1 2013-12-02 00:02:05 <rescrv> warren: but is his state getting corrupted? Or he just assumes he should start fresh?
2 2013-12-02 00:03:58 <warren> rescrv: his state apparently hasn't been corrupted since before your patch. something else leveldb related was crashing without apparent leveldb corruption , but with data invalid for bitcoin itself
3 2013-12-02 00:04:03 <hankcary> good morning to anyone approximately GMT +10
4 2013-12-02 00:04:42 <hankcary> Can I run a possible DDOS mitigation strategy passed someone knowledgeable ?
5 2013-12-02 00:06:01 <hankcary> PING ?
6 2013-12-02 00:06:25 <sevenqueue> something is going on
7 2013-12-02 00:06:41 <sevenqueue> I see errors in the debug log
8 2013-12-02 00:08:41 <nsh> hankcary, maybe try in #bitcoin or #bitcoin-offtopic if it's not about the development of bitcoind (or other implementations) software
9 2013-12-02 00:09:05 <hankcary> are most DDOS attacks based on overloading the BitcoinD process? generally with micro- or invalid transactions ?
10 2013-12-02 00:09:20 <hankcary> or are doing good old fashioned bandwidth floods ?
11 2013-12-02 00:09:50 <hankcary> I was just wondering if the -connect function where you only connect to a specified peer, could be used to sertup a reasonable proxy setup
12 2013-12-02 00:10:57 <hankcary> my thought was to have a ephemeral wallet server on a preloaded AWS template server, and if an inbound proxy server was overloaded, block all inbopund connections ( to try and find valid blocks amidst the chaff) and spawn a new server to pick up the load
13 2013-12-02 00:11:38 <hankcary> my real bitcoind server would then only connect to the proxy peer, which wold switch out, but hopefully always remain online, just via different proxies
14 2013-12-02 00:12:04 <hankcary> does this seem legitimate ?
15 2013-12-02 00:13:28 <phantomcircuit> sipa, is the blockundo database leveldb?
16 2013-12-02 00:13:33 <hankcary> or does anyone have a different forum that may better suit network function of bitcoind discussions ?
17 2013-12-02 00:13:37 <sipa> phantomcircuit: no
18 2013-12-02 00:14:17 <phantomcircuit> sipa, i switched out the mmap WriteableFile with a write() based implementation and now im getting "Failed to write undo data" when reindexing
19 2013-12-02 00:14:36 <sipa> that's strange :)
20 2013-12-02 00:14:39 <sipa> is your disk full? :p
21 2013-12-02 00:14:47 <phantomcircuit> nope
22 2013-12-02 00:14:56 <phantomcircuit> actually it's a vm disk let me check
23 2013-12-02 00:15:10 <phantomcircuit> oh hey
24 2013-12-02 00:15:11 <phantomcircuit> yes it is
25 2013-12-02 00:15:16 <phantomcircuit> brb clearing space
26 2013-12-02 00:16:51 <sipa> ACTION zZzZ
27 2013-12-02 00:20:09 <brandondahler> Can anyone link me to or give me a quick rundown on the databases/on-disk data structures and files referenced by them for bitcoin?
28 2013-12-02 00:29:58 <warren> hankcary: what os are you running?
29 2013-12-02 00:51:53 <pera> can anyone tell me how is it possible to do a mitm on bitcointalk when https is enforced?
30 2013-12-02 00:52:25 <pera> does rapidssl was compromised too? :p
31 2013-12-02 00:55:16 <brandondahler> So long as you are not accepting unknown/compromised SSL certificates, HTTPS ensures mitm attacks are impossible.
32 2013-12-02 00:56:00 <Tykling> brandondahler: well, difficult, with the number of CAs trusted by todays browser impossible is a pretty big word to throw around
33 2013-12-02 00:57:10 <Zarutian> well, that list needs a good pruning
34 2013-12-02 00:57:59 <pera> can we create an authentication protocol using bitcoin signatures?
35 2013-12-02 00:58:17 <c0rw1n> sure, why not
36 2013-12-02 00:58:29 <pera> and then an addon for firefox and chrome
37 2013-12-02 00:58:32 <brandondahler> Nothing any better than trusting CAs or particular web of trusts
38 2013-12-02 00:58:32 <c0rw1n> ask the user to sign a message using bitcoin
39 2013-12-02 01:00:36 <Zarutian> ACTION is on the opinion that we (collectivly) would be better off with yurls (basicly the first part of the domain contains the cryptohash of the linked to site public key.
40 2013-12-02 01:01:10 <brandondahler> Would that allow revoking of private keys
41 2013-12-02 01:02:24 <jakov> Zarutian that sounds good
42 2013-12-02 01:02:28 <jakov> in many situations
43 2013-12-02 01:03:08 <Zarutian> ACTION closes with )
44 2013-12-02 01:03:25 <jakov> how do you tell people about the url to begin with
45 2013-12-02 01:03:37 <jakov> since the hash isnt human memorazable
46 2013-12-02 01:03:41 <Zarutian> same way you tell them of .onion or .i2p sites
47 2013-12-02 01:04:06 <jakov> someone could mitm that
48 2013-12-02 01:04:09 <brandondahler> a random .onion site that contains a wiki of a bunch of other onion sites
49 2013-12-02 01:04:14 <jakov> replace it with their own link
50 2013-12-02 01:04:36 <Zarutian> or copy paste them around through signed email
51 2013-12-02 01:04:53 <jakov> we get infinite regress of keys to trust
52 2013-12-02 01:05:20 <Zarutian> the point is not trying to map realworld names to public keys
53 2013-12-02 01:06:02 <Zarutian> that is rather pointless.
54 2013-12-02 01:06:57 <Zarutian> when a friend or someone you meat face to face gives you an link you only care that that link links to the thing they intented to show you
55 2013-12-02 01:08:32 <Zarutian> but if you insists on squaring Zooko's triangle then just use namecoin to provide the well-known-name to public key mapping
56 2013-12-02 01:12:06 <nsh> Zarutian, where can i read more about this things with the stuff and words and all that?
57 2013-12-02 01:12:26 <nsh> (yurls, zooko's "triangle"?, etc.)
58 2013-12-02 01:12:42 <Zarutian> nsh: http://www.waterken.com/dev/YURL/
59 2013-12-02 01:13:02 <nsh> ty
60 2013-12-02 01:13:12 <Zarutian> nsh: https://en.wikipedia.org/wiki/Zooko%27s_triangle
61 2013-12-02 01:13:34 <nsh> oh, that's a real thing. i thought you were being figurative...
62 2013-12-02 01:13:42 <nsh> ah, right
63 2013-12-02 01:14:04 <Zarutian> nsh: http://www.aaronsw.com/weblog/squarezooko
64 2013-12-02 01:15:07 <nsh> ACTION nods, sighs
65 2013-12-02 02:51:07 <warren> It appears testnet has no miners again?
66 2013-12-02 02:51:46 <Luke-Jr> warren: so mine it?
67 2013-12-02 02:52:21 <saracen> It has begun: https://www.google.co.uk/search?q=%225HpHagT65TZzG1PH3CSu63k8DbpvD8s5ip4nEB3kEsrfg5cvjps%22
68 2013-12-02 02:52:24 <warren> Luke-Jr: I argued for removal of setgenerate, so it would be hypocritical to use it.
69 2013-12-02 02:53:27 <freewil> is there an open debate about removal of setgenerate?
70 2013-12-02 02:53:56 <Luke-Jr> warren: I didn't say to use setgenerate
71 2013-12-02 02:54:00 <kjj> isn't testnet the best argument in favor of keeping it?
72 2013-12-02 02:54:21 <Luke-Jr> kjj: and not a very strong argument reallty
73 2013-12-02 02:55:12 <freewil> when i started looking at litecoin i was kind of annoyed they removed it
74 2013-12-02 02:55:20 <freewil> but cpuminer is a great alternative
75 2013-12-02 02:55:36 <kjj> just remove mention of it from help
76 2013-12-02 02:55:40 <freewil> so it would probably be worth it just to remove some code to be maintained
77 2013-12-02 02:55:44 <warren> freewil: litecoin added it back
78 2013-12-02 02:55:49 <freewil> wha?
79 2013-12-02 02:55:50 <freewil> since when
80 2013-12-02 02:57:31 <freewil> i know its not in the latest release ... 0.8.5.1
81 2013-12-02 02:58:19 <freewil> thats why i use https://github.com/pooler/cpuminer in my litecoin-testnet-box
82 2013-12-02 02:58:27 <freewil> https://github.com/freewil/litecoin-testnet-box
83 2013-12-02 02:59:50 <rebroad> how does one mine on testnet?
84 2013-12-02 03:00:04 <andytoshi> i'll start mining, i've never mined a block on anything before..
85 2013-12-02 03:00:36 <andytoshi> rebroad: run bitcoind with -testnet, then you can use setgenerate :}
86 2013-12-02 03:04:59 <warren> the external cpuminers are much faster than what bitcoin can do itself
87 2013-12-02 03:06:16 <andytoshi> hmm, i'll track one of those down
88 2013-12-02 03:06:36 <kjj> who cares? testnet
89 2013-12-02 03:08:37 <danneu> i'm investing in testnet. lots of gains potential
90 2013-12-02 03:09:37 <freewil> heh i always wondered if testnet would eventually became it's own altcoin with a it's own market
91 2013-12-02 03:09:53 <freewil> but i think that has been killed with multiple resets in new releases of testnet
92 2013-12-02 03:11:10 <alex_fun> set generate is handy to mine with client
93 2013-12-02 03:11:24 <alex_fun> why would u want to remove it? :)
94 2013-12-02 03:12:50 <andytoshi> alex_fun: because it's awful
95 2013-12-02 03:13:26 <toffoo> hi warren 0.8.6-mactest1 just corrupted for me
96 2013-12-02 03:13:31 <gavinandresen> if testnet coins start to become worth something, "we" will reset the chain. Don't trade them, please, it just wastes everybody's time.
97 2013-12-02 03:13:43 <warren> toffoo: oh crap
98 2013-12-02 03:14:16 <toffoo> did not crash
99 2013-12-02 03:14:16 <toffoo> it was upon restart
100 2013-12-02 03:14:23 <toffoo> asking me if I want to rebuild the block database now
101 2013-12-02 03:14:27 <warren> toffoo: do not
102 2013-12-02 03:14:40 <andytoshi> is there a ./configure flag to disable building qt?
103 2013-12-02 03:14:45 <andytoshi> the qt gui*
104 2013-12-02 03:14:45 <warren> rescrv: corruption with your patch ...
105 2013-12-02 03:15:00 <warren> toffoo: please upload this chainstate again in tar.xz
106 2013-12-02 03:15:12 <toffoo> ok
107 2013-12-02 03:15:59 <andytoshi> --with-qt=no i think
108 2013-12-02 03:17:45 <rebroad> I wonder if there is a way to make testnet coins worth negative something to offset the potential of them being worth something..
109 2013-12-02 03:18:11 <andytoshi> right now they are worth grumpy messages from gavin ;)
110 2013-12-02 03:18:39 <warren> rescrv: corruption with this branch this time, no crash
111 2013-12-02 03:18:41 <warren> rescrv: https://github.com/bitcoin/bitcoin/commits/0.8.6
112 2013-12-02 03:18:54 <warren> cfields: ^
113 2013-12-02 03:19:11 <gavinandresen> mmm. If you want to trade a worthless coin for real money, there are PLENTY of options these daysâ¦.
114 2013-12-02 03:19:29 <rescrv> warren: where's the corrupted db?
115 2013-12-02 03:20:37 <freewil> you could always start adding exceptions to testnet, but then it behaves less and less like the main chain
116 2013-12-02 03:20:55 <warren> rescrv: asked him to upload it
117 2013-12-02 03:21:48 <freewil> gavinandresen, ha - good point
118 2013-12-02 03:22:27 <cfields> anyone around running fedora or some non-debian-based distro?
119 2013-12-02 03:22:48 <rescrv> cfields: I have one here (VM)
120 2013-12-02 03:23:52 <cfields> rescrv: mmm, i doubt you'd have all the glitter installed on a vm. I'm looking to know how they choose between qt4/qt5 tools
121 2013-12-02 03:24:59 <cfields> ubuntu uses 'qtchooser'. But it seems like a kludge that the other distro's wouldn't adopt
122 2013-12-02 03:25:00 <alex_fun> gavinandresen: 98 alts on cryptsy :D
123 2013-12-02 03:26:22 <freewil> how can you have a market for these coins that are so easily finney-attacked
124 2013-12-02 03:29:06 <alex_fun> easy
125 2013-12-02 03:29:14 <alex_fun> no one cares to hack them atm
126 2013-12-02 03:29:18 <gavinandresen> Maybe the biggest thing I've learned through bitcoin that I didn't know before: some people w
127 2013-12-02 03:29:24 <alex_fun> cryptsy itself can be double spent easily I think :)
128 2013-12-02 03:29:26 <rebroad> would anyone like some testnet coins?
129 2013-12-02 03:29:56 <alex_fun> gavin coins next? with youtube video to promote them? :D
130 2013-12-02 03:30:16 <gavinandresen> no thanks
131 2013-12-02 03:30:19 <freewil> alex_fun, right, wouldnt it trivial to finney attack one of these altcoins, trade them, and withdraw some BTC
132 2013-12-02 03:30:29 <alex_fun> I wonder if google try to sue users of android coins
133 2013-12-02 03:30:31 <alex_fun> lol
134 2013-12-02 03:30:37 <alex_fun> not that they can
135 2013-12-02 03:30:52 <alex_fun> hey diki ;)
136 2013-12-02 03:31:15 <alex_fun> freewil: depends on their total hash
137 2013-12-02 03:31:23 <alex_fun> you would have to rent alot of gpus ;)
138 2013-12-02 03:31:43 <freewil> im assuming most of these altcoins have a relatively low hash rate
139 2013-12-02 03:31:45 <gavinandresen> I wonder how many of the altcoin exchanges are just long-cons like Sheep Marketplace, waiting until they have enough deposits to pull up stakes and disappear.
140 2013-12-02 03:32:22 <freewil> cryptsy seems to have an actual legal corporation behind it
141 2013-12-02 03:32:29 <freewil> but i have done any actual looking into it
142 2013-12-02 03:32:32 <diki> alex_fun:hello, do I know you?
143 2013-12-02 03:32:39 <alex_fun> gavin satoshi successor coin - satoshi and fontas latest work :P
144 2013-12-02 03:32:40 <freewil> havent*
145 2013-12-02 03:32:44 <toffoo> corrupted chainstate upping, 21mins remaining...
146 2013-12-02 03:32:50 <alex_fun> diki yes from coin development lab
147 2013-12-02 03:32:52 <alex_fun> :)
148 2013-12-02 03:33:00 <diki> I am sorry, I don't remember.
149 2013-12-02 03:33:04 <alex_fun> np
150 2013-12-02 03:33:13 <diki> ah
151 2013-12-02 03:33:18 <alex_fun> I dont mind been not remembered :P
152 2013-12-02 03:33:18 <diki> you mean noobcoin was it?
153 2013-12-02 03:33:22 <alex_fun> yes :D
154 2013-12-02 03:33:22 <gavinandresen> mmm. cryptsy made the mistake (in my humble opinion) of issuing unregulated stockâ¦
155 2013-12-02 03:33:44 <alex_fun> gavinandresen: cryptsy is like fake penny stock thingy
156 2013-12-02 03:33:45 <diki> Yes, I remember now.
157 2013-12-02 03:35:01 <freewil> gavinandresen, but their website has "full compliance" in bold
158 2013-12-02 03:35:05 <freewil> /sarc
159 2013-12-02 03:35:12 <gavinandresen> oh, then it MUST be legit
160 2013-12-02 03:35:16 <cfields> gavinandresen: do you still have your qt5 environment around? For a test to see if it's drop-in replaceable when i get this knocked out?
161 2013-12-02 03:35:18 <gavinandresen> my mistake
162 2013-12-02 03:35:26 <freewil> lolz
163 2013-12-02 03:35:28 <gavinandresen> cfields: yes
164 2013-12-02 03:35:32 <alex_fun> fully complied - clients Google, NASDAQ, Coca Cola
165 2013-12-02 03:35:35 <alex_fun> my new exchange :D
166 2013-12-02 03:35:53 <cfields> gavinandresen: great, thanks
167 2013-12-02 03:36:09 <alex_fun> money can be deposited now and withdrawal function come when there is donations to build it :)
168 2013-12-02 03:36:22 <cfields> toffoo: could you pastebin your log in the meantime to be sure it's worth the trouble of uploading?
169 2013-12-02 03:36:49 <cfields> just the part where it's exiting, ofc
170 2013-12-02 03:37:42 <toffoo> okay 1 moment..
171 2013-12-02 03:37:58 <rebroad> I know a lot of people who are buying litecoins now, purely because they can't afford an entire bitcoin. the psychological need to own a whole coin seems to be the main drive for the alt coins now..
172 2013-12-02 03:38:21 <Cryo> ^ this
173 2013-12-02 03:39:13 <alex_fun> yes
174 2013-12-02 03:39:17 <cfields> i can attest to that as well. Among the top 3 questions is always "how much is a coin worth?". These days, thee answer puts them off
175 2013-12-02 03:39:21 <alex_fun> and satoshi knew it imo
176 2013-12-02 03:39:29 <alex_fun> btc to make crypto popular
177 2013-12-02 03:39:33 <alex_fun> :)
178 2013-12-02 03:39:33 <alex_fun> main task
179 2013-12-02 03:39:42 <kjj> educate them
180 2013-12-02 03:39:49 <alex_fun> and heck its 1000 usd now so reward is plentiful :)
181 2013-12-02 03:39:57 <gmaxwell> well, we could change things to use mbtc but people are aruging over that vs ubtc.
182 2013-12-02 03:39:57 <kjj> compare it to buying a half, quarter or tenth ounce of gold
183 2013-12-02 03:39:59 <rebroad> what we need is a new name for a mBTC... let's call 1 mBTC a REBcoin!
184 2013-12-02 03:40:46 <rebroad> a fixed exchange rate 1000 rebcoins will always equal 1 bitcoin
185 2013-12-02 03:40:55 <cfields> kjj: it's too ingrained. the smaller units need to catch hold.
186 2013-12-02 03:40:57 <rebroad> no need for any alts..
187 2013-12-02 03:41:04 <kjj> Buying 0.1 oz gold involves paying about 30% over spot. Buying fractional BTC has no premium
188 2013-12-02 03:41:18 <rebroad> cfields, they won't... people are dumb... the market is driven by psychology, not logic
189 2013-12-02 03:41:36 <kjj> cfields: if the units are a problem, they can come back at 10k when maybe they aren't so stupid any more
190 2013-12-02 03:42:08 <andytoshi> during the runup to 1000 (when the mailing list was alight regarding ubtc vs mbtc) i agreed with the people saying ubtc..
191 2013-12-02 03:42:15 <rebroad> we need to allow bitcoin to evolve, and not get stiffled by these alt coins and their novelty value of owning a whole coin
192 2013-12-02 03:42:34 <andytoshi> but if the price is gonna hold at ~1000, we can't have individual units worth hundredth of a cent, that is really hard to think about
193 2013-12-02 03:42:39 <alex_fun> :)
194 2013-12-02 03:42:39 <alex_fun> rebroad: yes however many people who buy alts purely cause they feel pissed offf at `missing` btc
195 2013-12-02 03:42:57 <cfields> i'm not arguing for them, of course it's not a reasonable argument. but we're talking about people who are on the verge of learning about it. Can't fault them for not understanding at that point.
196 2013-12-02 03:42:59 <alex_fun> so mbtc anything btc u may have hard time selling to them :)
197 2013-12-02 03:43:39 <gmaxwell> just for pratical reasons, typical transactions are now pretty small values, mbtc just may be easier for people to deal with.
198 2013-12-02 03:43:43 <rebroad> so we're just going to have a series of cryptocoins, all reaching $1000 each, and each one being a successor to the last that did that?
199 2013-12-02 03:44:02 <alex_fun> 1000 unlikely
200 2013-12-02 03:44:05 <gmaxwell> but I'm the last guy to ask I work all day with numbers in the range -1 .. 1. :)
201 2013-12-02 03:44:06 <alex_fun> for new alts
202 2013-12-02 03:44:47 <kjj> gmaxwell: from that range, I'd think you did calculus, but real calculists only use 4 or 5 numbers from that range, not the whole range itself
203 2013-12-02 03:45:20 <rebroad> it's a shame bitcoin wasn't patented... all these alt coins could be stopped then for infringement :)
204 2013-12-02 03:45:51 <alex_fun> dream on ;)
205 2013-12-02 03:45:58 <gmaxwell> kjj: most signal processing work, you work with numbers normalized to ±1 (unless you're doing fixed point stuff)
206 2013-12-02 03:46:19 <cfields> has the idea of a split ever been considered? Similar to stocks? You could argue that it's useless since it's purely psychological, but that doesn't mean psychology should be ignored.
207 2013-12-02 03:46:31 <kjj> cfields: only a billion times on the forums
208 2013-12-02 03:46:52 <cfields> figured as much. I try to avoid that place :)
209 2013-12-02 03:47:20 <kjj> it serves a purpose. nice to have all of the bad ideas in one place
210 2013-12-02 03:47:49 <cfields> actually, that's a really bad idea anyway. Not sure how that made it through my 60sec filter.
211 2013-12-02 03:47:58 <rebroad> we need a new name for mBTC....
212 2013-12-02 03:48:08 <rebroad> it needs to be as easy to say as bitcoin
213 2013-12-02 03:48:23 <rebroad> two syllables..
214 2013-12-02 03:48:53 <rebroad> and a new name for uBTC at the ready also
215 2013-12-02 03:49:30 <rebroad> how about a simple refactor.. let mBTC be the new bitcoin.. just move the decimal point.
216 2013-12-02 03:49:35 <kjj> as always, I propose that we wait for a name to show up
217 2013-12-02 03:49:53 <rebroad> or start counting in Satoshis
218 2013-12-02 03:50:15 <toffoo> warren cfields rescrv debug.log http://pastebin.com/VQbvZDDr
219 2013-12-02 03:50:17 <kjj> the client already counts in satoshis, which is why there are satoshis
220 2013-12-02 03:50:57 <cfields> toffoo: ok, thanks
221 2013-12-02 03:50:59 <rebroad> ok, so no one has missed the boat on satoshis... they are still dirt cheap..
222 2013-12-02 03:51:14 <kjj> and you are free to change your local node to present that value in whatever way you wish, but you are in for an interesting couple of days if you change it so that 1.0!=1.0
223 2013-12-02 03:51:18 <rebroad> let's start making 1 satoshi coins...
224 2013-12-02 03:52:25 <rebroad> or find a way to show numbers without a decimal point... no one likes the idea of a 0.01 bitcoin coin much..
225 2013-12-02 03:52:37 <cfields> kjj: many arches will happily agree that 1.0f != 1.0f
226 2013-12-02 03:52:43 <Cryo> toffoo, is the client compiled with gcc or clang
227 2013-12-02 03:53:17 <cfields> as i'm sure gmaxwell is all-too-aware ;)
228 2013-12-02 03:54:05 <toffoo> Cryo no idea, it's from warren
229 2013-12-02 03:54:22 <toffoo> warren cfields rescrv https://www.dropbox.com/s/gf9cahbqkewt6rx/toffoo0.8.6-mactest1-chainstateCORRUPTED.tar.xz
230 2013-12-02 03:54:24 <cfields> it's gcc
231 2013-12-02 03:54:27 <rebroad> a new name for mBTC.... hmmm... a 10 millibit coin..
232 2013-12-02 03:54:58 <Cryo> just wondering if the database corruption is a clang bug, and if building with gcc would 'fix it'
233 2013-12-02 03:58:37 <shesek> anyone knows if bitcointalk is going to be back any time soon?
234 2013-12-02 03:59:00 <alex_fun> shesek he got plenty of cash the forum owner so ask him
235 2013-12-02 03:59:01 <alex_fun> ;)
236 2013-12-02 03:59:06 <alex_fun> to rent more aws servers
237 2013-12-02 04:00:16 <alex_fun> what if someone ddos bitcoin dns resolvers?
238 2013-12-02 04:00:20 <alex_fun> would that make it stuck?
239 2013-12-02 04:00:23 <alex_fun> for a while
240 2013-12-02 04:00:27 <Luke-Jr> .. no?
241 2013-12-02 04:00:36 <Luke-Jr> they're only used as a last resory
242 2013-12-02 04:00:39 <alex_fun> just wonder how is dns connection work, versus old irc one
243 2013-12-02 04:00:39 <Luke-Jr> resort*
244 2013-12-02 04:00:41 <alex_fun> oki
245 2013-12-02 04:00:59 <alex_fun> I forgot that its indeed last resort :)
246 2013-12-02 04:01:48 <phantomcircuit> sipa, so i went and cleared a few hundred gb of space
247 2013-12-02 04:01:54 <phantomcircuit> and im still getting that same error
248 2013-12-02 04:02:01 <phantomcircuit> but only with my modified leveldb
249 2013-12-02 04:06:12 <Lifeofcray> hey guys
250 2013-12-02 04:06:22 <alex_fun> :)
251 2013-12-02 04:06:24 <alex_fun> hi
252 2013-12-02 04:06:39 <alex_fun> there are girls here too like Alina :D
253 2013-12-02 04:06:48 <Lifeofcray> i have a question, tryign to code a php page with json
254 2013-12-02 04:07:14 <Lifeofcray> do i have to wait for the bitcoin client to finish updating then etwork stuff
255 2013-12-02 04:07:26 <Lifeofcray> before i can run commands such as getnewadress
256 2013-12-02 04:07:47 <andytoshi> Lifeofcray: getnewaddress should work
257 2013-12-02 04:07:58 <Lifeofcray> getinfo() works fine, but as soon as i try to do that everything just breaks
258 2013-12-02 04:09:39 <phantomcircuit> Lifeofcray, getnewaddress should work
259 2013-12-02 04:09:51 <phantomcircuit> what is the exact error?
260 2013-12-02 04:10:04 <Lifeofcray> the php page just loads forever
261 2013-12-02 04:12:51 <Lifeofcray> meh, ill just let it finish anyway, continue tomorrow
262 2013-12-02 04:12:59 <andytoshi> i bet it'll work in a couple minutes..
263 2013-12-02 04:13:06 <andytoshi> but yeah, best to let it sync
264 2013-12-02 04:13:11 <phantomcircuit> possibly it's blocking waiting for a lock on the wallet
265 2013-12-02 04:13:29 <Lifeofcray> i have no idea what that means, i just started doing this today
266 2013-12-02 04:13:55 <phantomcircuit> Lifeofcray, the rpc stuff acquires a lock, so the blocksync might have that lock
267 2013-12-02 04:14:02 <Lifeofcray> ah
268 2013-12-02 04:14:02 <phantomcircuit> thus your rpc call will block until it's done
269 2013-12-02 04:14:13 <phantomcircuit> but it *should* only block for smallish periods of time
270 2013-12-02 04:14:23 <phantomcircuit> but i assume you're using some terrible vps
271 2013-12-02 04:14:29 <gdoteof> anyone have a good idea how electrum works that can give me some insight? there is/was a wordpress plugin that took only the 'master publickey' as its input
272 2013-12-02 04:14:34 <Lifeofcray> home test server
273 2013-12-02 04:15:00 <phantomcircuit> gdoteof, #electrum
274 2013-12-02 04:15:45 <gdoteof> then there is an issue where people are saying "dont upgrade to 1.9.4" since it will break the store. how can it possibly break the store
275 2013-12-02 04:15:51 <phantomcircuit> sipa, i wonder if im somehow using more filedescriptors or something
276 2013-12-02 04:16:02 <gdoteof> phantomcircuit: i know. i did, and i should be patient. but its dead
277 2013-12-02 04:16:30 <Lifeofcray> how about some more general questions
278 2013-12-02 04:16:37 <phantomcircuit> why would you think asking her would help you?
279 2013-12-02 04:16:47 <phantomcircuit> this channel is almost entirely about the bitcoin-qt reference client
280 2013-12-02 04:16:57 <Lifeofcray> from what I understand, the bitcoin client got some built in account management
281 2013-12-02 04:17:06 <phantomcircuit> Lifeofcray, do not use that
282 2013-12-02 04:17:12 <phantomcircuit> it's impossible to have a proper backup
283 2013-12-02 04:17:24 <Lifeofcray> well that's good to know
284 2013-12-02 04:17:29 <Lifeofcray> what should I do instead?
285 2013-12-02 04:17:30 <phantomcircuit> you'd have to do a complete backup after almost any rpc call
286 2013-12-02 04:17:46 <phantomcircuit> Lifeofcray, use an external database to maintain account/invoice information
287 2013-12-02 04:18:21 <Lifeofcray> so basically just store peoples information and how much BTC they have in mysql?
288 2013-12-02 04:18:24 <alex_fun> :)
289 2013-12-02 04:18:30 <phantomcircuit> Lifeofcray, yes
290 2013-12-02 04:18:41 <Lifeofcray> but how does that work with payments?
291 2013-12-02 04:18:59 <phantomcircuit> Lifeofcray, what are you trying to accomplish
292 2013-12-02 04:19:36 <Lifeofcray> a simple buy/sell site
293 2013-12-02 04:19:56 <Lifeofcray> user 1 puts up an ad, user 2 pays user 1
294 2013-12-02 04:19:56 <phantomcircuit> that doesn't mean anything to me
295 2013-12-02 04:20:07 <Lifeofcray> i take a small fee
296 2013-12-02 04:20:28 <andytoshi> Lifeofcray: the moral is, track balances yourself, just use bitcoind for address generation and network interaction
297 2013-12-02 04:21:02 <Lifeofcray> is it possible to give every user their own adress then?
298 2013-12-02 04:21:11 <Lifeofcray> and just have the site check balance on that one
299 2013-12-02 04:21:26 <Luke-Jr> Lifeofcray: um, with bitcoin every *transaction* needs its own address
300 2013-12-02 04:21:35 <Luke-Jr> addresses don't have balances
301 2013-12-02 04:22:21 <phantomcircuit> Lifeofcray, you should assign an address to each transactions they do, if you maintain a site "account" balance then change the displayed address when a transaction pays to that address
302 2013-12-02 04:23:25 <warren> cfields: i have fedora
303 2013-12-02 04:23:37 <phantomcircuit> Lifeofcray, https://gitorious.org/vibanko/vibanko/source/master:
304 2013-12-02 04:23:41 <cfields> warren: you have qt4/qt5 installed?
305 2013-12-02 04:23:45 <warren> cfields: I can
306 2013-12-02 04:23:49 <phantomcircuit> that's an example of a very simple correctly implemented web wallet
307 2013-12-02 04:24:02 <warren> cfields: installing
308 2013-12-02 04:24:07 <cfields> warren: thanks
309 2013-12-02 04:25:33 <warren> did toffoo upload his chainstate?
310 2013-12-02 04:25:40 <alex_fun> Lifeofcray: do you want to make exchange? :D
311 2013-12-02 04:25:47 <Lifeofcray> naa not exchange
312 2013-12-02 04:25:59 <Lifeofcray> mostly just playing around
313 2013-12-02 04:26:27 <alex_fun> well if you dont want to say how your expect some more advice? :)
314 2013-12-02 04:26:31 <alex_fun> *you
315 2013-12-02 04:26:45 <phantomcircuit> 2013-12-02 04:12:24 ERROR: CBlockUndo::WriteToDisk() : ftell failed
316 2013-12-02 04:26:46 <phantomcircuit> wat
317 2013-12-02 04:26:53 <Lifeofcray> well first thing i want to create is just a simple web wallet
318 2013-12-02 04:26:57 <Lifeofcray> something like blockchain
319 2013-12-02 04:27:32 <phantomcircuit> how is it even possible for that to fail
320 2013-12-02 04:27:46 <alex_fun> I think maybe Kim Dot Com can add we wallets function to his storage site :)
321 2013-12-02 04:28:04 <alex_fun> by the way ramnode have sale and asking users if they want to use btc :D
322 2013-12-02 04:28:11 <alex_fun> to pay for nodes
323 2013-12-02 04:29:13 <Lifeofcray> could you guys eli5 me real quick on how bitcoin wallets work?
324 2013-12-02 04:29:17 <andytoshi> phantomcircuit: the write stream must've closed somehow?
325 2013-12-02 04:29:46 <phantomcircuit> wait how does that ever work
326 2013-12-02 04:29:58 <phantomcircuit> CAutoFile fileout = CAutoFile(OpenUndoFile(pos), SER_DISK, CLIENT_VERSION);long fileOutPos = ftell(fileout);
327 2013-12-02 04:30:02 <phantomcircuit> fileout isn't a FILE*
328 2013-12-02 04:30:13 <andytoshi> c++ magic no doubt
329 2013-12-02 04:30:46 <andytoshi> there should be a feature request to port bitcoind to rust..
330 2013-12-02 04:31:01 <alex_fun> Lifeofcray: imo webwallet is most secure when stores notional balance only
331 2013-12-02 04:31:44 <alex_fun> then they telephone you to make transfer :D
332 2013-12-02 04:31:46 <andytoshi> phantomcircuit: in serialize.h CAutoFile is defined, it overloads * and ->
333 2013-12-02 04:31:48 <alex_fun> from cold wallet
334 2013-12-02 04:31:52 <andytoshi> to return a FILE*...
335 2013-12-02 04:32:06 <phantomcircuit> oh god it does...
336 2013-12-02 04:32:22 <andytoshi> ha! that is satoshi code for you
337 2013-12-02 04:32:31 <andytoshi> git blame even blames him..
338 2013-12-02 04:32:45 <alex_fun> his code is simple so people can make alts easily
339 2013-12-02 04:32:51 <alex_fun> works just fine
340 2013-12-02 04:33:23 <Lifeofcray> so an adress just tracks how much gets sent to it?
341 2013-12-02 04:33:48 <andytoshi> Lifeofcray: an address doesn't track anything, it is used interally to construct transactions
342 2013-12-02 04:34:18 <alex_fun> bitcoin can benefit from gamification
343 2013-12-02 04:34:29 <alex_fun> well crypto in general :)
344 2013-12-02 04:34:29 <andytoshi> oh man wtf,
345 2013-12-02 04:34:31 <andytoshi> FILE** operator&() { return &file; }
346 2013-12-02 04:34:35 <Lifeofcray> this hurts my head
347 2013-12-02 04:34:41 <phantomcircuit> andytoshi, lol
348 2013-12-02 04:34:43 <alex_fun> lolol
349 2013-12-02 04:35:32 <alex_fun> Lifeofcray: coins ownership is registered in the 2p2 register
350 2013-12-02 04:35:40 <alex_fun> when u want to sent coins you use key
351 2013-12-02 04:37:02 <phantomcircuit> Bad file descriptor
352 2013-12-02 04:37:05 <phantomcircuit> wtf
353 2013-12-02 04:37:19 <Lifeofcray> so i got an adress, someone sends me money, networks keeps track on how much money that adress is worth?
354 2013-12-02 04:37:31 <gdoteof> okay i don't want to spam this room; but #electrum is dead; if someone that knows has an understanding of how it works can give a me a few minutes in PM i would appreciateit
355 2013-12-02 04:37:57 <Lifeofcray> and the bitcoin client keeps track of how much has been sent to and from that adress with the help of the network?
356 2013-12-02 04:39:51 <phantomcircuit> ah i found it
357 2013-12-02 04:39:52 <Lifeofcray> Cant I just asign every user their own adress and have my website check the transactions to see how much money that adress is worth?
358 2013-12-02 04:39:58 <phantomcircuit> apparently leveldb can call close() multiple times
359 2013-12-02 04:40:05 <phantomcircuit> tricky tricky
360 2013-12-02 04:40:32 <freewil> Lifeofcray, yes that is the basic idea
361 2013-12-02 04:40:54 <freewil> although depending on what you are doing you may want to use a new address per transaction (purchase) rather than just per user
362 2013-12-02 04:41:11 <phantomcircuit> Lifeofcray, you can track how much has been sent to an address, but not how much has been sent from it
363 2013-12-02 04:41:48 <Lifeofcray> i see
364 2013-12-02 04:41:48 <phantomcircuit> Lifeofcray, a wallet is a collection of spendable outputs, when you call sendtoaddress() it will pick the outputs that make the best transaction
365 2013-12-02 04:41:59 <phantomcircuit> so you cannot calculate an address balance that means anything
366 2013-12-02 04:42:14 <phantomcircuit> you can calculate the total bitcoins sent to an address though
367 2013-12-02 04:42:26 <phantomcircuit> Lifeofcray, does that help?
368 2013-12-02 04:42:50 <Lifeofcray> so if i make a site, keep track on how much they send, and how much is transfered to their adress, i can just do ammount_in - ammount_out?
369 2013-12-02 04:43:14 <phantomcircuit> Lifeofcray, take a look at this
370 2013-12-02 04:43:15 <phantomcircuit> https://gitorious.org/vibanko/vibanko/source/master:
371 2013-12-02 04:43:25 <phantomcircuit> it's a webwallet but i suspect it's similar to what you want to do
372 2013-12-02 04:43:27 <Lifeofcray> i will, just trying to get the basic gist of it
373 2013-12-02 04:43:37 <phantomcircuit> Lifeofcray, it's a *very* basic example
374 2013-12-02 04:44:11 <kjj> fuck. I just had this exact same conversation with someone last night, and a couple more times earlier in the week. we need a FAQ
375 2013-12-02 04:44:35 <alex_fun> i can help with FAQ
376 2013-12-02 04:44:45 <alex_fun> I can collate it a bit
377 2013-12-02 04:44:48 <freewil> theres probably one on the wiki
378 2013-12-02 04:44:57 <alex_fun> yeppa
379 2013-12-02 04:44:57 <Lifeofcray> if(bccomp($balance,$_POST['amount'],8) < 0)
380 2013-12-02 04:45:03 <alex_fun> its a bit random yet good
381 2013-12-02 04:45:04 <Lifeofcray> seems as if that's how they do it
382 2013-12-02 04:45:46 <alex_fun> :)
383 2013-12-02 04:45:46 <alex_fun> (6:41:55 AM) phantomcircuit: so you cannot calculate an address balance that means SR coins that were confiscated could of been spent already?
384 2013-12-02 04:46:49 <Lifeofcray> well, doesnt seem all that hard now
385 2013-12-02 04:47:23 <Lifeofcray> thanks guys, i feel a bit smarter
386 2013-12-02 04:48:04 <Lifeofcray> i take it litecoin and other coins work in the same manner?
387 2013-12-02 04:48:27 <Lifeofcray> say if i code somethign that works for bitcoin, it wouldnt be hard to copy that code to work for the otehrs?
388 2013-12-02 04:48:44 <freewil> litecoin's jsonrpc is exactly the same
389 2013-12-02 04:48:48 <freewil> cant comment on the other ones
390 2013-12-02 04:49:15 <kjj> you should get nearly 100% coverage of the scamcoin market
391 2013-12-02 04:49:48 <alex_fun> hehehe
392 2013-12-02 04:49:50 <kjj> since those are all just bitcoin with a different genesis hash and if they are really clever, a different port
393 2013-12-02 04:50:10 <alex_fun> kjj they do use diff port
394 2013-12-02 04:50:16 <alex_fun> and diff magic bits by some :D
395 2013-12-02 04:50:36 <alex_fun> http://devtome.com/doku.php?id=scrypt_altcoin_cloning_guide
396 2013-12-02 04:50:48 <alex_fun> misses some details yet overall fine
397 2013-12-02 04:52:25 <Lifeofcray> haha
398 2013-12-02 04:52:45 <Lifeofcray> well, it would be nice to see bitcoin stop jumping around in value
399 2013-12-02 04:52:51 <alex_fun> Lifeofcray: so u plan to make coin now? let me know when its ready to mine :P
400 2013-12-02 04:53:13 <Lifeofcray> ready to mine?
401 2013-12-02 04:53:41 <alex_fun> crypto can stop jumping in value when you premine 100% and back by gold or silver
402 2013-12-02 04:53:45 <alex_fun> or something like that
403 2013-12-02 04:53:48 <alex_fun> or 1:1 to usd
404 2013-12-02 04:54:02 <alex_fun> in essense same as yandex money but p2p
405 2013-12-02 04:54:53 <Lifeofcray> i just know that's it's going to be fun to see how an deflatory currency will work in our world
406 2013-12-02 04:55:42 <Lifeofcray> but at it is now, it's growing to much in value to be used as a "real" currency
407 2013-12-02 04:55:56 <alex_fun> is it deflatory? hehe
408 2013-12-02 04:56:09 <alex_fun> depends on people point of view
409 2013-12-02 04:56:42 <Lifeofcray> there's a fixed amount of bitcoins, coins get lost every day
410 2013-12-02 04:56:45 <Lifeofcray> so yeah
411 2013-12-02 04:56:51 <gdoteof> okay, i take back my requests for help w/r/t electrum
412 2013-12-02 04:56:57 <gdoteof> i figured out what was going on.
413 2013-12-02 04:57:47 <Lifeofcray> it's kind of hard to justify using bitcoins on buying groceries if the value goes up by 20% the next day
414 2013-12-02 04:58:31 <kjj> I hear that often. but I still bought a bunch of computer parts from bitcoinstore.com the other day (around $700 rate at the time)
415 2013-12-02 04:59:52 <Lifeofcray> well, the average consumer
416 2013-12-02 05:00:34 <Lifeofcray> and it kind of sucks if you get paid 1000 usd in a monthly salary in btc
417 2013-12-02 05:00:38 <kjj> I'm not sure what property I have that makes me different from "the average consumer"
418 2013-12-02 05:00:40 <Lifeofcray> and it crashes the next day
419 2013-12-02 05:00:59 <Lifeofcray> and you have to buy stuff tomorrow
420 2013-12-02 05:01:05 <freewil> kjj, you're on irc
421 2013-12-02 05:01:13 <gdoteof> Lifeofcray: i bought a bunch of food on foodler and then bitcoin crashed 20% the next day
422 2013-12-02 05:01:14 <Lifeofcray> and the market doesnt go up again for another two weeks
423 2013-12-02 05:01:16 <gdoteof> so your argument is invalid
424 2013-12-02 05:01:36 <Lifeofcray> gdoteof it's not really invalid, people want security
425 2013-12-02 05:01:52 <kjj> it's the same old circular argument. bitcoin can't get big because it isn't big. which is nonsense
426 2013-12-02 05:02:06 <Lifeofcray> that's another argument completly
427 2013-12-02 05:02:12 <gdoteof> Lifeofcray: people clearly want bitcoin though
428 2013-12-02 05:02:14 <Lifeofcray> if bitcoins gets bigger
429 2013-12-02 05:02:18 <Lifeofcray> it'll be more stable
430 2013-12-02 05:02:35 <gdoteof> this whole "people dont want it bexause of xyz" is pure conjecture that has historically been false
431 2013-12-02 05:02:54 <Lifeofcray> but if i work, and I get paid 1000 usd a month, and rent it 600
432 2013-12-02 05:03:02 <Lifeofcray> i want a stable currency
433 2013-12-02 05:03:08 <gdoteof> then you would be an idiot to keep allof your moneyh in bitcoin
434 2013-12-02 05:03:26 <Lifeofcray> and that's why it's more of an investment than a "real" currency right now
435 2013-12-02 05:03:31 <Lifeofcray> which stability would fix
436 2013-12-02 05:03:38 <kjj> this is way off topic, by the way
437 2013-12-02 05:03:48 <Lifeofcray> because you would indeed be an idiot to have all your cash in bitcoins
438 2013-12-02 05:03:54 <Happy_Meal> Hi, I asked this in bitcoin and haven't gotten an answer so I was wondering if someone here could point me to a resource explaining what exactly bitcoin is. I keep reading that new coins are created once an sha-256 hash has been found that meets certain characteristics. I get this but what doesn't make sense is, what exactly is the thing that is created?
439 2013-12-02 05:04:18 <kjj> Happy_Meal: what is created is an entry in the ledger
440 2013-12-02 05:04:37 <kjj> and bitcoin is, essentially, a global distributed ledger
441 2013-12-02 05:04:57 <gdoteof> Happy_Meal: check out http://bitcoin.org/bitcoin.pdf
442 2013-12-02 05:05:05 <phantomcircuit> warren, https://github.com/pstratem/bitcoin/tree/leveldbnommap
443 2013-12-02 05:05:19 <gdoteof> Happy_Meal: the thing that gets created is an entry in the ledger
444 2013-12-02 05:05:22 <phantomcircuit> i bet that fixes effectively all the issues on os x
445 2013-12-02 05:05:36 <phantomcircuit> and bitcoin already internally buffers everything so the performance hit is trivial
446 2013-12-02 05:05:38 <warren> phantomcircuit: rescrv claims this approach won't help other things, but I'll try it.
447 2013-12-02 05:05:56 <phantomcircuit> warren, im sure there are other issues with leveldb
448 2013-12-02 05:06:08 <phantomcircuit> specifically it doesn't handle a bunch of fairly common failure modes for hdds
449 2013-12-02 05:06:09 <gdoteof> imagine you were playing a game with your friends called "get to punch someone" and you played the game by watching southpark and everytime cartman said bitch you rolled a dice
450 2013-12-02 05:06:19 <warren> phantomcircuit: how marginal is marginal?
451 2013-12-02 05:06:22 <phantomcircuit> but the fix for that would not be backwards compatible
452 2013-12-02 05:06:28 <gdoteof> and the number the dice corresponded to decided who got a token
453 2013-12-02 05:06:40 <phantomcircuit> warren, 1-100k blocks i measured a 1 second increase in processing time
454 2013-12-02 05:06:45 <warren> OMG
455 2013-12-02 05:06:49 <gdoteof> then with that token, they could spend it by giving it to osmeone; which gives them the right within the rules of the game
456 2013-12-02 05:06:52 <gdoteof> to punch them
457 2013-12-02 05:06:58 <gdoteof> bitcoin are like those tokens
458 2013-12-02 05:07:11 <gdoteof> there can be infinite of them that have nothing to do with the game
459 2013-12-02 05:07:13 <kjj> except there are no punches
460 2013-12-02 05:07:17 <gdoteof> but to play the game fairly, which everyone enforces
461 2013-12-02 05:07:20 <phantomcircuit> warren, i'll run them again to 250k
462 2013-12-02 05:07:31 <gdoteof> only the tokens that came into the game through the legit rules get accepted
463 2013-12-02 05:07:43 <warren> phantomcircuit: do you think this adds any risk to linux/win32 that were otherwise fine?
464 2013-12-02 05:07:55 <phantomcircuit> warren, no
465 2013-12-02 05:07:56 <warren> phantomcircuit: we're on the edge of making a release ...
466 2013-12-02 05:07:58 <alex_fun> ACTION run kji average customer test, 50% 60% 90% IRC presense detected, failed :D
467 2013-12-02 05:08:19 <phantomcircuit> warren, look at the patch, it's actually very simple
468 2013-12-02 05:08:49 <rescrv> phantomcircuit, warren: you may want to error out on short writes too
469 2013-12-02 05:08:54 <Happy_Meal> The entry in the ledger is a list of all transactions between the current block and the new block created right? With each successful win, comes (currently) 25 bitcoins. That's what I was asking. What determines how many bitcoins are newly created. Also, thanks for the link to the pdf.
470 2013-12-02 05:08:59 <phantomcircuit> rescrv, this does
471 2013-12-02 05:09:01 <phantomcircuit> i think
472 2013-12-02 05:09:08 <phantomcircuit> let me double check
473 2013-12-02 05:09:25 <phantomcircuit> oh it doesn't
474 2013-12-02 05:09:27 <phantomcircuit> let me fix that
475 2013-12-02 05:09:52 <gdoteof> Happy_Meal: no the ledger is a list of all transactions from the first block ever (aka genesis block) to the currently created block
476 2013-12-02 05:10:19 <warren> phantomcircuit: rescrv: I'm burning myself trying to get 10 things done before moving in 6 days. I'm relying on you folks to figure out how to fix this. I can do builds.
477 2013-12-02 05:10:20 <gdoteof> you can think of it essentially just as a list of addresses that have balances
478 2013-12-02 05:10:42 <gdoteof> Happy_Meal: it is just part of the protocol. it is 25 now
479 2013-12-02 05:10:43 <kjj> Happy_Meal: there is a subsidy function. the finder of a block can create no more than 25 BTC (currently)
480 2013-12-02 05:10:53 <gdoteof> it used to be 50, in a couple years it will be 12.5 then 6.25 etc..
481 2013-12-02 05:11:02 <warren> rescrv: saw toffoo's chainstate and debug.log?
482 2013-12-02 05:11:17 <phantomcircuit> rescrv, if only write set errno on a short write this would be so easy
483 2013-12-02 05:11:20 <kjj> gdeteof: except that there are no balances, and not everything in the ledger is or has an address
484 2013-12-02 05:11:28 <Happy_Meal> So essentially the points are purely conceptual and not enforced by math but by group agreement?
485 2013-12-02 05:11:42 <gdoteof> Happy_Meal: we should take this to #bitcoin
486 2013-12-02 05:11:56 <Happy_Meal> sorry and OK
487 2013-12-02 05:15:46 <rescrv> warren: I saw toffoo's post. Here's my current take: We're working with many customers with LevelDB in production. So is Basho (the Riak folks). We've not encountered such a corruption error in LevelDB other than the one that would be explained by my patch. Toffoo's current error looks like a bitflip or a memory corruption bug, and the crashes he was getting would indicate the same. I would
488 2013-12-02 05:15:48 <rescrv> strongly suggest that toffoo run a memtest, and that he run a build linking against ElectricFence for the next round of debugging.
489 2013-12-02 05:16:47 <warren> rescrv: I know very little about mac dev and I'm out of time. someone else would need to supply him builds. I suggest gavinandresen, as my build environment is modeled after his.
490 2013-12-02 05:17:35 <warren> rescrv: while toffoo is among the rare users who experience any corruption, the behavior he sees has been nearly identical to coblee
491 2013-12-02 05:17:39 <phantomcircuit> rescrv, the way bitcoin writes to leveldb (sequentially in large chunks from a single thread) the difference between mmap and pread/write is almost nonexistent
492 2013-12-02 05:19:34 <phantomcircuit> rescrv, do you know if leveldb has version numbers for their databases?
493 2013-12-02 05:19:49 <phantomcircuit> i would like to add sequence numbers to the journal but that is not backwards compatible
494 2013-12-02 05:19:54 <rescrv> phantomcircuit: version numbers signalling?
495 2013-12-02 05:20:18 <gmaxwell> rescrv: we've had many reports of corruption, though indeed, it may well be that any particular tester has flaky hardware. Though I would be somewhat surprised if bitflips in _ram_ only manifest this way: bitcoin probably does 10000x the amount of sha256 than it does leveldb operations, and all of it is checked and produces distinctive screams that toffoo hasn't been seeing.
496 2013-12-02 05:20:27 <phantomcircuit> rescrv, like can i increment a version number somewhere to indicate this leveldb cannot read any previous versions database
497 2013-12-02 05:20:34 <gmaxwell> rescrv: bitflips in sata or the drive would be more likely than in ram, if bitflips are indeed part of his issue.
498 2013-12-02 05:21:01 <toffoo> also, bitcoin0.7.2 (and all previous versions) work perfectly, as does all other software
499 2013-12-02 05:21:08 <rescrv> phantomcircuit: no version number. You'd probably do best to modify the CURRENT file
500 2013-12-02 05:22:27 <warren> FYI: We're considering partially sponsoring cfields to go to the Vegas Bitcoin conference. I hope others see value in donating to him so he can attend.
501 2013-12-02 05:23:07 <warren> Or maybe it would be more valuable to have a Bitcoin dev meeting elsewhere.
502 2013-12-02 05:24:00 <phantomcircuit> rescrv, alright modified to fail on short writes
503 2013-12-02 05:24:33 <rescrv> gmaxwell: I'm leaning more heavily on memory corruption, meaning something writing where it shouldn't. The crash he posted is most likely the result of reading from an invalid pointer. The corruption he just posted is the result of something overwriting state used during compaction, causing checksums to fail. The previous corruption bugs, reported by toffoo and reproduced by gavinandresen were
504 2013-12-02 05:24:35 <rescrv> different in that data was simply not making it to disk.
505 2013-12-02 05:26:03 <rescrv> gmaxwell: further, toffoo seems to have no issue until he restarts overnight per https://github.com/bitcoin/bitcoin/issues/2770#issuecomment-29594370; toffoo: do you leave the process running overnight?
506 2013-12-02 05:26:11 <toffoo> no
507 2013-12-02 05:26:21 <toffoo> i turn off the laptop
508 2013-12-02 05:26:33 <toffoo> today, it was not upon first restart,
509 2013-12-02 05:26:41 <toffoo> it open and ran well for a few hours,
510 2013-12-02 05:26:42 <toffoo> then I closed it,
511 2013-12-02 05:26:54 <toffoo> and upon the next start got the error
512 2013-12-02 05:27:21 <warren> toffoo: coblee has been having nearly the same test results as toffoo, coblee is a veteran engineer and would know if he has any hardware issues.
513 2013-12-02 05:28:20 <toffoo> i'm a former sunOS/solaris/winNT sysadmin
514 2013-12-02 05:28:37 <toffoo> (but never a programmer) :(
515 2013-12-02 05:28:42 <alex_fun> :)
516 2013-12-02 05:29:00 <gmaxwell> rescrv: I'm still skeptical, we hit users with unreliable memory with some regularity. Bitcoin is _very_ sensitive because all the data and processing it does is authenticated. He was reliable on many pre-level-db verions of Bitcoin, and totally unreliable on many post level-db versions, so it seems unlikely that the earlier versions just got lucky memory layouts.
517 2013-12-02 05:29:16 <toffoo> gmaxwell agreed
518 2013-12-02 05:29:22 <gmaxwell> Though memtesting is _always_ prudent, so I agree toffoo should run something like memtest86.
519 2013-12-02 05:29:31 <gmaxwell> (does that work on macs?)
520 2013-12-02 05:29:39 <phantomcircuit> this is probably premature but https://github.com/bitcoin/bitcoin/pull/3340
521 2013-12-02 05:29:44 <toffoo> someone posted a mac memtest in the forum thread,
522 2013-12-02 05:29:50 <toffoo> anyone know if it's good?
523 2013-12-02 05:30:08 <toffoo> I discounted the idea when I first read it, but I'll try it now if you think it's a good idea
524 2013-12-02 05:31:05 <toffoo> duh .. that's right, the forum is AWOL
525 2013-12-02 05:31:06 <gmaxwell> even if memory isn't your only problem, if you also have memory corruption that will complicated diagnosis, and testing is pretty easy.
526 2013-12-02 05:31:22 <gmaxwell> toffoo: it's just dns thats stuck flush your dns caching or peg the ip.
527 2013-12-02 05:31:42 <gmaxwell> toffoo: 109.201.133.195
528 2013-12-02 05:31:51 <toffoo> agh thanks
529 2013-12-02 05:37:40 <phantomcircuit> gmaxwell, that is a pretty hilarious attack vector
530 2013-12-02 05:37:50 <phantomcircuit> afaict none of the dns registrars are secure at all
531 2013-12-02 05:38:06 <toffoo> Memtest version 4.22 in progress...
532 2013-12-02 05:40:07 <rescrv> gmaxwell toffoo warren: let's assume no hardware issues: The file that's corrupt in toffoo's latest chainstate is 010255.sst. That file was generated by compaction, merging other files to create it. Those other files were not corrupt; otherwise, the compaction would have failed and not written the file. It follows then, that it corrupted during that rewrite process. Is there any easy way to discern
533 2013-12-02 05:40:09 <rescrv> what the value stored in the chainstate should be? As in, if I copy someone's valid chain state, should the two be bytewise equal for present keys?
534 2013-12-02 05:41:44 <gmaxwell> rescrv: if you also know what blocknumber his node was at, you can reindex another node and make it stop at that height and you will end up with an identicate chainstate state. (the gettxoutsetinfo rpc will even iterate through the chainstate and return the sha256, and it will agree between nodes)
535 2013-12-02 05:43:07 <rescrv> gmaxwell: is each key in the chainstate an immutable value?
536 2013-12-02 05:43:30 <toffoo> blocknumber should be in the debug.log I pastebined
537 2013-12-02 05:45:30 <shesek> I'm getting a lot (in the 1000's) of ORPHAN BLOCK messages after not opening bitcoind for awhile. is that normal?
538 2013-12-02 05:46:03 <shesek> I'm guessing that it isn't :-\
539 2013-12-02 05:46:10 <shesek> what could be causing it?
540 2013-12-02 05:49:53 <gmaxwell> rescrv: yes, if the key is in there its value will be the same for anyone other system that has the same key. (keys are inserted in removed, but never changed, IIRC)
541 2013-12-02 05:50:02 <gmaxwell> shesek: it's fineish.
542 2013-12-02 05:50:26 <gmaxwell> shesek: it just means that you got a new block announcement before you'd caught up the forward direction, and now its enumerating backwards to find where it connects.
543 2013-12-02 05:51:12 <gmaxwell> (fine-ish because it sucks but it's not specific to you, its expected behavior, and beyond causing a spike in memory usage and some delays it should be harmless)
544 2013-12-02 05:51:47 <warren> I'm pushing new builds, alternatively rescrv and phantomcircuit's patch. I'm included memory barrier just for the heck of it.
545 2013-12-02 05:54:57 <shesek> gmaxwell, ah, I see. thanks.
546 2013-12-02 06:11:41 <shesek> it went all the way back to 2009
547 2013-12-02 06:12:05 <shesek> and I'm still getting about an even mixture of accepted and orphan messages
548 2013-12-02 06:12:58 <shesek> oh, wait, no, not even -- mostly accepted, but still lots of orphaned
549 2013-12-02 06:22:00 <warren> http://download1.rpmfusion.org/~warren/bitcoin-0.8.5-OMG7/macosx/
550 2013-12-02 06:22:14 <warren> Both options of rescrv and phantomcircuit's patch.
551 2013-12-02 06:25:29 <madthanu> rescrv: stupid suggestion probably, but does compaction verify checksums even if you don't run in paranoid mode?
552 2013-12-02 06:26:02 <madthanu> rescrv: never mind, i get it now
553 2013-12-02 06:29:31 <toffoo> I passed my memtest: http://pastebin.com/Xu52eJr7
554 2013-12-02 06:30:25 <toffoo> warren should I do a -reindex with OMG7 or can I copy over a pre-corruption copy of chainstate/ ?
555 2013-12-02 06:31:40 <toffoo> oh and which OMG7 should I try?
556 2013-12-02 06:39:04 <Burritoh> Someone in #bitcoin is claiming that the client has the ability to receive a signed message from developers, to be displayed to the user, warning them not to transact while the Bitcoin network is under attack.
557 2013-12-02 06:39:11 <Burritoh> Are they right? Is there such an alarm?
558 2013-12-02 06:39:56 <gmaxwell> Burritoh: https://en.bitcoin.it/wiki/Alerts
559 2013-12-02 06:40:31 <Burritoh> Aaah..
560 2013-12-02 06:42:59 <Krellan> If Satoshi ever came back, he could use the alerts system to prove himself, right? Assuming he still had his keys.
561 2013-12-02 06:43:17 <vsrinivas> currently the 'setgenerate' rpc in bitcoind's source is marked as thread safe. is that really appropriate?;
562 2013-12-02 06:44:10 <gmaxwell> Krellan: no, more than satoshi has the alert key.
563 2013-12-02 06:44:43 <gmaxwell> Krellan: he could potentially sign with the key in the genesis block, but he might not have it. regardless he has a somewhat well known pgp key.
564 2013-12-02 06:44:54 <vsrinivas> concurrent callers of GenerateBitcoins() might end up trying to delete minerthreads twice, i think.
565 2013-12-02 06:45:09 <Krellan> gmaxwell: Didn't know it was a shared key. OK thanks that makes sense, he could always sign a message to verify an address from one of his early blocks.
566 2013-12-02 06:45:20 <sevenqueue> Warning: Displayed transactions may not be correct! You may need to upgrade, or other nodes may need to upgrade.
567 2013-12-02 06:45:31 <gmaxwell> Krellan: the only block we know for sure is his .. is the genesis block.
568 2013-12-02 06:45:39 <Burritoh> gmaxwell: these days, wouldn't the alerts need to start being multilingual?
569 2013-12-02 06:45:52 <Auctus> so a bitcoin transaction has only a to address, and then a transaction id for you to look up what the from address was?
570 2013-12-02 06:45:59 <gmaxwell> Burritoh: we've put URLs in them in the past, and the content of the url was translated.
571 2013-12-02 06:46:05 <Burritoh> I really didn't expect something like an alert system in Bitcoin-Qt :|
572 2013-12-02 06:46:06 <Burritoh> ah
573 2013-12-02 06:46:07 <Auctus> like a previous transaction id i mean
574 2013-12-02 06:46:17 <Burritoh> not that it's a particularly bad thing
575 2013-12-02 06:46:24 <gmaxwell> Auctus: there is no real from address at allâ thats not a "from" it's a prior to, which is subtly different.
576 2013-12-02 06:46:57 <gmaxwell> Burritoh: well it's kinda ugly but too valuable not to have, the potential for harm is minimized as much as we know how
577 2013-12-02 06:47:06 <Auctus> can you explain the difference/where can i read about it? I mean i understand the difference, but in the end it amounts to the same thing?
578 2013-12-02 06:47:54 <Krellan> gmaxwell: Interesting, I thought Satoshi exclusively mined most of the early blocks, not just the genesis block. No real way to conclusively prove who owns what early coin, unless they revealed themselves voluntarily.
579 2013-12-02 06:47:57 <gmaxwell> vsrinivas: yea that doesn't sound quite right to me, how'd you find that.
580 2013-12-02 06:48:07 <vsrinivas> gmaxwell: source inspection;
581 2013-12-02 06:48:46 <gmaxwell> Krellan: well, it's a reasonable guess, but we don't really know for sure, there was a small gap between the initial timestamps and the post (e.g. he waited to see that it was working) but he could have also announced it to other people privately in that time.
582 2013-12-02 06:49:07 <gmaxwell> so as far as I know there is no way to be sure that block 1 is his, though its likely.
583 2013-12-02 06:49:34 <gmaxwell> there is that nonce correlation stuff, but that may be distinguishing someone else instead of him.
584 2013-12-02 06:52:43 <Krellan> gmaxwell: Makes sense. The mined coins were not spent until Block 8. Blocks 1-7, each to a unique address, have never been spent (as we know, Block 1 *can't* ever be spent).
585 2013-12-02 06:53:38 <gmaxwell> block 0 can't be spent.
586 2013-12-02 07:01:21 <vsrinivas> while we're observing things ... DetectShutdownThread in the no-ui code wakes up every 200 mS, just to check a variable; its a minor change, but perhaps StartShutdown() / ShutdownRequested() should rendezvous by something other than a 200 mS sleep and a volatile variable.
587 2013-12-02 07:04:12 <gmaxwell> phantomcircuit: IIRC at least on some OSX versions fdatasync() just doesn't exist at all. The normal fallback code on hosts without fdatasync() is fsync() but that doesn't work as expected on OSX.
588 2013-12-02 07:05:03 <phantomcircuit> gmaxwell, all i know is that it's #define'd as the fnctl
589 2013-12-02 07:05:08 <vsrinivas> gmaxwell: the OS X fallback for fdatasync is fcntl F_FULLSYNC usually.
590 2013-12-02 07:05:34 <gmaxwell> vsrinivas: I mean in most software written preOSX stupidity with fsync.
591 2013-12-02 07:05:45 <gmaxwell> e.g. what you'd do on AIX or something.
592 2013-12-02 07:06:18 <gmaxwell> vsrinivas: we use fcntl F_FULLSYNC, of course, but wumpus and phantomcircuit are in the process of trying to convince themselves that it's not needed.
593 2013-12-02 07:06:41 <vsrinivas> ah ok.
594 2013-12-02 07:06:54 <phantomcircuit> gmaxwell, hmm?
595 2013-12-02 07:07:02 <phantomcircuit> gmaxwell, no im not
596 2013-12-02 07:09:50 <phantomcircuit> gmaxwell, lol im pretty sure im the one who told you osx is a liar about fsync
597 2013-12-02 07:10:38 <gmaxwell> phantomcircuit: you might have told me, but I knew from elsewhere too. (it was some drama on the postgresql list a long time ago IIRC, plus people claiming osx rigs benchmarks etc)
598 2013-12-02 07:12:02 <phantomcircuit> gmaxwell, lol
599 2013-12-02 07:16:28 <vsrinivas> should 'setgenerate' really be okay to start mining threads while bitcoind is in safe mode?
600 2013-12-02 07:31:17 <warren> toffoo: pre-corruption copy should be fine
601 2013-12-02 07:48:47 <rweichler> hey whats the min amount of ram you need to run bitcoind?
602 2013-12-02 07:49:18 <rweichler> im running it on a 512MB VPS and it keeps getting killed
603 2013-12-02 07:49:28 <rweichler> im afraid that it might be too little ram or something
604 2013-12-02 07:50:35 <diki> rweichler:Sadly, it's indeed a resource hog
605 2013-12-02 07:50:41 <diki> I am afraid you will need to upgrade.
606 2013-12-02 07:50:57 <Krellan> rweichler: My bitcoind is taking 1.4GB resident, 2.9GB virtual, now.
607 2013-12-02 07:50:57 <warren> 1GB RAM would be safe for now
608 2013-12-02 07:51:02 <warren> whoa
609 2013-12-02 07:51:11 <warren> it's using like 700MB resident for me with disablewallet=1
610 2013-12-02 07:51:13 <diki> Not impossible.
611 2013-12-02 07:51:18 <lianj> maybe do less connection and more swap
612 2013-12-02 07:51:27 <Krellan> rweichler: Although I'm running 64-bit and I've had it up for a long time so it has around 70 connections or so.
613 2013-12-02 07:51:40 <gmaxwell> Krellan: what version is that? 1.4 GB resident sounds broken.
614 2013-12-02 07:51:43 <Krellan> And I have a wallet, and I'm mining.
615 2013-12-02 07:52:02 <Krellan> gmaxwell: It's from github, not THE latest, but pretty close to it.
616 2013-12-02 07:52:03 <rweichler> warren: safe... for now?
617 2013-12-02 07:52:09 <rweichler> does it take more RAM the bigger the blockchain is?
618 2013-12-02 07:52:12 <gmaxwell> Krellan: mine is 424MB resident.
619 2013-12-02 07:52:20 <gmaxwell> rweichler: no, not meaningfully.
620 2013-12-02 07:52:51 <Krellan> gmaxwell: That's nice and small. I wonder if optimization/etc. flags make a difference when compiling?
621 2013-12-02 07:52:52 <rweichler> ah, okay
622 2013-12-02 07:53:36 <Krellan> gmaxwell: What's your virtual (not resident) RAM footprint of your bitcoind?
623 2013-12-02 07:54:16 <rweichler> i guess i shall turn on swap
624 2013-12-02 07:54:20 <gmaxwell> Krellan: infinity squared or something like that, on x86_64 leveldb makes extensive use of mmap, so the virtual numbers are high and not very meaningful.
625 2013-12-02 07:54:53 <Krellan> gmaxwell: I noticed the same thing. Hard to actually know the real usage. There's all sorts of mapped chainstate files around.
626 2013-12-02 07:55:14 <warren> I'm making no-mmap builds to see if it changes anything.
627 2013-12-02 07:55:36 <gmaxwell> Krellan: not just that, thread stacks and other stuff inflate virt. really on any multthreaded program virt is useless, doubly so if it uses mmap.
628 2013-12-02 07:56:07 <gmaxwell> warren: on x86_64 those would still use mmap on the reading side, I believeâ I think phantomcircuit only changed the writer.
629 2013-12-02 07:56:23 <warren> oh
630 2013-12-02 07:58:40 <Krellan> gmaxwell: Thanks, yes there's a lot of threads. I played with "ps" more and now I think it's only 0.7 GB resident 1.4 GB virtual. Sounds better.
631 2013-12-02 08:02:36 <Krellan> It should be possible to run bitcoind on a small RAM system, if you're willing to tolerate lots of swapping, and the slowness that results.
632 2013-12-02 08:03:44 <Krellan> On the other extreme, a large RAM system, with entire blockchain on RAM disk, fast!
633 2013-12-02 08:03:45 <warren> wumpus: https://github.com/pstratem/bitcoin/commit/d4c99c02a8fb2f6b50b9fcf86e3398dcbf7cea1c#commitcomment-4743856
634 2013-12-02 08:03:54 <warren> wumpus: basically this means this patch needs more work to be safe on mac?
635 2013-12-02 08:04:15 <wumpus> @warren, no, he explained it in a latter post
636 2013-12-02 08:04:43 <wumpus> in leveldb, fdatasync *is* defined as fcntl(fileno(fileout), F_FULLFSYNC, 0);
637 2013-12-02 08:04:57 <warren> ok thanks
638 2013-12-02 08:05:11 <warren> wumpus: I'm throwing this at testers. if any survive we'll get reports.
639 2013-12-02 08:05:38 <wumpus> well it's another step in the right direction :)
640 2013-12-02 08:05:54 <wumpus> it's crazy that macosx is such a data-eating trap
641 2013-12-02 08:06:22 <warren> wumpus: https://bitcointalk.org/index.php?topic=337294.msg3793820#msg3793820
642 2013-12-02 08:15:56 <phantomcircuit> gmaxwell, no i also commented out the code that selects mmap for reading
643 2013-12-02 08:19:04 <warren> phantomcircuit: oh! then i'm very interested in seeing this on linux
644 2013-12-02 08:19:46 <warren> horray for last minute changes
645 2013-12-02 08:22:50 <wumpus> woohoo
646 2013-12-02 08:23:14 <warren> not like we're responsible for billions of dollars or anything
647 2013-12-02 08:23:18 <warren> oh wait
648 2013-12-02 08:23:31 <rweichler> nother question
649 2013-12-02 08:23:43 <rweichler> is there a way to be notified, like, immediately when a block is found
650 2013-12-02 08:23:46 <rweichler> and run some code
651 2013-12-02 08:23:49 <rweichler> based on that block?
652 2013-12-02 08:23:54 <rweichler> i cant find anything in the wiki
653 2013-12-02 08:26:04 <wumpus> rweichler: yes, the -blocknotify option can run any script
654 2013-12-02 08:26:23 <wumpus> Execute command when the best block changes (%s in cmd is replaced by block hash)
655 2013-12-02 08:27:19 <rweichler> sweet!
656 2013-12-02 08:27:21 <rweichler> thanks :D
657 2013-12-02 08:27:39 <cfields> wumpus: you run ubuntu, right?
658 2013-12-02 08:30:53 <wumpus> cfields: yep!
659 2013-12-02 08:31:04 <wumpus> I can test with 12.04 and 13.10
660 2013-12-02 08:31:15 <cfields> wumpus: mind doing 1 or 2 quick checks real quick?
661 2013-12-02 08:31:26 <cfields> not ready to test yet, i can push up tomorrow
662 2013-12-02 08:31:48 <cfields> just need to verify a few things quickly on an older version
663 2013-12-02 08:33:00 <cfields> wumpus: if you don't mind, and happen to have qt4/qt5 installed side-by-side, i'd be interested to know the results of:
664 2013-12-02 08:33:15 <cfields> ls -l `which moc`
665 2013-12-02 08:34:05 <wumpus> ok I have qt4 and qt5 installed on the 13.10, I'll take a look
666 2013-12-02 08:34:25 <gmaxwell> saracen: I hate to say it, but I think directory.io might need a "ITS A JOKE" thing.
667 2013-12-02 08:34:33 <gmaxwell> 00:28 < Olipro> looks like Bitcoin-Qt might be generating "weak" keys
668 2013-12-02 08:34:33 <gmaxwell> 00:30 < Olipro> gmaxwell: mmm, but they're all within the range & length that I see the Qt client generate
669 2013-12-02 08:35:14 <cfields> wumpus: ah perfect, i'm dev'ing on 13.04.
670 2013-12-02 08:36:05 <cfields> wumpus: main thing i'm trying to figure out is how the distros manage the qt4/qt5 split as far as locating qt build-side tools. Seems everyone does it differently of course.
671 2013-12-02 08:37:14 <cfields> I believe I have ubuntu/fedora worked out. Trying to cover as many bases as possible initially
672 2013-12-02 08:39:16 <throughnothing> I'm trying to wrap my head around bip32, and don't understand the reason/differenc for public and private child key derivation, can anyone explain this?
673 2013-12-02 08:39:25 <wumpus> qt4: lrwxrwxrwx 1 root root 35 okt 13 15:18 /usr/bin/moc-qt4 -> ../lib/x86_64-linux-gnu/qt4/bin/moc
674 2013-12-02 08:39:38 <wumpus> general: lrwxrwxrwx 1 root root 9 aug 7 16:46 /usr/bin/moc -> qtchooser
675 2013-12-02 08:40:04 <gmaxwell> throughnothing: public derivation has a non-intutive vulnerability that private does not have.
676 2013-12-02 08:40:12 <cfields> wumpus: perfect. that's 13.10?
677 2013-12-02 08:40:13 <wumpus> I don't see one for qt5 specifically, but the general one seems to choose qt5
678 2013-12-02 08:40:19 <wumpus> yes
679 2013-12-02 08:40:33 <gmaxwell> throughnothing: if you give someone one private key, and the extended public key then they can derrive all the private keys.
680 2013-12-02 08:40:36 <cfields> ok. could you verify that you can force qt4? sec.
681 2013-12-02 08:40:43 <gmaxwell> throughnothing: the private derrivation doesn't have this property.
682 2013-12-02 08:40:58 <wumpus> force qt4? well I can build the current bitcoin-qt which forces qt4 :p
683 2013-12-02 08:41:07 <throughnothing> gmaxwell: hmm ok. So does that equate toâ¦don't use public derivations?
684 2013-12-02 08:41:08 <cfields> QT_SELECT=qt4 moc -v
685 2013-12-02 08:41:18 <cfields> QT_SELECT=qt5 moc -v
686 2013-12-02 08:41:27 <wumpus> ok will test
687 2013-12-02 08:41:28 <cfields> should do the obvious things.
688 2013-12-02 08:41:40 <throughnothing> gmaxwell: and i was confused b/c public/private has a different meaning when talking about public/private keys (or so it seems), so public/private derivations is confusing when each derivation type has an associated public/private keypair
689 2013-12-02 08:41:43 <throughnothing> or am I misunderstanding that
690 2013-12-02 08:41:53 <gmaxwell> throughnothing: no, they're insanely useful. And there is no alternative to their usefulness. They allow greately improved security for some applications.
691 2013-12-02 08:41:54 <wumpus> Qt Meta Object Compiler version 63 (Qt 4.8.4)
692 2013-12-02 08:41:59 <wumpus> Qt Meta Object Compiler version 67 (Qt 5.0.2)
693 2013-12-02 08:42:05 <wumpus> respectively... so yeah it works!
694 2013-12-02 08:42:16 <cfields> wumpus: great. So they haven't screwed it up yet, at least :)
695 2013-12-02 08:42:34 <cfields> wumpus: thanks a bunch. I'll have you something to try tomorrow.
696 2013-12-02 08:42:38 <gmaxwell> throughnothing: its not a seperate meaning, the private derrivation can only be done using private key material. The public derrivation can be done using (semi-)public data.
697 2013-12-02 08:42:42 <wumpus> cfields: great!
698 2013-12-02 08:43:18 <throughnothing> gmaxwell: as i understand the bip32 spec, the first half of child nodes use public derivation, and the second half (of the 32-bit space) uses private
699 2013-12-02 08:43:25 <throughnothing> is this b/c of the mathematics, or just arbitrary choice by the spec
700 2013-12-02 08:44:48 <gmaxwell> throughnothing: it just allows the single index to signal which is being used, it's an arbitrary choice.
701 2013-12-02 08:45:40 <throughnothing> hmm ok
702 2013-12-02 08:46:36 <throughnothing> If I have my master key, and I derive m/0
703 2013-12-02 08:46:51 <throughnothing> that node will have a public and private keypair, and thus a usable address
704 2013-12-02 08:46:52 <throughnothing> right?
705 2013-12-02 08:47:04 <throughnothing> and that would be in the 'public' derivation space for the master node
706 2013-12-02 08:54:48 <throughnothing> oops
707 2013-12-02 09:10:41 <throughnothing> gmaxwell: ah, I think i see the difference, if you have a 'public derivation' keypair, you can generate sub-public keys, but not sub-private keys
708 2013-12-02 09:10:53 <throughnothing> with a private derived key, you must have the private key to even derive sub-public keys
709 2013-12-02 09:10:56 <throughnothing> ?
710 2013-12-02 09:10:56 <throughnothing> is that corerct
711 2013-12-02 09:24:15 <gmaxwell> throughnothing: thats correct.
712 2013-12-02 09:37:35 <throughnothing> gmaxwell: perfect, thanks
713 2013-12-02 10:27:22 <warren> toffoo: it's crashing in the same function, just called elsewhere =(
714 2013-12-02 10:27:26 <warren> fucking mac