:22 <maaku> and the result would only be that the meta-chain's difficulty is correspondingly lower
436 2012-12-04 20:02:44 <gmaxwell> maaku: Thats worthless. It doesn't actually provide security because there would be zero incentive to not lie in it.
437 2012-12-04 20:03:33 <maaku> miners who do merge-mine the meta-chain would validate it, and reject any blocks containing lies
438 2012-12-04 20:04:41 <gmaxwell> maaku: and the best that can give you is old data, which isn't useful because you need the current UTXO set. And worse, a short reorg of the secondary chain would make the real chain unvalidatable for nodes using the utxo set.
439 2012-12-04 20:06:23 <Diapolo> Luke-jr: I said I don't want to spend a whole lot of time setting up gitian, just because the documentation is not good. What you meant by profit?
440 2012-12-04 20:06:34 <maaku> gmaxwell: the meta-chain would be linked to a bitcoin chain, e.g. "this is that hash of the UTxO index for block #210100"--you then download the UTxO set, convert that into an ultra prune index, and then replay blocks from #210100 onward
441 2012-12-04 20:06:39 <maaku> a couple of blocks out of date isn't an issue
442 2012-12-04 20:08:18 <jgarzik> sipa: "might gain more", BTIM using diff-style techniques to store the CCoins index on disk, as a stream of new-record and update-record operations in append-only format
443 2012-12-04 20:08:47 <jgarzik> versus a diff-style format for the block index, versus simply storing new-record or update-whole-record operations for the block index
444 2012-12-04 20:09:01 <jgarzik> trying to think through the implications of dropping leveldb altogether ;p
445 2012-12-04 20:09:29 <sipa> ah
446 2012-12-04 20:10:13 <Diapolo> yet another db engine :)?
447 2012-12-04 20:11:51 <jaromil> nah he is talking about restructuring how data gets saved
448 2012-12-04 20:12:12 <jaromil> i'd say there is some gain in that direction for sure, not sure how much / worth
449 2012-12-04 20:12:16 <jaromil> sounds more like a filesystem actually
450 2012-12-04 20:13:54 <jaromil> the kind of stuff that, when made wrong, can make programmers go bonkers and killl their wife and shit
451 2012-12-04 20:15:24 <jgarzik> well if we face months of leveldb debugging, it is a pain
452 2012-12-04 20:15:33 <jgarzik> and our use case for block index is rather simple
453 2012-12-04 20:16:22 <jgarzik> we already load all of the block index into RAM
454 2012-12-04 20:17:07 <jgarzik> flat file is straightforward there
455 2012-12-04 20:17:16 <sipa> the only leveldb problem i've seen so far is on windows
456 2012-12-04 20:17:20 <jgarzik> that leaves the {TX hash} -> CCoins index
457 2012-12-04 20:17:55 <jgarzik> sipa: nod. leveldb is woefully under-tested on that platform in general, and that is a large userbase for bitcoin.
458 2012-12-04 20:18:06 <sipa> so it's easy
459 2012-12-04 20:18:10 <sipa> we just drop windows support :p
460 2012-12-04 20:18:15 <jgarzik> hah
461 2012-12-04 20:18:28 <sipa> let's ship a VM with bitcoin preinstalled
462 2012-12-04 20:18:50 <Diapolo> sipa: may a thunder strike your computer ... just NOW :-P
463 2012-12-04 20:18:52 <jgarzik> heh
464 2012-12-04 20:18:53 <sipa> probably cheaper in developer time than support windows :p
465 2012-12-04 20:19:15 <sipa> (i'm joking, if that wasn't clear)
466 2012-12-04 20:19:20 <Diapolo> it was :D
467 2012-12-04 20:19:31 <Diapolo> I hope you understood I was also joking ^^
468 2012-12-04 20:19:44 <sipa> well, the fact that it would be cheaper in developer time was serious, but the suggestion to drop windows wasn't :D
469 2012-12-04 20:20:15 <Diapolo> What's so bad about LevelDB and Win?
470 2012-12-04 20:21:34 <sipa> maybe nothing, but we've seen users with corrupted leveldb's on windows
471 2012-12-04 20:21:38 <sipa> and no where else
472 2012-12-04 20:22:00 <Diapolo> I never observed that yet...
473 2012-12-04 20:22:17 <sipa> well, ideally it shouldn't ever happen
474 2012-12-04 20:22:43 <sipa> but for example on the forum there was some user compaining it at some point used close to 1GB of RAM during IBD
475 2012-12-04 20:22:45 <Diapolo> indeed
476 2012-12-04 20:22:52 <sipa> which is also unexpected
477 2012-12-04 20:23:03 <sipa> and the windows leveldb uses a very different backend
478 2012-12-04 20:23:13 <Diapolo> the client process ate 1GB RAM? I need to take a look at that
479 2012-12-04 20:23:57 <Diapolo> (watch how much it need when I'm testing next IBD)
480 2012-12-04 20:23:58 <Diapolo> (watch how much it need when I'm testing next IBD)
481 2012-12-04 20:24:01 <Diapolo> +s
482 2012-12-04 20:24:51 <jaromil> speaking of RAM, have you looked into redis?
483 2012-12-04 20:24:51 <sipa> jgarzik: well the requirement is very easy... bytearray to bytearray map with atomic updates :)
484 2012-12-04 20:26:33 <TD> jgarzik: leveldb is used in chrome, it's hardly untested. the issue is chrome uses a different platform mapping
485 2012-12-04 20:26:51 <maaku> i was about to suggest redis as well :)
486 2012-12-04 20:26:52 <sipa> chrome uses its own backend
487 2012-12-04 20:27:00 <TD> anyway, i don't think you're going to beat it with a custom implementation. if there are issues on windows, someone is just going to have to sit down and do some debugging with it
488 2012-12-04 20:27:11 <TD> well "backend" - you mean classes that implement file IO and threading
489 2012-12-04 20:27:20 <jgarzik> yes
490 2012-12-04 20:27:35 <jgarzik> IOW the "platform stuff" that runs on Windows and not Linux
491 2012-12-04 20:27:40 <Diapolo> if it's used in Chrome it's rather well tested on Windows no?
492 2012-12-04 20:27:49 <jgarzik> Diapolo: see what sipa said
493 2012-12-04 20:28:08 <Diapolo> I commented before proceeding to read further...
494 2012-12-04 20:28:10 <TD> chrome has its own OS abstraction layer that the leveldb integration there uses. if there's a corruption issue, it's likely to be some subtle detail of how file IO or threading is mapped to win32
495 2012-12-04 20:28:20 <jgarzik> yes
496 2012-12-04 20:28:20 <TD> ie, not in the actual leveldb code
497 2012-12-04 20:28:28 <sipa> yes, quite sure about that
498 2012-12-04 20:28:30 <jgarzik> a distinction without difference
499 2012-12-04 20:28:43 <jgarzik> it's filed under leveldb, since it's not bitcoin
500 2012-12-04 20:28:49 <sipa> we should look into upgrading to leveldb 1.7
501 2012-12-04 20:29:08 <jgarzik> a lot of LOC just for one index
502 2012-12-04 20:29:11 <TD> it can't really hurt
503 2012-12-04 20:29:29 <sipa> jgarzik: i think leveldb is quite small, compared so some other systems :)
504 2012-12-04 20:29:31 <TD> jgarzik: you could say the same about bigtable
505 2012-12-04 20:29:33 <sipa> in terms of lines of code
506 2012-12-04 20:29:59 <Diapolo> TD: any changelog link for me I love reading such stuff ^^
507 2012-12-04 20:30:46 <sipa> https://code.google.com/p/leveldb/downloads/detail?name=leveldb-1.7.0.tar.gz&can=2&q=
508 2012-12-04 20:30:55 <sipa> https://code.google.com/p/leveldb/downloads/detail?name=leveldb-1.6.0.tar.gz&can=2&q=
509 2012-12-04 20:31:20 <sipa> not much is changed, but it seems the build environment should be more flexible now
510 2012-12-04 20:31:36 <sipa> so we may less ad-hoc patches
511 2012-12-04 20:31:37 <jgarzik> I dunno, seems like you could do an index in less than 24kLOC
512 2012-12-04 20:32:48 <TD> sure. but it'll probably suck.
513 2012-12-04 20:32:49 <sipa> TD: i really wonder what awful experience you had with linux on the desktop - i really feel the opposite... i never expected it to become a competitor to mainstream desktops, but there are certainly more users now than 10y ago
514 2012-12-04 20:33:08 <TD> sipa: on all the graphs i've seen it never made it beyond 1%
515 2012-12-04 20:33:12 <TD> (of usage)
516 2012-12-04 20:33:14 <sipa> so?
517 2012-12-04 20:33:20 <TD> so that's not a successful product
518 2012-12-04 20:33:48 <Diapolo> TD: what's your favorite OS? Mac?
519 2012-12-04 20:33:54 <TD> i hate all of them :)
520 2012-12-04 20:34:10 <TD> at work i use linux. it's alright, given i only need to run a command line and emacs normally
521 2012-12-04 20:34:19 <TD> (and intellij for when i work on java stuff)
522 2012-12-04 20:34:28 <TD> it's not anything i'd recommend to non-geeks
523 2012-12-04 20:34:48 <Diapolo> you hate all of them ^^ alright
524 2012-12-04 20:34:56 <D34TH> Diapolo, i hate so do this, but how do you change the commit message? i have never had to change it
525 2012-12-04 20:34:59 <sipa> neither do i recommend it to others frequently, but for me it's the best choice i know of
526 2012-12-04 20:35:04 <TD> sure, for you
527 2012-12-04 20:35:32 <sipa> yes, if it's a good choice for me and some others, but not for yet others... how is that unsuccesful? it's not dying
528 2012-12-04 20:35:41 <TD> the market spoke clearly though - the vast majority of people i know who are not programmers are now using macs, a few are on windows
529 2012-12-04 20:35:54 <sipa> sure, it lives in a niche
530 2012-12-04 20:35:59 <TD> well, i'd define the word successful to mean more than "not dying" :)
531 2012-12-04 20:36:11 <sipa> evolving is probably better
532 2012-12-04 20:36:17 <sipa> OS/2 hasn't died either
533 2012-12-04 20:36:18 <TD> a platforms success is really about how many users it can attract. otherwise people don't bother writing apps for it
534 2012-12-04 20:36:22 <Diapolo> git rebase origin -i and then reword
535 2012-12-04 20:36:29 <TD> same for a currency
536 2012-12-04 20:36:35 <TD> network effects ??ber alles
537 2012-12-04 20:37:01 <sipa> network effect certainly is important
538 2012-12-04 20:37:07 <TD> a ton of companies and individuals put a ton of work into desktop linux, not for themselves but to try and make it a successful product
539 2012-12-04 20:37:10 <D34TH> Diapolo, i did it through the github editor
540 2012-12-04 20:37:21 <TD> in the end those efforts were basically a failure. "linux for the family" never happened
541 2012-12-04 20:37:34 <Diapolo> D34TH: is it saved then into the commit, dunno
542 2012-12-04 20:37:34 <sipa> but in my opinion, linux is doing good... certainly better than 10y ago
543 2012-12-04 20:37:54 <sipa> hell, even game companies are developing for linux now
544 2012-12-04 20:37:56 <TD> i used to work for codeweavers, that was a little company that tried to build its fortune by making products for desktop linux users. when apple announced x86 mac they were very happy and ported their stuff immediately
545 2012-12-04 20:38:07 <sipa> who would even have dreamt about that 10y ago?
546 2012-12-04 20:38:22 <TD> well, i've been using linux for 10 years and little has changed. yes, games companies were releasing for linux 10 years ago
547 2012-12-04 20:38:35 <TD> see http://en.wikipedia.org/wiki/Loki_Software
548 2012-12-04 20:38:38 <sipa> yeah, tuxracer
549 2012-12-04 20:38:44 <TD> Although successful in its goal of bringing games to the Linux platform, the company was eventually forced to close due to financial troubles,[3] with it declaring Chapter 11 Bankruptcy Protection in August 2001,[4][5] and finally being disbanded in January 2002.
550 2012-12-04 20:39:11 <TD> yep, loki went bankrupt almost exactly 11 years ago
551 2012-12-04 20:39:20 <ByronJohnson> doublec: go'i
552 2012-12-04 20:39:25 <sipa> i feel you're completely off if you consider the only success scenario to be mainstream acceptance
553 2012-12-04 20:40:00 <Diapolo> I fear game development will switch over to all those mobile and tablet things, as core-gaming is not where they make their money ... so Linux and gaming, I don't think that will cause a raise in users
554 2012-12-04 20:40:02 <helo> ACTION busts out his 'loki: games linux people play' tshirt
555 2012-12-04 20:40:15 <sipa> Diapolo: indeed, it won't
556 2012-12-04 20:40:19 <TD> well, it depends on your goals for sure, but how could you argue a backwater system hardly anyone uses is a "success" without getting into "well it entertained me personally for a while" type definitions
557 2012-12-04 20:40:59 <TD> my definition of success is, would the man on the street agree that it was a success
558 2012-12-04 20:41:25 <maaku> ...meanwhile, Valve releases Steam client for Linux...
559 2012-12-04 20:41:43 <sipa> ^
560 2012-12-04 20:41:59 <TD> Diapolo: in fairness to linux, the games industry pretty much abandoned the entire PC/Mac market in favor of consoles years ago
561 2012-12-04 20:42:06 <Diapolo> they have enough money to do such adventures ^^ EA would not do this :D
562 2012-12-04 20:42:07 <Diapolo> they have enough money to do such adventures ^^ EA would not do this :D
563 2012-12-04 20:42:22 <Diapolo> TD: that is what I was trying to say ^^
564 2012-12-04 20:42:35 <TD> Diapolo: though linux has always been behind even macos for gaming. not that this is the only metric that matters
565 2012-12-04 20:42:45 <Diapolo> and I hate mobile gaming -_-
566 2012-12-04 20:46:25 <gmaxwell> TD: My definiton of success is the Haber process.
567 2012-12-04 20:46:53 <TD> the fertilizer process?
568 2012-12-04 20:47:06 <TD> because it led to the green revolution etc
569 2012-12-04 20:47:07 <TD> ?
570 2012-12-04 20:48:36 <gmaxwell> Yes. Something on the order of 2 billion extra people exist because of the haber process. Nitrogen fixing is one of the limiting factors for life on earth, and its the major bottleneck in industrial farming in most of the world.
571 2012-12-04 20:48:43 <TD> btw i disagree that android is linux. yes it uses the same kernel (mostly, it's quite heavily forked). but you can't write an app that runs on both android phones and ubuntu. everything other than the kernel is completely different.
572 2012-12-04 20:48:53 <TD> hmm
573 2012-12-04 20:49:01 <TD> not sure bitcoin can create an additional 2 billion people :)
574 2012-12-04 20:49:28 <Diapolo> so let's create 2 billion coins ... well no
575 2012-12-04 20:49:31 <TD> i'd be happy with it taking a good chunk out of paypal
576 2012-12-04 20:49:39 <TD> i mean, i don't feel like bitcoin has to become the one world currency or anything
577 2012-12-04 20:49:48 <gmaxwell> TD: I do agree somewhat with your dis-agreement on android. Android isnt't really something functionally equal to linux, the kernel is invisible to most people and largely 'incidental'??? although the android system is a very open system compared to the alternatives which was part of the linux desktop story.
578 2012-12-04 20:49:51 <TD> but i'd like it to eventually at least be a household name used by people who don't care about technology or politics
579 2012-12-04 20:50:04 <TD> (and who aren't a tiny specialized minority, like drug dealers)
580 2012-12-04 20:50:05 <gmaxwell> I think very few people would be sad if we missed the linux desktop and got the freebsd desktop or the opensolaris desktop instead. :P
581 2012-12-04 20:50:13 <TD> heh
582 2012-12-04 20:50:20 <TD> i'm hoping one day android on the desktop happens
583 2012-12-04 20:50:39 <gmaxwell> seems right now that the fad mongers think the desktop is dead. 0_o
584 2012-12-04 20:50:41 <TD> macos is heading into some kind of dystopian fairy tale world. windows 8 is just bizarre. unity broke my desktop and i didn't really forgive it yet
585 2012-12-04 20:50:50 <Diapolo> Paypal needs a kick in it's ass their model is pure arbitrariness
586 2012-12-04 20:51:11 <TD> the history of paypal is interesting
587 2012-12-04 20:51:17 <sipa> age grading :p
588 2012-12-04 20:51:26 <TD> when it was first started, it was way closer to bitcoin in terms of its philosophy and politics than most people realize
589 2012-12-04 20:51:33 <TD> they were borderline crypto anarchists
590 2012-12-04 20:51:35 <gmaxwell> TD: I dunno, how household a word is most of the most important financial instruments until they fail?
591 2012-12-04 20:52:04 <TD> then size happened, and with it craptons of regulation
592 2012-12-04 20:52:15 <Diapolo> windows 8 metro is the most ugly, user-unfriedly and unneeded crap I've ever seen
593 2012-12-04 20:52:56 <sipa> at least it is evolution
594 2012-12-04 20:52:58 <helo> i don't know anyone that uses metro... step 1: click to desktop
595 2012-12-04 20:53:00 <TD> gmaxwell: i'm not sure a payment system is really like a financial instrument. but things like visa, mastercard, paypal, maestro, $BANK etc are all pretty much household names because people interact with them a lot
596 2012-12-04 20:53:03 <sipa> (note: haven't touched or tried it yet)
597 2012-12-04 20:53:08 <gmaxwell> I guess a fundimental point... I don't think the basic design (blockchain consensus) of bitcoin makes for a great end user technology. I don't mind it being a better one, but I don't see much point in trying to cram it in. So I think of infrastructure as a better goal than being a household name.
598 2012-12-04 20:53:25 <TD> sipa: it's pretty cool ???.. on a tablet. i have no clue what they were thinking when they welded it to the side (literally) of the classic windows desktop
599 2012-12-04 20:53:46 <TD> gmaxwell: you mean a backbone for institutions?
600 2012-12-04 20:53:47 <gmaxwell> E.g. I think??? just on technology??? bitcoin is a pretty poor replacement for visa. A currency visa could trade in? sure. But a replacement? It's just ... yuck. :)
601 2012-12-04 20:53:55 <TD> heh
602 2012-12-04 20:54:05 <TD> well i'm hopeful we can make it work. heck, it runs fine on my smartphone.
603 2012-12-04 20:54:12 <TD> and that's before even basic optimizations like bloom filtering is done.
604 2012-12-04 20:54:40 <sipa> gmaxwell: any results from running parallel checking on your 32-core machine? :p
605 2012-12-04 20:54:41 <TD> automatic sync at night made a world of difference
606 2012-12-04 20:54:47 <Diapolo> that Android client is pretty neat yes
607 2012-12-04 20:54:51 <TD> now it's never more than about 30 seconds sync away from head
608 2012-12-04 20:55:06 <TD> even though it's downloading full blocks
609 2012-12-04 20:55:07 <gmaxwell> I like that. And I do want to see that work go further??? but my thinking is that it'll always kinda suck compared to visa. But it means that when visa does offer transactions in BTC they'll have good competition from SPV clients working directly, and so they'll have to step up their pricing, security, and auditablity game.
610 2012-12-04 20:55:21 <TD> yes
611 2012-12-04 20:55:42 <TD> i can easily see the end-game for bitcoin being "and then visa got a lot better and most people ended up using that"
612 2012-12-04 20:56:12 <Diapolo> For comparing these 2 I miss one thing ... a Bitcoin-payment-card or such.
613 2012-12-04 20:56:33 <sipa> Diapolo: but how is a bitcoin payment card any different from a dollar payment card?
614 2012-12-04 20:56:34 <sipa> Diapolo: but how is a bitcoin payment card any different from a dollar payment card?
615 2012-12-04 20:57:03 <gmaxwell> sipa: running the coins rebuild now??? took a while to netsync.
616 2012-12-04 20:57:12 <TD> phones can be a good enough replacement. heck the entire US payment industry wants/wanted to move to using nfc in phones to pay for things
617 2012-12-04 20:57:17 <sipa> i'm not saying that that would be a bad thing, but the reason why it is considered good won't have much to do with bitcoin anyway
618 2012-12-04 20:57:17 <TD> not that it makes a ton of sense for them
619 2012-12-04 20:57:39 <Diapolo> I didn't think, just Visa is a card of plastic for me and I think most other normal users. Perhaps I was somewhere else with my thoughts,
620 2012-12-04 20:58:06 <jgarzik> Fun way to help bitcoin survive an Internet blackout: broadcast the 80-byte block headers via packet radio and other low bandwidth means
621 2012-12-04 20:58:42 <jgarzik> And another tangent: was pondering UDP for inv, tx, and block header messages
622 2012-12-04 20:58:46 <TD> the card format is legacy/backwards compatibility. Visa have introduced keyfobs and the like for doing touch-to-pay
623 2012-12-04 20:59:05 <TD> jgarzik: well, it might be better to let the chain split and have the worlds biggest re-org when connectivity comes back
624 2012-12-04 20:59:14 <TD> jgarzik: not sure if you can do much with just the headers
625 2012-12-04 20:59:24 <Diapolo> TD: Pay-wave and such stuff ... but I have ZERO trust in using that from a company like Visa or Mastercard
626 2012-12-04 20:59:33 <TD> Diapolo: why?
627 2012-12-04 20:59:36 <jgarzik> well, you could header + coinbase TX
628 2012-12-04 20:59:43 <jgarzik> once coinbase TX has UTXO merkle root
629 2012-12-04 21:00:06 <TD> jgarzik: but you aren't moving transactions in or out, if all that happens is you broadcast headers over the radio.
630 2012-12-04 21:00:12 <Diapolo> big company, intransparent, dunno how their security standards are and of course data-privacy concerns
631 2012-12-04 21:00:36 <TD> Diapolo: right, that applies to all payments even if it's not paywave
632 2012-12-04 21:00:59 <TD> ACTION doesn't recall reading about visa getting hacked directly
633 2012-12-04 21:01:05 <Diapolo> I like hard coins ... which most other Germans love, too. We seem a little special in that case,
634 2012-12-04 21:01:22 <TD> of course the entire card infrastructure makes a sieve look like a cargo tanker
635 2012-12-04 21:01:38 <TD> germany has an unusually low level of credit card penetration. <0.3 iirc
636 2012-12-04 21:01:42 <TD> vs >2 for the USA
637 2012-12-04 21:02:52 <helo> it doesn't matter if visa is insecure, as You Are Protected
638 2012-12-04 21:02:53 <helo> it doesn't matter if visa is insecure, as You Are Protected
639 2012-12-04 21:03:42 <sipa> i wonder: how many people today use bitcoin because of its inherent properties (and not ideology, technology, speculation, ...)
640 2012-12-04 21:03:44 <Diapolo> So perhaps the point to make Bitcoin more popular is it's open and transparent concept, not it's Visa replacement capabilities. That's what is important to me.
641 2012-12-04 21:03:44 <gmaxwell> jgarzik: Old C-band sat bandwidth is not terribly expensive. I got a quote a while back for $350/month for 100KHz bandwidth on a bird that should be visible to all of north america, central america, and some of south america.
642 2012-12-04 21:04:17 <gmaxwell> and 100KHz is wide enough that you could fit the blockchain maxrate there with low order modulation and lots of FEC and hopefully pick it up with a fairly small dish even though its cband.
643 2012-12-04 21:04:45 <TD> sipa: probably not many
644 2012-12-04 21:05:04 <Diapolo> TD: oh and I don't like to get profiled, I don't want Visa to know what I bought, when, where, when and all such stuff ... I don't trust in american companies data-protection policies ^^
645 2012-12-04 21:05:06 <TD> sipa: that said, i know one guy at google zrh does buy coins off me semi-regularly. he says he buys things online with them as he doesn't have a credit card
646 2012-12-04 21:05:13 <gmaxwell> jgarzik: thats even better bitcoin security than packet radio??? but people would need to invest in the infrastructure in advance.
647 2012-12-04 21:05:23 <TD> sipa: so i guess it's at least 1 :)
648 2012-12-04 21:05:34 <sipa> TD: well, i suppose SD players count as well...
649 2012-12-04 21:05:42 <TD> that's true
650 2012-12-04 21:07:20 <jgarzik> gmaxwell: yep, I mentioned satellite in thread: https://bitcointalk.org/index.php?topic=128702.msg1378305#msg1378305
651 2012-12-04 21:07:26 <jgarzik> ACTION whips out the FidoNet reference
652 2012-12-04 21:08:23 <gmaxwell> jgarzik: bandwidth on ku band stuff is much more expensive though. :(
653 2012-12-04 21:08:59 <gavinandresen> sipa or luke-jr : trying to gitian-build 0.7.2rc2 I get this: http://pastebin.com/T2DjEXXL
654 2012-12-04 21:09:31 <gavinandresen> my ../bitcoin directory does have the 0.7.2rc2 tag from gitorious in it, checked out
655 2012-12-04 21:09:51 <Diapolo> *damn, why can't I connect to IPv6 nodes*
656 2012-12-04 21:10:12 <sipa> gavinandresen: gitian tries to download, it doesn't copy
657 2012-12-04 21:10:34 <Luke-Jr> gavinandresen: can you confirm the gitian.yml file is pointing at the gitorious repo?
658 2012-12-04 21:10:35 <sipa> gavinandresen: i have a script that fetches in my normal repo dir, and then injects into gitian
659 2012-12-04 21:11:09 <sipa> i suppose what Luke-Jr says is te problem indeed
660 2012-12-04 21:11:11 <Diapolo> TD: https://ipv6.google.com/ should be reachable via 6to4, right?
661 2012-12-04 21:11:11 <gavinandresen> Luke-Jr: yes, the gitian.yml in ../bitcoin/contrib/etc points to gitorious
662 2012-12-04 21:11:45 <sipa> Diapolo: sure
663 2012-12-04 21:12:06 <gmaxwell> sipa: the utilization over 193k is really low. :(
664 2012-12-04 21:12:11 <gmaxwell> sipa: like 500%
665 2012-12-04 21:12:47 <Luke-Jr> gavinandresen: I seem to recall a bug in newer gitians where it refused to use the provided URL once one had been established - maybe?
666 2012-12-04 21:12:48 <sipa> gmaxwell: quite expected
667 2012-12-04 21:12:50 <gavinandresen> Luke-Jr: mac 0.7.2rc2 build is at https://s3.amazonaws.com/gavinandresen-bitcoin/bitcoin-0.7.2rc2-macosx.dmg
668 2012-12-04 21:13:03 <Luke-Jr> if that's the case, copying the tag to github/master repo will "workaround" it
669 2012-12-04 21:13:12 <Diapolo> sipa: seems the last FW update for my router broke IPv6 ...
670 2012-12-04 21:13:35 <gmaxwell> sipa: hm. suspending a sched_idle computing job increased it to 800% with bursts to 1000%.
671 2012-12-04 21:13:41 <Luke-Jr> devrandom: poke
672 2012-12-04 21:13:52 <gavinandresen> Luke-Jr: I'd rather it get fixed, or I'll waste another hour of my life trying to figure out why it doesn't work next time.
673 2012-12-04 21:14:08 <sipa> gavinandresen: i got tired of it long ago :)
674 2012-12-04 21:14:34 <sipa> i just want to build using the tags in my own repo
675 2012-12-04 21:14:35 <Luke-Jr> ACTION ended up just sticking to his old gitian version :/
676 2012-12-04 21:14:36 <Luke-Jr> ACTION ended up just sticking to his old gitian version :/
677 2012-12-04 21:14:50 <gavinandresen> I'm very tempted to cherry-pick the move bug and just spin a 0.7.2 final and be done with it.
678 2012-12-04 21:15:39 <gmaxwell> sipa: still just 20-25% in most of the threads.
679 2012-12-04 21:15:49 <gavinandresen> Luke-Jr: I don't understand your 0.7.2rc2 commit history; why all the "merge from old releases" commits that do nothing?
680 2012-12-04 21:17:12 <Luke-Jr> gavinandresen: try deleting inputs/bitcoin/, that should force gitian to refetch
681 2012-12-04 21:17:42 <jgarzik> TD: I would think there would be -some- value in multiple operators broadcasting their notion of current best block [header]
682 2012-12-04 21:17:42 <sipa> gmaxwell: i'm not so much worried about not maximizing utilization... we knew that wouldn't be possible with this mechanism
683 2012-12-04 21:17:43 <sipa> gmaxwell: i'm not so much worried about not maximizing utilization... we knew that wouldn't be possible with this mechanism
684 2012-12-04 21:17:44 <gavinandresen> Luke-Jr: that seems to be working.
685 2012-12-04 21:18:25 <sipa> gmaxwell: however, if saying 32 threads makes your system 80% loaded, for only a 3x speed gain compared to 1 thread... there is a problem
686 2012-12-04 21:18:31 <Luke-Jr> looks like I reported this 2 months ago: https://github.com/devrandom/gitian-builder/issues/26 -.-
687 2012-12-04 21:18:45 <sipa> gavinandresen: http://bitcoin.sipa.be/builds/bitcoin-build.sh.txt is my script
688 2012-12-04 21:19:00 <sipa> it builds from whatever is in ../bitcoin-build
689 2012-12-04 21:19:18 <sipa> (and it also signs, creates packages, some statistics, ...)
690 2012-12-04 21:19:19 <gavinandresen> sipa: spiffy
691 2012-12-04 21:19:34 <Luke-Jr> sipa: nice
692 2012-12-04 21:19:49 <gmaxwell> sipa: it's not responsive to sigterm .. at all. in this config (32 threads, bit dbcace) :P (trying to restart it to run again with the system totally unloaded)
693 2012-12-04 21:20:21 <jgarzik> huh. hash rate took a nose-dive, and price is going up.
694 2012-12-04 21:20:37 <sipa> gmaxwell: yeah, noticed that myself
695 2012-12-04 21:20:52 <gmaxwell> I think it's the dbcache, as I didn't notice that on the other systems.
696 2012-12-04 21:21:09 <sipa> gmaxwell: i think it's just not checking fShutdown
697 2012-12-04 21:21:44 <sipa> hmm, it does
698 2012-12-04 21:24:44 <sipa> oh, it should check fRequestShutdown instead
699 2012-12-04 21:24:45 <sipa> oh, it should check fRequestShutdown instead
700 2012-12-04 21:25:16 <sipa> lolwut???
701 2012-12-04 21:25:35 <sipa> the network message handler threads polls fRequestShutdown to initialize the shutdown process? :o
702 2012-12-04 21:25:36 <sipa> the network message handler threads polls fRequestShutdown to initialize the shutdown process? :o
703 2012-12-04 21:25:58 <TD> hmm
704 2012-12-04 21:26:06 <TD> i wonder if this is the anticipated adjustment to 25 BTC rewards
705 2012-12-04 21:26:25 <maaku> TD? that happened a week ago
706 2012-12-04 21:26:32 <TD> yes i know
707 2012-12-04 21:26:44 <TD> i mean, the falling speed and the rising price is what you'd expect to see when inflation suddenly halves
708 2012-12-04 21:27:23 <gmaxwell> TD: dunno that speed has actually fallen, dips of this magntiude on the graph happen every week or two.
709 2012-12-04 21:27:25 <sipa> rising price.. unsure
710 2012-12-04 21:27:34 <sipa> falling speed seems expectable
711 2012-12-04 21:27:34 <TD> guess we'll see soon enough
712 2012-12-04 21:27:57 <gmaxwell> ;;bc,ticker
713 2012-12-04 21:27:58 <gribble> Error: "bc,ticker" is not a valid command.
714 2012-12-04 21:28:00 <gmaxwell> ;;ticker
715 2012-12-04 21:28:01 <gribble> BTCUSD ticker | Best bid: 13.44001, Best ask: 13.49300, Bid-ask spread: 0.05299, Last trade: 13.49490, 24 hour volume: 59795.42159723, 24 hour low: 12.55000, 24 hour high: 13.50000, 24 hour vwap: 12.98225
716 2012-12-04 21:28:02 <sipa> note that my graph tend to exaggerate some sudden changes
717 2012-12-04 21:28:26 <TD> yeah
718 2012-12-04 21:28:36 <gmaxwell> GPU mining is still profitable over power for me, but not very much.
719 2012-12-04 21:28:47 <maaku> price already rose 2x in the 6mo period leading up to december.. i wouldn't expect it to rise much more because we actually reached halving day
720 2012-12-04 21:29:20 <sipa> maaku: seems reasonable, but i wouldn't have made a bet on that beforehand
721 2012-12-04 21:29:35 <sipa> the bitcoin market isn't particularly rational, i think :)
722 2012-12-04 21:29:45 <maaku> sipa: me neither :)
723 2012-12-04 21:30:07 <gmaxwell> sipa: more rational than some of the pundits!
724 2012-12-04 22:01:06 <midnightmagic> ACTION is having trouble reading past the visa-using-bitcoin discussion without saying something derogatory
725 2012-12-04 22:03:16 <maaku> midnightmagic: just mentally replace "visa" with "my future visa-killing startup" and it reads better
726 2012-12-04 22:08:12 <gmaxwell> meh
727 2012-12-04 22:08:13 <gmaxwell> 12/04/12 22:20:38 SetBestChain: new best=00000000839a8e6886ab5951d76f411475428afc90947ee320161bbf18eb6048 height=1 work=8590065666 tx=2 date=01/09/09 02:54:25
728 2012-12-04 22:08:16 <gmaxwell> 12/04/12 23:03:39 SetBestChain: new best=000000000000048b95347e83192f69cf0366076336c639f9b7228e9ba171342e height=210000 work=628963747775700992096 tx=9344662 date=11/28/12 15:24:38
729 2012-12-04 22:08:16 <midnightmagic> :-)
730 2012-12-04 22:08:17 <midnightmagic> :-)
731 2012-12-04 22:08:49 <sipa> 43 minutes? :o
732 2012-12-04 22:09:20 <midnightmagic> i thought you had it down to just a few dozen minutes?
733 2012-12-04 22:09:21 <midnightmagic> i thought you had it down to just a few dozen minutes?
734 2012-12-04 22:10:08 <sipa> depends what we're benchmarking, with which options, on which machine, with which settings :)
735 2012-12-04 22:12:25 <sipa> best result so far: 12 threads, rebuild of coindb until 210000 blocks, hexacore xeon CPU, 2 GB dbcache, Hal's optimized signature verification, scripts only verified after checkpoint: 13m51s
736 2012-12-04 22:13:23 <sipa> the same machine did a more-or-less noncheating sync from network in 37 minutes
737 2012-12-04 22:13:50 <midnightmagic> that's pretty cool.
738 2012-12-04 22:17:36 <bonks> So I can't find much about the public notes on blockchain.info. Is this a feature of theirs or the bitcoin protocol?
739 2012-12-04 22:17:47 <sipa> theirs.
740 2012-12-04 22:17:57 <D34TH> now tar.xz that and upload it for people
741 2012-12-04 22:17:58 <D34TH> :D
742 2012-12-04 22:18:12 <sipa> D34TH: tar.xz what?
743 2012-12-04 22:18:18 <sipa> my machine?
744 2012-12-04 22:18:26 <bonks> sipa: Ah okay. So it's a comment when you send coins from their online wallet service
745 2012-12-04 22:18:39 <sipa> bonks: and it's stored in their database
746 2012-12-04 22:18:54 <bonks> Ok thanks. What is the comment and comment-to in bitcind used for?
747 2012-12-04 22:19:42 <sipa> they used to be used for send-to-IP transactions i think, now they're mostly useless
748 2012-12-04 22:20:06 <sipa> you can set them for outgoing transactions when using the RPC interface
749 2012-12-04 22:20:08 <sipa> that's it
750 2012-12-04 22:21:30 <sipa> bonks: note that the payment protocol which is being developed on top of bitcoin will have messages attached to transactions
751 2012-12-04 22:22:23 <bonks> awesome :D
752 2012-12-04 23:30:54 <jgarzik> ACTION perplexed
753 2012-12-04 23:31:12 <jgarzik> ACTION connects to P2P nodes, and gets "inv" messages but never "version" or "verack"
754 2012-12-04 23:32:35 <jgarzik> maybe this is related to shy-client
755 2012-12-04 23:32:40 <edcba> very old nodes ?
756 2012-12-04 23:32:46 <jgarzik> maybe my picocoin client should _not_ stuff "version" into the send buffer
757 2012-12-04 23:33:07 <jgarzik> edcba: am connecting to at least one 0.8 node, and seeing this