1 2012-07-04 00:24:36 <gmaxwell> BlueMatt: Not?  False perhaps?
  2 2012-07-04 00:25:37 <BlueMatt> gmaxwell: sure, I found not connected to make sense, but...yea false is better
  3 2012-07-04 00:27:25 <BlueMatt> (should be "false" not false though)
  4 2012-07-04 00:28:08 <gmaxwell> The hazard there is someone writes some rpc reading code and only sees 'true' .. they might write if connected=="false" as the opposite of true.  Alternatively.  connected: 0  connected: 1  (or more, enh? if you have more than one?)
  5 2012-07-04 00:28:31 <gmaxwell> False is fine
  6 2012-07-04 01:17:54 <jgarzik> BlueMatt: stuff like "Remove duplicate GetHash() in ConnectBlock." should be closer to git HEAD
  7 2012-07-04 01:17:58 <jgarzik> BlueMatt: would love to upstream that one now
  8 2012-07-04 01:28:03 <jgarzik> mine, you miners.  mine!
  9 2012-07-04 02:14:27 <gribble> New news from bitcoinrss: jgarzik opened pull request 1554 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/1554>
 10 2012-07-04 06:05:44 <cande> about doublespending, Finney attack, how quickly after the attacker has found his block, must he initiate the attack?
 11 2012-07-04 06:45:47 <Keverw> Hi. I'm setting up a VM to install BTC to use TestNet. Can my local PC connect to the RPC on my VM if I know the password, or does RPC only take connections from Localhost?
 12 2012-07-04 06:47:55 <Keverw> new to the API, gonna play with Node.js and Bitcoins! I guess I can wait and try.  Don't see anything in the docs that would say why it won't work& so idk.
 13 2012-07-04 06:47:57 <sipa> See the -rpcallowip opion
 14 2012-07-04 06:48:02 <sipa> option
 15 2012-07-04 06:48:37 <Keverw> oh. okay.
 16 2012-07-04 06:48:57 <Keverw> so rpcallowip=10.0.1.* should work
 17 2012-07-04 06:49:04 <sipa> indeed
 18 2012-07-04 06:49:33 <Keverw> Sweet! Thanks.
 19 2012-07-04 06:51:34 <Keverw> I'm worried about fees. hehe. TestNet does fees like normal one right? Want to make sure the fee handling code is good.
 20 2012-07-04 06:59:42 <gmaxwell> Keverw: worried about fees how?
 21 2012-07-04 07:00:38 <gmaxwell> Testnet's behavior isn't entirely the same for everything (no isstandard check), but at the moment I don't recall there being a fee difference. Though its 5 am here and I should be asleep.
 22 2012-07-04 07:01:08 <Keverw> Like if I'm storing multiple peoples money
 23 2012-07-04 07:01:43 <gmaxwell> In any case even in the insane cases were your wallet is full of nothing but tiny inputs so you have to make a maximum size transaction the highest an anti-dos (mandatory) fee can be is 0.05 BTC, and getting to that requires your client to create a 100KB transaction.
 24 2012-07-04 07:01:47 <Keverw> like need to make sure they withdraw too much(like someone told me the RPC do not ask about fees like the client, it just sends them)
 25 2012-07-04 07:03:08 <gmaxwell> Keverw: right the rpc will just apply the mandatory fees (0.0005 BTC/kb) where required (when the inputs don't have enough priority to qualify as a free txn or when you have an output less than 0.01 BTC).
 26 2012-07-04 07:03:38 <Keverw> hmm
 27 2012-07-04 07:04:16 <Keverw> What if I use the accounts feature?
 28 2012-07-04 07:04:40 <Keverw> Where does the fees come from? The main one, or the account and if there is no funds in that users account, give an error?
 29 2012-07-04 07:04:43 <gmaxwell> Later you regred using the accounts feature. :)
 30 2012-07-04 07:05:39 <gmaxwell> (in particular the accounts stuff is wonky with respect to fees, and its defiantly too late for me to try to remember.  This is stuff you can test on testnet.)
 31 2012-07-04 07:05:50 <gmaxwell> (or with testnet in a box)
 32 2012-07-04 07:06:02 <Keverw> Just trying to figure out how to write my own little EWallet type service. I'm a noob at this :( but my product requires it.
 33 2012-07-04 07:06:24 <sipa> quite sure fees are taken from the source account
 34 2012-07-04 07:07:14 <Keverw> like the account I put first in the sendfrom command?
 35 2012-07-04 07:07:21 <sipa> yes
 36 2012-07-04 07:07:23 <Keverw> and if that account can't afford it, error?
 37 2012-07-04 07:07:32 <sipa> no, it'll just go negative iirc
 38 2012-07-04 07:07:59 <Keverw> &.. That's bad?
 39 2012-07-04 07:08:04 <Keverw> You can have negative btw?
 40 2012-07-04 07:08:13 <sipa> account balances can be negative
 41 2012-07-04 07:08:20 <sipa> the total wallet balance obviously can't
 42 2012-07-04 07:08:39 <Keverw> oh& so it takes the rest of the money from some other account on it?
 43 2012-07-04 07:08:57 <sipa> accounts have *nothing* to do with input coin selection
 44 2012-07-04 07:09:06 <Keverw> Just want to prevent overspending I guess.
 45 2012-07-04 07:09:10 <gmaxwell> sipa> quite sure fees are taken from the source account  < yes but IIRC fees can draw it negative.
 46 2012-07-04 07:09:16 <sipa> indeed
 47 2012-07-04 07:09:30 <sipa> the wallet is just one pile of coins, and they are always considered for every transaction
 48 2012-07-04 07:09:41 <sipa> and account balances are just virtual counters
 49 2012-07-04 07:09:55 <gmaxwell> Like 'categories' in a checkbook.
 50 2012-07-04 07:09:58 <Keverw> yeah&. Can't each user get their own wallet somehow?
 51 2012-07-04 07:10:13 <Keverw> or would that require writing my own client instead of RPC?
 52 2012-07-04 07:10:22 <gmaxwell> not with the current software, it's not entirely desirable though because then you couldn't do transactionless moves.
 53 2012-07-04 07:10:33 <gmaxwell> And you'd also spend more on fees.
 54 2012-07-04 07:10:46 <gmaxwell> (because each wallet would be constrained in its coin selection)
 55 2012-07-04 07:11:17 <Keverw> oh& hmm. I guess I could write my own code to figure out what the fee is going to be, and then account for it?
 56 2012-07-04 07:11:34 <Keverw> or is there a fee calc in the RPC?
 57 2012-07-04 07:11:42 <gmaxwell> No. Unfortunately, for part of the same reason the RPC doesn't prompt you.
 58 2012-07-04 07:11:55 <gmaxwell> The fee isn't  "determinstic", at least not as far as you're concerned.
 59 2012-07-04 07:12:15 <Keverw> oh& Hmm.
 60 2012-07-04 07:12:19 <gmaxwell> It depends on the state of all the transactions in the wallet, on the blockchain, and on the specific selected inputs which are randomized when the txn is created.
 61 2012-07-04 07:12:35 <Keverw> I know the litecoin wallet will fail to send the money if you can't cover it fee wise.
 62 2012-07-04 07:12:53 <Keverw> and It uses the same API as Bitcoin I was told when chatting with the owner.
 63 2012-07-04 07:13:14 <Keverw> it doesn't tell you how much the fee was, but just fails to send. if you lower it, it sends.
 64 2012-07-04 07:13:20 <gmaxwell> If the wallet is empty then sure. It's not possible to create a transaction that would make you negative.
 65 2012-07-04 07:13:37 <gmaxwell> but that doesn't apply to accounts, only the whole wallet.
 66 2012-07-04 07:14:01 <Keverw> hmm& Is it normal to be kinda confused when it comes to this?
 67 2012-07-04 07:14:25 <gmaxwell> For most wallet usage patterns the fee is almost always 0. In cases where it's not zero its almost always 0.0005.  Most bitcoin businesses just always charge their users a fixed transaction fee (and make a small profit on it), or don't charge one and just eat it as a cost of doing business.
 68 2012-07-04 07:14:52 <gmaxwell> Keverw: people either ignore it, or they drive themselves crazy over it. There seems to be no middle ground. :)
 69 2012-07-04 07:16:15 <Keverw> Yeah& I guess I'm sorta driving my self crazing thinking about it before really messing with the API
 70 2012-07-04 07:17:47 <gmaxwell> I think the uncertanty bothers people far out of proportion to the actual issue. This has been a major factor in how I think about the fee setup changing in the future. Whatever we do it'll be important to make people feel well informed and in control (er, even if its mostly the same as today) so they don't burn a lot of effort worrying about it. :)
 71 2012-07-04 07:19:49 <Keverw> Yeah. I'm not sure if i should use the account feature or just keep track of it in my own database.
 72 2012-07-04 07:20:50 <Keverw> I was just thinking the built in account feature would be better.
 73 2012-07-04 07:21:38 <gmaxwell> yea, I don't know what to advise. The account feature was _made_ for your kind of usage. ... but it's got a lot of sharp edges. I do recommend that whatever you do you also track everything on your own and keep lots of logs... so if something goes wrong you can pick up the pieces.
 74 2012-07-04 07:22:41 <Keverw> yeah
 75 2012-07-04 07:23:27 <Keverw> I wish the RPC had better fees tho. I know like one site won't allow less then 1 BTC to be sent because of it.
 76 2012-07-04 07:23:40 <Keverw> Are there any work on improving it?
 77 2012-07-04 07:23:46 <gmaxwell> That seems ... a bit dishonest.
 78 2012-07-04 07:23:57 <Keverw> instawallet
 79 2012-07-04 07:24:08 <Keverw> let me check the FAQ.
 80 2012-07-04 07:24:15 <gmaxwell> 0.01 would make a lot of sense.
 81 2012-07-04 07:24:20 <Keverw> oh& lower than 0.01
 82 2012-07-04 07:24:22 <gmaxwell> 1 BTC would make no sense at all.
 83 2012-07-04 07:24:23 <Keverw> I got it wrong.
 84 2012-07-04 07:24:25 <gmaxwell> Right. okay!
 85 2012-07-04 07:24:41 <Keverw> I can do the same on my service maybe?
 86 2012-07-04 07:25:13 <gmaxwell> any txn under 0.01 always needs a fee. (to discourage people from sending thousands of 1e-8 bitcoin payments to everyone and clogging their wallets)
 87 2012-07-04 07:25:54 <gmaxwell> You should though that doesn't guarantee no fees, though thats the only case where you'll always get them.
 88 2012-07-04 07:26:31 <Keverw> even 0.02 needs a fee.
 89 2012-07-04 07:26:44 <Keverw> or so it says in the gui
 90 2012-07-04 07:26:44 <sipa> not necessarily
 91 2012-07-04 07:26:46 <MC1984> why arent the pools asspained about SD just up thier fees
 92 2012-07-04 07:27:00 <gmaxwell> Keverw: No, it depends on priority.
 93 2012-07-04 07:27:52 <Keverw> ?
 94 2012-07-04 07:28:00 <gmaxwell> Keverw: if there is an output less than 0.01 then a fee of 0.0005btc/kb is needed always. Otherwise the priority is measured, and if less than a threshold you need 0.0005btc/kb. Otherwise it's free.
 95 2012-07-04 07:28:20 <sipa> s/free/voluntary/
 96 2012-07-04 07:28:28 <gmaxwell> sipa: thanks, yes.
 97 2012-07-04 07:28:59 <Keverw> oh& So I could really always charge 0.0005 just in case? and maybe make a little profit. like X.X% withdraw fee
 98 2012-07-04 07:29:27 <gmaxwell> The priority depends on the coins you are spending.  sum(value * confirmations) / data_size = priority.  The threshold is so that 1 BTC which was last sent a day ago, as a single input result in a regular size txn being free.
 99 2012-07-04 07:30:14 <gmaxwell> Keverw: yes. sometimes it could be greater than 0.0005 though thats vanishingly rare, and I would expect anyone charging 0.0005 to make a tiny profit on it.
100 2012-07-04 07:31:03 <gmaxwell> Bitparking (a namecoin / bitcoin exchange) charages 0.01 BTC per withdraw.
101 2012-07-04 07:31:37 <Keverw> after I send, does the results tell you if there was a fee?
102 2012-07-04 07:31:57 <Keverw> so then I can take that 0.01 - fee = ???, then add that to my own account
103 2012-07-04 07:31:57 <sipa> you can see the used fee with gettransaction
104 2012-07-04 07:32:16 <gmaxwell> the absolute maximum the mandatory fees can be is 0.05 BTC. ... though this requires the txn have 100 kb of data, which is only generally possible if for example you've been getting lots of 0.01 btc payments and nothing else then send 10 BTC out of them.
105 2012-07-04 07:32:26 <gmaxwell> Keverw: yes you can use the move command to do that.
106 2012-07-04 07:33:03 <Keverw> hmm. I think we are getting somewhere now :)
107 2012-07-04 07:33:21 <gmaxwell> realistically, for most businesses if you're doing enough traffic that fees could even possibly be a concern.. thats a great problem to have. :)
108 2012-07-04 07:33:53 <Keverw> so the send from sends a tx id, not a json object
109 2012-07-04 07:34:03 <Keverw> so I guess I could look that up, and see if a fee was paid?
110 2012-07-04 07:34:24 <gmaxwell> and if your business model has room, I advise just eating the fees and not bothering users.. or if you don't have anywhere to make up for it, just stick on a mandatory withdraw fee.
111 2012-07-04 07:34:32 <gmaxwell> Keverw: yes you can gettransaction on that txid.
112 2012-07-04 07:35:07 <Keverw> oh okay. Does it return if a fee was paid?
113 2012-07-04 07:35:27 <sipa> it just tells you the amount of fee used
114 2012-07-04 07:35:29 <Keverw> I don't think so& but I think it's get stored& or in the gui version
115 2012-07-04 07:35:29 <sipa> which can be 0
116 2012-07-04 07:35:46 <Keverw> yeah. maybe the Wiki doesn't list all the returned stuff?
117 2012-07-04 07:35:53 <sipa> wiki?
118 2012-07-04 07:35:56 <Keverw> I guess I have to just console.log
119 2012-07-04 07:36:02 <Keverw> Looking the the bitcoin.it site
120 2012-07-04 07:36:09 <sipa> yes yes, but which page?
121 2012-07-04 07:36:32 <gmaxwell> Keverw: it tells you the fee.
122 2012-07-04 07:36:44 <sipa> but gettransaction will give you a JSON object with among others a field "fee", and the amount of fee used in that transaction
123 2012-07-04 07:36:50 <gmaxwell> "fee" : -0.01000000,
124 2012-07-04 07:37:28 <Keverw> Let me get you the url. I'm viewing the web on my PC next me me, while using IRC o nmy Mac
125 2012-07-04 07:37:34 <gmaxwell> watch out with the wiki, there is a lot of out of date stuff on it.
126 2012-07-04 07:37:44 <gmaxwell> although I can't remember it not having the fee.
127 2012-07-04 07:37:50 <sipa> neither can i
128 2012-07-04 07:37:57 <keverw> laptop|https://en.bitcoin.it/wiki/Original_Bitcoin_client/API_Calls_list
129 2012-07-04 07:38:03 <gmaxwell> IIRC listtrasnaction doesn't show the field when it's zero.
130 2012-07-04 07:38:10 <keverw> laptop|gettransaction?
131 2012-07-04 07:38:20 <keverw> laptop|That's the one I'm looking at.
132 2012-07-04 07:38:38 <gmaxwell> Keverw: thats just woefully incomplete.
133 2012-07-04 07:38:38 <sipa> it seems incomplete
134 2012-07-04 07:38:49 <Keverw> or is that older?
135 2012-07-04 07:39:09 <Keverw> I guess I'll just have to play some, and console.log stuff
136 2012-07-04 07:39:19 <sipa> i believe fee has been there as long as gettransaction exist
137 2012-07-04 07:39:32 <sipa> you don't need to look at the console, just use gettransaction!
138 2012-07-04 07:39:37 <Keverw> but my VM software is down updating. So I'm gonna install a few VMs to test in. Different wallets.
139 2012-07-04 07:40:11 <Keverw> oh& I'm talking about using Node.js, forgot you can also run the commands from cmd
140 2012-07-04 07:40:41 <gmaxwell> Keverw: e.g. http://pastebin.mozilla.org/1690498
141 2012-07-04 07:41:18 <Keverw> Sweet. Someone should update the Wiki :)
142 2012-07-04 07:41:40 <gmaxwell> it's a wiki, that someone could be you!
143 2012-07-04 07:41:50 <gmaxwell> (it has open editing, you just need to create an account)
144 2012-07-04 07:42:15 <gmaxwell> you're better than me to update it, I already know how all that works. :) you'll find all the dumb spots.
145 2012-07-04 07:42:39 <Keverw> Yeah& I guess I can.
146 2012-07-04 07:42:54 <Keverw> I'm just learning the API working on a project with a business partner.
147 2012-07-04 07:43:08 <Keverw> but later in the future, I do want to work on something like Paypal.
148 2012-07-04 07:43:24 <Keverw> but don't feel like it's a good idea yet. Need to learn more.
149 2012-07-04 07:46:31 <Keverw> done. Added fee :)
150 2012-07-04 08:02:00 <doublec> gmaxwell: it was reduced to 0.001 recently
151 2012-07-04 08:02:20 <doublec> gmaxwell: about a week ago
152 2012-07-04 08:02:47 <sipa> what is reduced to 0.001?
153 2012-07-04 08:03:17 <doublec> sipa: the fee for the namecoin exchange that gmaxwell commented on
154 2012-07-04 08:03:43 <doublec> (fee for withdrawals)
155 2012-07-04 08:06:58 <sipa> ah
156 2012-07-04 08:11:47 <Keverw> I wonder how large test net block chain is...
157 2012-07-04 08:16:06 <ersi> not that large really
158 2012-07-04 08:16:06 <genjix> sipa: is this bip 32?https://github.com/sipa/bitcoin/tree/hdw
159 2012-07-04 08:16:10 <ersi> Keverw: ^
160 2012-07-04 08:16:44 <Keverw> oh. That's good.
161 2012-07-04 08:17:18 <Keverw> I wonder if i could mine some with my crappy Nvidia GeForce 9300M GS 512
162 2012-07-04 08:17:59 <Keverw> once I get one wallet set up in my VM. Can I just copy the VM? Start it, delete the wallet.dat and then start bit coin in the new VM to have a separate wallet?
163 2012-07-04 08:18:08 <Keverw> No need to redownload the chain.
164 2012-07-04 08:19:55 <genjix> sipa: i assume that if your gap limit with type2 keys is Z then with hd keys, you must monitor i branches to a depth of Z
165 2012-07-04 08:21:16 <SomeoneWeird> yes Keverw
166 2012-07-04 08:22:00 <Keverw> oh. okay good!
167 2012-07-04 08:22:08 <genjix> sipa: sum^Z{i_z}
168 2012-07-04 08:24:53 <sipa> genjix: that's part of bip32 yes, but i'm working on other things now
169 2012-07-04 08:25:05 <genjix> ok thx
170 2012-07-04 08:25:21 <sipa> genjix: and bip32 specifies wallet/account/chain/key
171 2012-07-04 08:25:33 <sipa> you can have different gap limits for each
172 2012-07-04 08:26:03 <sipa> for example, inner chains don't need a gap at all
173 2012-07-04 08:26:47 <sipa> and you could watch for let's say 4 accounts, each watch both subchains, the inner one with limit 1 and the outer one with limit 20
174 2012-07-04 08:27:22 <genjix> i see
175 2012-07-04 08:29:50 <sipa> wait... that branch doesn't have key derivation code? i'm sure i already implemented that
176 2012-07-04 08:30:54 <genjix> it would be nice if clients standardised on a key derivation
177 2012-07-04 08:31:01 <genjix> so that's why i was looking at this
178 2012-07-04 08:31:02 <sipa> genjix: my detwallet branch contains more
179 2012-07-04 08:31:06 <genjix> ok
180 2012-07-04 10:25:31 <BlueMatt> jgarzik: yea...but you aren't really gonna get good performance on master anyway, I was just like "we've got this thing threaded...lets see how fast we can make it go"
181 2012-07-04 10:26:23 <doublec> sipa: ping
182 2012-07-04 10:26:29 <sipa> yes?
183 2012-07-04 10:26:54 <doublec> sipa: I'm seeing high cpu usage in a bitcoind in some circumstances due to looping in addrmain.cpp, line 418-ish
184 2012-07-04 10:27:13 <doublec> sipa: there's an infinite while loop there, with a test that does a 'continue'
185 2012-07-04 10:27:18 <doublec> sipa: and it gets stuck in that
186 2012-07-04 10:27:24 <doublec> sipa: does that sound familiar?
187 2012-07-04 10:27:35 <sipa> addrmain.cpp ?
188 2012-07-04 10:27:41 <doublec> addrman.cpp
189 2012-07-04 10:28:03 <sipa> doublec: is your address database near-empty?
190 2012-07-04 10:28:18 <doublec> sipa: it contains one address
191 2012-07-04 10:28:24 <doublec> it's a 'network in a box'
192 2012-07-04 10:28:28 <doublec> with only two nodes
193 2012-07-04 10:28:47 <doublec> that I'm using for testing mining pool software
194 2012-07-04 10:28:54 <sipa> right
195 2012-07-04 10:29:05 <sipa> it should probably deal better with very small address databases
196 2012-07-04 10:29:14 <doublec> ok, just checking that it's known
197 2012-07-04 10:29:16 <doublec> thanks
198 2012-07-04 10:29:18 <sipa> like sleeping if nothing is found after N iterations
199 2012-07-04 10:40:52 <Keverw> hmm. If i had a PHP script that uses Paypal's IPN. I would how hard it would be to convert it to use BTC.
200 2012-07-04 10:44:15 <Eliel_> Keverw: easiest route would probably be either bit-pay or paysius. Their APIs should be quite close to what paypal has. (Said without actually knowing what Paypal's API looks like)
201 2012-07-04 10:44:44 <Keverw> yeah& I was just wondering.
202 2012-07-04 11:18:59 <jrmithdobbs> gmaxwell: what do i have to do to actually use those shallot genn'ed keys?
203 2012-07-04 11:19:12 <jrmithdobbs> gmaxwell: i replaced the contents of hostname and private_key but it looks like tor overwrote on start
204 2012-07-04 11:19:22 <jrmithdobbs> gmaxwell: did i miss something?
205 2012-07-04 11:22:29 <sipa> it shouldn't overwrite them
206 2012-07-04 11:23:13 <jrmithdobbs> errr, it's not changing the private_key but the onion address is not what shallot reported on generation
207 2012-07-04 11:23:20 <jrmithdobbs> (just noticed the rsa key wasn't changing)
208 2012-07-04 11:25:11 <jrmithdobbs> sipa: your seed having issues serving aaaa right now?
209 2012-07-04 11:25:13 <jrmithdobbs> getting timeouts
210 2012-07-04 11:26:30 <sipa> ooh, it seems to have died :(
211 2012-07-04 11:26:31 <jrmithdobbs> oh nm, it's my resolver
212 2012-07-04 11:26:40 <jrmithdobbs> oh weird my resolver was dead too
213 2012-07-04 11:26:46 <BlueMatt> jrmithdobbs: getting timeouts here too
214 2012-07-04 11:26:47 <jrmithdobbs> 99.99% sure i applied that bind patch ;p
215 2012-07-04 11:27:11 <sipa> bah, and my data file corrupted
216 2012-07-04 11:27:18 <sipa> i should fix that bug
217 2012-07-04 11:27:48 <jrmithdobbs> bleh, i need to finish migrating my shit off bind
218 2012-07-04 11:28:02 <jrmithdobbs> the recent 9 versions have just been annoying as fuck
219 2012-07-04 11:28:49 <jrmithdobbs> anyways
220 2012-07-04 11:29:29 <jrmithdobbs> sipa: so, i haven't re-read but the code still wont advert rfc1918 ipv4 addresses right? so i can safely have the tor hidden service redirect to an rfc1918 address and use an existing listening socket instead of having to listen on 127.0.0.1 right?
221 2012-07-04 11:29:44 <jrmithdobbs> s/code/code, but it/
222 2012-07-04 11:30:18 <sipa> it wont advertize rfc1918 addresses no
223 2012-07-04 11:30:28 <sipa> i dont understand thr rest of what you asked
224 2012-07-04 11:31:10 <jrmithdobbs> sipa: i didn't want adding a -bind=10.x.x.x with tor configured to forward the hidden service 8333 -> 10.x.x.x:8333 without it advertising 10.x.x.x is all i mean
225 2012-07-04 11:42:16 <sipa> jrmithdobbs: -bind only binds, it does not affect which address is advertized
226 2012-07-04 11:43:10 <sipa> use -externalip to set what is advertized
227 2012-07-04 11:43:33 <jrmithdobbs> i'm rebuilding for boost re-linkage atm
228 2012-07-04 11:44:27 <jrmithdobbs> sipa: do you have any idea how gmaxwell got private keys from that shallot generator bruteforce thing to actually use the address they said? the key works but it picks a different .onion for it than it said it had generated ;p
229 2012-07-04 11:45:11 <jrmithdobbs> also, gj whoever is getting rid of all those compile warnings ;p
230 2012-07-04 11:46:28 <sipa> jrmithdobbs: iirc, there's a bug in shallot where it sometimes produces bogus addresses
231 2012-07-04 11:47:02 <jrmithdobbs> sipa: bah, that means i accidentally threw away the valid one instead of the invalid one for that pattern, ha
232 2012-07-04 11:52:50 <jrmithdobbs> sipa: can you connect to deezs52qlm2tdbtc.onion ?
233 2012-07-04 11:53:54 <jrmithdobbs> waiting for my os x build to finish so i can test from here, heh
234 2012-07-04 11:55:26 <sipa> jrmithdobbs: trying
235 2012-07-04 11:55:47 <jrmithdobbs> bitcoind compile is now almost 100% warning free on 32bit debian at least, one signed compare warning that doesn't matter left ;p (bitcoin-qt not so much)
236 2012-07-04 11:56:33 <sipa> receive version message: version 60001, blocks=187503, us=0.0.0.0:0, them=deezs52qlm2tdbtc.onion:8333, peer=deezs52qlm2tdbtc.onion:8333
237 2012-07-04 11:56:46 <jrmithdobbs> nice
238 2012-07-04 11:57:00 <sipa> i'm syncing up over tor from you now
239 2012-07-04 11:57:14 <jrmithdobbs> sipa: that box is now on all 3 p2p nets ;p (v4, v6, and tor)
240 2012-07-04 11:57:27 <sipa> nice
241 2012-07-04 11:57:38 <sipa> so is the one running on vps.sipa.be
242 2012-07-04 11:57:39 <BlueMatt> jrmithdobbs: odd, aside from warnings like "src/net.h:374: warning: invalid access to non-static data member ???CMessageHeader::nMessageSize??? of NULL object" 2 warnings in qt/guituil.cpp and some in boost, I dont have any other warnings
243 2012-07-04 11:57:47 <BlueMatt> what else do you have?
244 2012-07-04 11:59:11 <BlueMatt> I get a log of stuff like https://svn.boost.org/trac/boost/ticket/6652
245 2012-07-04 11:59:52 <jrmithdobbs> bitcoinbulletin: i get all sorts on os x
246 2012-07-04 11:59:56 <jrmithdobbs> erm
247 2012-07-04 11:59:58 <jrmithdobbs> BlueMatt: ^
248 2012-07-04 12:00:02 <BlueMatt> can you file bugs for them?
249 2012-07-04 12:00:31 <jrmithdobbs> I don't think any of them actually matter, but ya next time i build i'll keep the log
250 2012-07-04 12:00:49 <BlueMatt> well its nice to get rid of them, even though 99% of them dont matter
251 2012-07-04 12:01:22 <jrmithdobbs> ya, the bitcoind ones went from ~4-5 in a few files to 1 left over the last couple days, always nice to see clean build output
252 2012-07-04 12:01:53 <BlueMatt> probably mostly https://github.com/bitcoin/bitcoin/pull/1548
253 2012-07-04 12:03:19 <jrmithdobbs> yes mostly, something in main.h too
254 2012-07-04 12:04:11 <jrmithdobbs> ../bitcoin/src/net.h:374: warning: invalid access to non-static data member 'CMessageHeader::nMessageSize' of NULL object
255 2012-07-04 12:04:32 <BlueMatt> yea, thats the only one I see thats left
256 2012-07-04 12:05:06 <jrmithdobbs> also on 380 and 381 with nCheckSum same thing
257 2012-07-04 12:05:24 <BlueMatt> yep
258 2012-07-04 12:05:45 <jrmithdobbs> ya actually after the last pull i did that's it
259 2012-07-04 12:06:05 <BlueMatt> nice
260 2012-07-04 12:06:09 <jrmithdobbs> nice work
261 2012-07-04 12:06:24 <jrmithdobbs> and that's not even using the official build method, that's just opening the .pro file in qt creator ;p
262 2012-07-04 12:06:34 <jrmithdobbs> and building via default release settings
263 2012-07-04 12:07:15 <jrmithdobbs> (my build/macports env is already pointed at the right places for all the deps)
264 2012-07-04 12:08:55 <BlueMatt> heh, I love british parliamentary testimonies, so much more direct, essentially: "should you be in jail right now?"
265 2012-07-04 12:21:11 <jrmithdobbs> usr/local/bin/bitcoind -datadir=/home/bitcoin -printtoconsole -maxconnections=512 -tor=127.0.0.1:9050 -bind=10.xx.xx.yy -bind=69.41.162.72 -bind=127.0.0.1 -bind=[2001:470:1f0f:9a2::1] -externalip=69.41.162.72 -externalip=[2001:470:1f0f:9a2::1] -externalip=deezs52qlm2tdbtc.onion -addnode=6hgmaxwellgpv2oe.onion -addnode=kjy2eqzk4zwi5zd3.onion -addnode=[2001:888:10:bbb::2]
266 2012-07-04 12:21:52 <jrmithdobbs> (that last addnode is a host -t aaaa to make sure i get at least one ipv6 peer off the bat, don't know that it's necessary any more)
267 2012-07-04 12:23:20 <jrmithdobbs> that doesn't look to be actually listing either of the onion peers in getpeerinfo ?
268 2012-07-04 12:24:38 <sipa> it should show outgoing onion connections
269 2012-07-04 12:24:46 <sipa> (if there are any)
270 2012-07-04 12:25:12 <jrmithdobbs> sipa: that second one is your node right?
271 2012-07-04 12:25:56 <BlueMatt> how does -onlynet=tor bootstrap?
272 2012-07-04 12:26:07 <sipa> it doesn't :)
273 2012-07-04 12:26:57 <sipa> jrmithdobbs: kjy2... yes
274 2012-07-04 12:27:40 <jrmithdobbs> 2012-07-04_14:26:30.04942 SOCKS5 connecting kjy2eqzk4zwi5zd3.onion
275 2012-07-04 12:27:41 <BlueMatt> sipa: can we put the tor-formatted ipv6's in dnsseeds?
276 2012-07-04 12:27:42 <jrmithdobbs> 2012-07-04_14:26:30.04946 ERROR: Proxy error: general failure
277 2012-07-04 12:27:43 <jrmithdobbs> hmmmm
278 2012-07-04 12:28:15 <jrmithdobbs> debian just pushed a torrc update, let me see if they set a random password on the proxy or some shit
279 2012-07-04 12:28:35 <sipa> BlueMatt: yes, but not very useful, as people behind tor shouldn't be doing DNS lookups
280 2012-07-04 12:28:54 <sipa> (and you cannot do DNS lookups through tor)
281 2012-07-04 12:29:03 <jrmithdobbs> you kind of can
282 2012-07-04 12:29:06 <BlueMatt> mmm, I thought there was some extension to do that...
283 2012-07-04 12:29:10 <BlueMatt> guess we dont use it though
284 2012-07-04 12:29:32 <sipa> you can ask an exit node to resolve an address for you, but that's not a lookup (you don't get to see the result yourself, you just get connected to it)
285 2012-07-04 12:29:58 <BlueMatt> yea...
286 2012-07-04 12:30:06 <sipa> and you can implement a DNS client yourself, and connect to an open TCP-capable DNS recursive server via an exit node
287 2012-07-04 12:30:39 <sipa> but both are hacky and suboptimal
288 2012-07-04 12:30:42 <BlueMatt> yuck...I though I remembered someone saying there was some special tor-specific proxy command thinggy...
289 2012-07-04 12:30:46 <jrmithdobbs> sipa: well, we could make it workable with dnscurve actually, since the seed would be encrypted data channel all the way to client
290 2012-07-04 12:30:47 <BlueMatt> nevermind
291 2012-07-04 12:30:51 <jrmithdobbs> but that's a lot of code for not a lot of benefit
292 2012-07-04 12:30:55 <sipa> meh
293 2012-07-04 12:30:59 <jrmithdobbs> (on dns over ssl)
294 2012-07-04 12:31:01 <jrmithdobbs> s/on/or/
295 2012-07-04 12:31:20 <sipa> i'd just wait until some stable set of onion nodes is up, and then add some as hardcoded seeds
296 2012-07-04 12:31:41 <BlueMatt> yea
297 2012-07-04 12:31:50 <sipa> and make oneshot connections to them
298 2012-07-04 12:31:58 <jrmithdobbs> sipa: or we could setup some tor directory servers that hand out the seeds
299 2012-07-04 12:32:09 <sipa> what's the difference?
300 2012-07-04 12:32:33 <jrmithdobbs> not much, just thinking out loud because hard coded seeds seems like a poor solution ;p
301 2012-07-04 12:32:52 <sipa> well, otherwise the directory server is hardcoded, no?
302 2012-07-04 12:33:35 <jrmithdobbs> ya but in the same way tors are, tbqh we may even be able to work something out with the tor project folk that run the directory servers once the code's mature
303 2012-07-04 12:34:20 <jrmithdobbs> seems that being able to bootstrap an overlaid network is something they'd be interested in helping with
304 2012-07-04 12:35:07 <jrmithdobbs> anyone asked any of them how they think would work well overlaid on tor? heh
305 2012-07-04 12:36:12 <sipa> well, if someone runs a tor-only server with stable uptime and such, and you add the address hardcoded to the client, and the client just connects to it, requests a getaddr, waits for an addr response, and disconnects... more or less the same thing
306 2012-07-04 12:36:22 <sipa> (which is what oneshot does)
307 2012-07-04 12:37:28 <jrmithdobbs> sipa: but that's no better than doing the dns query over tor
308 2012-07-04 12:37:52 <jrmithdobbs> is it?
309 2012-07-04 12:38:03 <sipa> this way it never leaves the onion network
310 2012-07-04 12:38:26 <jrmithdobbs> so router before can't read the plaintext because no exit
311 2012-07-04 12:38:40 <jrmithdobbs> why not setup a dnssed with an onion address then?
312 2012-07-04 12:39:04 <jrmithdobbs> that way seed can still do connectability testing
313 2012-07-04 12:39:04 <sipa> that'd require a dns client implementation in bitcoin
314 2012-07-04 12:39:26 <sipa> i'd rather add minimal bitcoin p2p protocol ability in the dns seed
315 2012-07-04 12:39:44 <sipa> (no relaying of anything, only version/verack getaddr/addr supported)
316 2012-07-04 12:40:39 <jrmithdobbs> that sounds like a good solution, actually, over tor it's even authenticated in a way that isn't true of a direct tcp connection
317 2012-07-04 12:41:02 <BlueMatt> jrmithdobbs: its somewhat ugly, but if does fix the warnings: https://github.com/bitcoin/bitcoin/pull/1555
318 2012-07-04 12:41:04 <sipa> indeed
319 2012-07-04 12:42:09 <jrmithdobbs> tbqh once it's tested a bit more i think tor only connections should be recomended, it does nothing but help network help, solves nat and everything
320 2012-07-04 12:42:26 <jrmithdobbs> s/network help/network health/
321 2012-07-04 12:44:16 <gribble> New news from bitcoinrss: TheBlueMatt opened pull request 1555 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/1555>
322 2012-07-04 12:44:20 <BlueMatt> except for the fact that you would make everyone install tor, which is kinda ridiculous, and miners  would see significantly worse propagation times (which could be worked around using the bitcoin backbone thing)
323 2012-07-04 12:47:14 <jrmithdobbs> BlueMatt: for everyone but miners it makes sense, though, basically anyone currently using p2p would benefit from at least having tor connectivity if not being tor only
324 2012-07-04 12:47:52 <BlueMatt> most people dont want to install additional optional software if the only difference is they now dont have to use upnp...
325 2012-07-04 12:47:55 <jrmithdobbs> and having nodes on all transports is good for validation reasons, etc, but for the average end user tor-only connections make sense if it can be made easy (and it can by installing on top of tails or similar)
326 2012-07-04 12:48:01 <sipa> ok, todo list for bitcoin-seeder: 1) rework scheduling using less magic constants  2) when querying a node for hosts, also push some good addresses to it  3) support listening for p2p connections  4) safe database dumping
327 2012-07-04 12:58:04 <TD> sipa: what is bitcoin-seeder?
328 2012-07-04 12:58:24 <BlueMatt> dnsseed crawler/server
329 2012-07-04 12:58:47 <jrmithdobbs> sipa: hrm, this is weird, the connect.c i use as a proxy command for ssh works fine through this tor socks proxy, but bitcoin keeps generating this in the tor log:
330 2012-07-04 12:58:50 <jrmithdobbs> Jul 04 09:57:26.000 [warn] Socks version 0 not recognized. (Tor is not an http proxy.)
331 2012-07-04 12:58:53 <jrmithdobbs> Jul 04 09:57:26.000 [warn] Fetching socks handshake failed. Closing.
332 2012-07-04 12:59:28 <sipa> hmm?
333 2012-07-04 12:59:45 <sipa> TD: yes, my bitcoin network crawler + integrated dns seed server
334 2012-07-04 13:00:44 <sipa> jrmithdobbs: what does that connect.c do?
335 2012-07-04 13:02:08 <jrmithdobbs> sipa: it works with ssh's ProxyCommand option like this: alias torssh='ssh -2AYC -oProxyCommand="/usr/local/bin/socksconnect -4 -R remote -S localhost:9050 %h %p"'
336 2012-07-04 13:02:25 <jrmithdobbs> so it initiates the socks connection for session using the hostname/ip and port passed to it
337 2012-07-04 13:02:39 <jrmithdobbs> (-R is wether to use remote or local name resolution)
338 2012-07-04 13:02:42 <sipa> -4 means SOCKS4 ?
339 2012-07-04 13:03:00 <sipa> no, it probably means IPv4
340 2012-07-04 13:03:02 <sipa> hmm
341 2012-07-04 13:03:13 <jrmithdobbs> yes, it works with -5 too, just holdover in my alias when name resolution didn't work in tor for socks5
342 2012-07-04 13:03:22 <TD> sipa: cool
343 2012-07-04 13:03:35 <TD> sipa: what logic does it use? i'm starting to think about how to manage the local addr db on mobile phones
344 2012-07-04 13:03:39 <jrmithdobbs> this
345 2012-07-04 13:03:50 <TD> i'm wondering if it should just be the same as satoshis code, or whether something totally different makes more sense
346 2012-07-04 13:04:05 <jrmithdobbs> sipa: source is on there
347 2012-07-04 13:04:15 <sipa> TD: bitcoind switched to "addrman" (which i wrote) in v0.5
348 2012-07-04 13:04:37 <sipa> TD: addrman.h has some documentation about the design
349 2012-07-04 13:04:59 <sipa> the seeder keeps all addresses it knows around for a very long time, with much more statistics than a normal node needs
350 2012-07-04 13:05:08 <gavinandresen> BlueMatt: what happens if I addnode an IP address?
351 2012-07-04 13:06:18 <TD> thanks
352 2012-07-04 13:06:22 <jrmithdobbs> sipa: if it matters: Jul 04 10:06:04.965 [notice] Tor v0.2.3.18-rc (git-a2015428e4698ff9) running on Linux.
353 2012-07-04 13:06:37 <sipa> gavinandresen: what you expect, probably
354 2012-07-04 13:06:45 <sipa> gavinandresen: as the IP resolves to just one address
355 2012-07-04 13:07:08 <BlueMatt> gavinandresen: connects to that ip and keeps it open, if you are asking about getaddednodeinfo output, look at https://github.com/bitcoin/bitcoin/pull/1549#issuecomment-6751340 at the first addednde "10.232.4.10"
356 2012-07-04 13:08:08 <gavinandresen> BlueMatt: hmmm.   See  http://pastebin.com/pCzLWXy2
357 2012-07-04 13:08:19 <gavinandresen> Maybe I was already connected to that peer?
358 2012-07-04 13:08:31 <jrmithdobbs> sipa: only thing i don't like about your seeder is that it speaks dns directly instead of outputing zones at an interval or similar
359 2012-07-04 13:08:46 <BlueMatt> gavinandresen: onetry doesnt add the node to addednodeinfo, it only connects once and moves on
360 2012-07-04 13:08:51 <gavinandresen> Ah, ok
361 2012-07-04 13:08:54 <BlueMatt> its not considered an "added node" at that point
362 2012-07-04 13:08:55 <sipa> jrmithdobbs: that was exactly the reason why i wrote it - i wanted every query to return a different result
363 2012-07-04 13:09:20 <jrmithdobbs> sipa: you can get same effect with options on the dns servers if you just write out a zone file though ;p
364 2012-07-04 13:09:41 <gavinandresen> BlueMatt: help for getaddednodeinfo should mention that
365 2012-07-04 13:09:52 <BlueMatt> gavinandresen: was just going to add that, yep
366 2012-07-04 13:09:53 <TD> sipa: btw just a heads up. apparently somebody wrote a paper (in german) claiming that Windows 7 automatically changes IPv6 addresses every N hours by default for "privacy reasons", even if there are active TCP connections on it
367 2012-07-04 13:10:03 <jrmithdobbs> sipa: and then you can dnssec sign and such ;p
368 2012-07-04 13:10:06 <BlueMatt> TD: old news
369 2012-07-04 13:10:12 <TD> orly
370 2012-07-04 13:10:37 <BlueMatt> it losing tcp connections is new, but changing ipv6 addresses it has always done
371 2012-07-04 13:10:38 <D34TH> my ipv6 addr hasnt changed once
372 2012-07-04 13:10:39 <D34TH> D:
373 2012-07-04 13:10:54 <jrmithdobbs> err ya
374 2012-07-04 13:11:03 <sipa> jrmithdobbs: my seeder can serve >20k DNS queries per second, with each getting a different subset of the known addresses
375 2012-07-04 13:11:04 <jrmithdobbs> i thought it kept the old addresses until connections dropped
376 2012-07-04 13:11:06 <BlueMatt> it keeps 2, one permanent-ish mac-address-derived one (the way linux does), and one that is "private" which rotates
377 2012-07-04 13:11:31 <Dagger2> it doesn't technically change every N hours, but rather it adds a new address and then deletes old ones after (on the default settings) a week
378 2012-07-04 13:11:36 <jrmithdobbs> sipa: unbound can server >500k a second on crap hardware with different query for every request
379 2012-07-04 13:11:39 <jrmithdobbs> sipa: ;p
380 2012-07-04 13:11:46 <jrmithdobbs> s/server/serve/
381 2012-07-04 13:11:51 <Dagger2> I don't know if it will hang on to an address even after a week if there are connections open to it though
382 2012-07-04 13:11:54 <TD> hmm
383 2012-07-04 13:12:04 <D34TH> 2001:XXX:X:9b1::2
384 2012-07-04 13:12:09 <D34TH> it has naver changed
385 2012-07-04 13:12:18 <BlueMatt> well thats because you manually set it to ::2
386 2012-07-04 13:12:19 <D34TH> **never
387 2012-07-04 13:12:30 <jrmithdobbs> sipa: just as an example of one implementation (and that number is probably low, tbqh)
388 2012-07-04 13:12:32 <D34TH> thats what i was automatically assigned
389 2012-07-04 13:12:35 <BlueMatt> or its a tunnel endpoint, in which case it was manually set to ::2
390 2012-07-04 13:12:51 <BlueMatt> (by you or the tunnel software)
391 2012-07-04 13:13:15 <sipa> jrmithdobbs: to be honest, i'm more sure about the efficiency of my implementation than the test setup i used to benchmark it
392 2012-07-04 13:13:52 <sipa> jrmithdobbs: anyway, pull requests welcome if you want it to write out zone files
393 2012-07-04 13:14:27 <sipa> TD: i've heard something like that before as well
394 2012-07-04 13:14:44 <jrmithdobbs> sipa: ya, i've not had time to do anything of the sort lately, your impl is just hard to shoehorn into existing dns servers, it works though
395 2012-07-04 13:14:57 <jrmithdobbs> s/hard/annoying/
396 2012-07-04 13:15:01 <sipa> jrmithdobbs: the DNS server code is completely separate
397 2012-07-04 13:15:22 <sipa> disable it, and add a thread that occassionally queries the database and dumps it
398 2012-07-04 13:15:40 <jrmithdobbs> oh? you must have re architected majorly since i last looked?
399 2012-07-04 13:15:52 <sipa> all dns code is in dns.c
400 2012-07-04 13:15:56 <sipa> and always was
401 2012-07-04 13:16:03 <BlueMatt> gavinandresen: pushed
402 2012-07-04 13:16:45 <sipa> main.cpp provides a callback function to dns.c that queries the database
403 2012-07-04 13:16:55 <jrmithdobbs> anyways, back to this tor issue, any ideas?
404 2012-07-04 13:17:31 <jrmithdobbs> i'm about to see if it connects correctly on my macbook through the tor proxy ... it's running whatever latest browser bundle stable is
405 2012-07-04 13:17:58 <jrmithdobbs> that should be a good version of tor on that server though, everything else seems to work through it :(
406 2012-07-04 13:18:37 <sipa> TD: i guess we'll see what happens when 0.7 is deployed on Win7+ setups
407 2012-07-04 13:18:42 <TD> yeah
408 2012-07-04 13:19:10 <jrmithdobbs> you've got a while before anyone but comcast turns it up anyways really in the us at least
409 2012-07-04 13:19:45 <TD> i don't think it'd be a big issue for us anyway. just i didn't know about that before
410 2012-07-04 13:20:28 <BlueMatt> I believe linux also supports (though not sure about defaults) the rotating ipv6 address "for privacy reasons" stuff
411 2012-07-04 13:21:24 <Dagger2> BlueMatt: it does; you normally have to turn it on, but recent versions of Network Manager (and thus Ubuntu 12.05) turn it on by default
412 2012-07-04 13:21:45 <TD> ugh
413 2012-07-04 13:21:54 <BlueMatt> yea...
414 2012-07-04 13:21:54 <TD> this could result in a lot of unnecessary network thrashing
415 2012-07-04 13:22:23 <BlueMatt> and it really doesnt help privacy...
416 2012-07-04 13:23:15 <jrmithdobbs> sipa: it's just the default torrc from the package in wheezy with hidden service and port 8333 127.0.0.1:8333 configured right now, am I missing something?
417 2012-07-04 13:23:40 <sipa> jrmithdobbs: i haven't looked at your socksconnect stuff
418 2012-07-04 13:23:41 <BlueMatt> Dagger2: odd, latest debian testing network-manager doesnt though
419 2012-07-04 13:24:32 <jrmithdobbs> sipa: it just connects via socks proxy on behalf of ssh (or other things that have a proxycommand like option), torsocks/etc work as well browser configured to use that socks proxy ;/
420 2012-07-04 13:24:47 <Dagger2> BlueMatt: hm. I'm going on second-hand knowledge, since none of my systems have Network Manager installed
421 2012-07-04 13:25:13 <BlueMatt> I wouldnt be surprised if either debian is turning it off or canonical is turning it on, though
422 2012-07-04 13:25:29 <jrmithdobbs> i'm saying my tor setup seems to work, but bitcoin keeps throwing a proxy connect error using -tor=127.0.0.1:9050 and: tcp        0      0 127.0.0.1:9050          0.0.0.0:*               LISTEN      25291/tor
423 2012-07-04 13:25:50 <sipa> BlueMatt: canonical turning it on would be my guess
424 2012-07-04 13:25:59 <BlueMatt> yea, probably
425 2012-07-04 13:27:49 <jrmithdobbs> sipa: it works correctly on os x using the current stable tor browser bundle for outbound connections
426 2012-07-04 13:28:10 <jrmithdobbs> sipa: I'm connected to the hidden service on the node that can't make outbound tor connections from my laptop ...
427 2012-07-04 13:28:20 <jrmithdobbs> using same commit
428 2012-07-04 13:28:24 <sipa> hmm, weird
429 2012-07-04 13:29:42 <jrmithdobbs> works fine with this version of tor:
430 2012-07-04 13:29:43 <jrmithdobbs> Jul 04 10:29:13.600 [notice] Tor v0.2.2.37 (git-fce6eb1c44e87bc2). This is experimental software. Do not rely on it for strong anonymity. (Running on Darwin x86_64)
431 2012-07-04 13:29:46 <jrmithdobbs> Tor version 0.2.2.37 (git-fce6eb1c44e87bc2).
432 2012-07-04 13:29:58 <jrmithdobbs> but not this one:
433 2012-07-04 13:29:59 <jrmithdobbs> $ tor --version
434 2012-07-04 13:30:02 <jrmithdobbs> Jul 04 10:29:50.169 [notice] Tor can't help you if you use it wrong! Learn how to be safe at https://www.torproject.org/download/download#warning
435 2012-07-04 13:30:05 <jrmithdobbs> Tor version 0.2.3.18-rc (git-a2015428e4698ff9).
436 2012-07-04 13:31:04 <jrmithdobbs> is it how you're doing the socks handshake maybe? (like i said other things work fine with both) i know in the 2.3 series there's a lot of work going on on isolating different connection types across identities and such and i think there may be more strict error checking on the socks connections maybe?!
437 2012-07-04 13:31:09 <jrmithdobbs> weird
438 2012-07-04 13:31:43 <jrmithdobbs> also sync'ing over tor is surprisingly not completely slow, wow
439 2012-07-04 13:32:47 <BlueMatt> jrmithdobbs: syncing is rarely network-limited...even if none of it is threaded
440 2012-07-04 13:37:22 <jrmithdobbs> BlueMatt: oh it's significantly slower than when i do it straight over the vpn to that box
441 2012-07-04 13:38:11 <BlueMatt> yea...because its not properly threaded
442 2012-07-04 13:38:27 <BlueMatt> it shouldnt be limited by network speed, but it does slow it down
443 2012-07-04 13:38:58 <jrmithdobbs> sipa: it's weird, everything else i try works fine
444 2012-07-04 13:43:31 <jrmithdobbs> just double checked the iceweasel package on that box and it works correctly as both socks4 and 5
445 2012-07-04 13:44:58 <gribble> New news from bitcoinrss: Diapolo opened issue 1556 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/issues/1556>
446 2012-07-04 13:54:44 <jrmithdobbs> sipa: i'm comparing how that connect.c vs the socks code in bitcoin do the handshake
447 2012-07-04 13:55:16 <jrmithdobbs> sipa: something in the bitcoin version is causing the tor proxy to return the 0x01 error bit on connect() request
448 2012-07-04 13:59:45 <jrmithdobbs> sipa: i think I see the problem
449 2012-07-04 14:00:36 <jrmithdobbs> sipa: testing
450 2012-07-04 14:11:05 <jrmithdobbs> hrm, nope
451 2012-07-04 14:19:40 <midnightmagic> hah!
452 2012-07-04 14:23:41 <BlueMatt> dammnnn...my bitcoin-qt built using clang's lto is only 4.6M (vs gcc's lto is 6.9M)
453 2012-07-04 14:24:41 <drizztbsd> BlueMatt: gcc with -Os + strip is smaller
454 2012-07-04 14:25:21 <drizztbsd> gcc with -O2/-O3 creates PERFORMANT binaries, not small binaries
455 2012-07-04 14:25:51 <BlueMatt> yea, and I didnt run clang to target size either
456 2012-07-04 14:26:11 <BlueMatt> (sorry, should be 4.0M vs 5.6M)
457 2012-07-04 14:28:16 <BlueMatt> clang-lto: " block index            6523ms" gcc-lto: " block index            8497ms"
458 2012-07-04 14:28:43 <drizztbsd> wow
459 2012-07-04 14:28:44 <BlueMatt> thats gcc -O3, clang -O4
460 2012-07-04 14:29:16 <jrmithdobbs> sipa: -tor= doesn't seem to honor -socks=
461 2012-07-04 14:30:22 <BlueMatt> wallet load goes from 106ms to 89ms and address load goes from 22ms to 15ms
462 2012-07-04 14:39:03 <jrmithdobbs> sipa: if i specific -socks after -tor it honors it, but it still fails to connect on 2.3.18-rc :(
463 2012-07-04 14:39:32 <jrmithdobbs> via 5 or 4
464 2012-07-04 15:09:28 <kkulhavy> hi all
465 2012-07-04 15:09:40 <kkulhavy> We want to set up our business to accept bitcoins
466 2012-07-04 15:10:02 <kkulhavy> But I couldn't find an information how to do it
467 2012-07-04 15:10:09 <kkulhavy> Not even in the "Merchants sections"
468 2012-07-04 15:10:13 <kkulhavy> What do we have to do=
469 2012-07-04 15:19:58 <jgarzik> sipa: so...  ACK #1554 (dup GetHash) for now?
470 2012-07-04 16:31:50 <galambo> http://cms.web.cern.ch/news/observation-new-particle-mass-125-gev
471 2012-07-04 16:33:12 <PK> so Cern won the race to finding the higgs boson.
472 2012-07-04 16:33:19 <galambo> it will be very neat if they did infact find it
473 2012-07-04 16:33:50 <galambo> happy 4th of july ;O
474 2012-07-04 16:34:14 <galambo> 125 gev thats a big firecracker
475 2012-07-04 16:36:48 <galambo> for a subatomic particle
476 2012-07-04 16:43:42 <jgarzik> hrm
477 2012-07-04 16:44:08 <jgarzik> what is the quickest way -- without spending bitcoins -- to verify that bitcoin address 123456789 is present in the bitcoind wallet?
478 2012-07-04 16:45:35 <ersi> Isn't that what the sign message feature is for?
479 2012-07-04 16:54:28 <luke-jr> jgarzik: do you really need "quickest" (ie, don't spend CPU signing, or require unlocking the wallet) or is ersi's solution OK?
480 2012-07-04 16:56:58 <jgarzik> easiest from command line.  I found and used signmessage about the time ersi was typing :)
481 2012-07-04 16:57:21 <jgarzik> signmessage correctly distinguishes between (a) you have the private key, (b) you don't have the private key, and (c) invalid address
482 2012-07-04 16:57:34 <jgarzik> so it functions as a super-sized validateaddress :)
483 2012-07-04 16:58:09 <ersi> Glad to hear
484 2012-07-04 17:29:56 <jrmithdobbs> sipa: it ok if i post pcaps with your onion address in them to tor's trac to try and track this down?
485 2012-07-04 17:30:03 <jrmithdobbs> gmaxwell: same question ^
486 2012-07-04 17:30:21 <gmaxwell> jrmithdobbs: fine for mine.
487 2012-07-04 17:30:57 <jrmithdobbs> the socks handshake looks right on the wire afaict, it's got to be related to the circuit sep logic stuff in .2.3 i think
488 2012-07-04 17:33:11 <jrmithdobbs> works fine on .2.2.37
489 2012-07-04 17:39:24 <gmaxwell> 09:28 < BlueMatt> thats gcc -O3, clang -O4
490 2012-07-04 17:39:46 <gmaxwell> Seriously, every time I've seen a major speedup with clang it was miscompiling the code.
491 2012-07-04 17:39:56 <gmaxwell> Check carefully. :)
492 2012-07-04 17:44:49 <BlueMatt> gmaxwell: well there aren't any obvious problems cropping up, downloading blocks appears to work
493 2012-07-04 17:46:30 <BlueMatt> gmaxwell: somehow I dont see clang being /that/ unreliable...I mean every iphone app is compiled using it and most mac apps...
494 2012-07-04 17:47:08 <gmaxwell> It's reliable, I didn't say it wasn't, I said that every time I've seen a major speedup out of it it was miscompiling. Important distinction.
495 2012-07-04 17:47:49 <BlueMatt> well I find it hard to believe that even rare miscompiles happen that easily
496 2012-07-04 17:47:56 <gmaxwell> In particular, for a while it had problems with reordering locks out of loops, which created really subtle bugs.
497 2012-07-04 17:48:33 <BlueMatt> mmm, fun
498 2012-07-04 17:48:50 <gmaxwell> For libtheora it was manging to just not emit some ASM blocks.
499 2012-07-04 17:49:02 <BlueMatt> heh
500 2012-07-04 17:49:11 <BlueMatt> what version was this?
501 2012-07-04 17:49:59 <gribble> New news from bitcoinrss: TheBlueMatt opened pull request 1557 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/1557>
502 2012-07-04 17:50:28 <gmaxwell> An older one, I don't know the clang version, though it was the apple xcode release prior to the most recent I think.  though apple's xcode stuff lags clang development, so it was a major pain.
503 2012-07-04 17:50:57 <BlueMatt> hmm...well maybe they've fixed some stuff, I think xcode releases are only yearly...
504 2012-07-04 17:52:18 <gmaxwell> Oh sure. In any case, it's useful to actually look for where the speedup is and look at the emitted code. You may discover that it's breaking something, or you may find whatever moronic thing the code does that clang really can optimize and gcc can't and by fixing it make both better.
505 2012-07-04 17:52:43 <BlueMatt> yep, was gonna profile it later, hadnt gotten around to it yet
506 2012-07-04 17:57:03 <jrmithdobbs> sipa: ping
507 2012-07-04 18:04:24 <paraipan> ;;ping
508 2012-07-04 18:04:25 <gribble> pong
509 2012-07-04 18:04:41 <paraipan> nice :)
510 2012-07-04 18:05:33 <Cubox> hmm, I asked here a few days ago, and you said me to ask on forums for my fork's problem. I don't have an answer. There is others ways to get some help (mailing lists ?) ?
511 2012-07-04 18:08:28 <jrmithdobbs> ;;tell sipa hey is it ok if your onion address is in a trac entry on torproject.org
512 2012-07-04 18:08:45 <jrmithdobbs> did i do that right
513 2012-07-04 18:13:42 <BlueMatt> jrmithdobbs: should be ;;later tell ...
514 2012-07-04 18:14:01 <jrmithdobbs> ;;later tell sipa hey is it ok if your onion address is in a trac entry on torproject.org
515 2012-07-04 18:14:02 <gribble> The operation succeeded.
516 2012-07-04 18:14:11 <gmaxwell> Cubox: did you ask on the forums?
517 2012-07-04 18:15:07 <Cubox> gmaxwell: yes
518 2012-07-04 18:15:18 <Cubox> July 03, 2012, 05:30:24 PM
519 2012-07-04 18:15:47 <Cubox> I got 5 or 4 answers, with most of they saying "oh no, please, not a new scam fork"
520 2012-07-04 18:16:02 <Cubox> but not the excepted answer :(
521 2012-07-04 18:16:11 <gmaxwell> hahah
522 2012-07-04 18:16:24 <Cubox> I said that is just for fun >.<
523 2012-07-04 18:17:30 <Cubox> "I'm making a fork of litecoin, for fun, and looking how bitcoin works." -> "SCAM ALERT. PONZI!" or "Please god not another pointless fork."
524 2012-07-04 18:19:40 <sipa> jrmithdobbs: -tor is always socks5
525 2012-07-04 18:20:56 <sipa> jgarzik: validateaddress also distinguishes between those 3 cases
526 2012-07-04 18:21:26 <jrmithdobbs> sipa: ya i saw that
527 2012-07-04 18:21:38 <sipa> jrmithdobbs: yeah you can post the onion address
528 2012-07-04 18:21:54 <jrmithdobbs> https://trac.torproject.org/projects/tor/ticket/6297
529 2012-07-04 18:21:59 <jrmithdobbs> if you want to see the packet dumps
530 2012-07-04 18:22:10 <jrmithdobbs> i'm honestly not sure which side the problem is on
531 2012-07-04 18:23:12 <jrmithdobbs> sipa: kinda realized it's a public node whose address is propigating, asking is kind of unnecessary and posted it a couple minutes before you answered
532 2012-07-04 18:23:22 <jrmithdobbs> heh
533 2012-07-04 19:07:31 <BlueMatt> gmaxwell: had to reload my chain so Ive only profiled -loadblock, but from looking at this (where there is only a very small difference in performance) the performance difference is because llvm is inlining a /ton/ of stuff
534 2012-07-04 19:31:47 <devrandom> hi BlueMatt
535 2012-07-04 19:42:43 <devrandom> (just saying hi, nothing special)
536 2012-07-04 19:51:01 <gmaxwell> BlueMatt: ah, you can cause gcc to do that too.. but it usually reduces performance due to l1i/l2 cache pressure.
537 2012-07-04 19:54:51 <BlueMatt> whats up devrandom?
538 2012-07-04 19:55:17 <BlueMatt> gmaxwell: yep, though it appears to be doing it at least somewhat smartly and actually succeeding
539 2012-07-04 19:59:42 <BlueMatt> though there do appear to be some obvious functions that are only ever called in one place that could be trivially inlined (Id think)
540 2012-07-04 19:59:59 <BlueMatt> but...meh maybe its getting duplicated due to some other really weird optimization...
541 2012-07-04 20:14:54 <devrandom> BlueMatt: just saying hi.  anything new and exciting?
542 2012-07-04 20:15:34 <BlueMatt> devrandom: around here? not much lots of pulls getting merged, but more piling up faster, so...nothing new
543 2012-07-04 20:59:30 <BlueMatt> gmaxwell: it appears to have done quite a number on the serialization code, with CTransaction::GetHash() nearly dropping by 3x in total cpu cycles and being dominated by SHA256 instead of serialization
544 2012-07-04 21:00:30 <[Tycho]> :)
545 2012-07-04 21:00:38 <BlueMatt> meh, its easy enough to remove duplicate GetHash() calls without doing that
546 2012-07-04 21:01:08 <sipa> also easy enough to add a counter as well, and see how many rehashings it saves us in practice
547 2012-07-04 21:01:10 <BlueMatt> but if llvm can do this much for our serialization, there has to be a better way to do it...
548 2012-07-04 21:01:16 <gmaxwell> might be fun to wrap sha256 with code that screams if you ever send it the same data twice.
549 2012-07-04 21:02:08 <BlueMatt> its pretty easy to see when looking at it in kcachegrind
550 2012-07-04 21:02:15 <BlueMatt> all the sha256 blocks are pretty evident
551 2012-07-04 21:03:58 <BlueMatt> I like when idiots protesting banks are put on the news "Barclays is stupid because they are falsely [DECREASING] interest rates which hurts everyone who has a loan" wait????
552 2012-07-04 21:04:58 <freewil> well they should be protesting
553 2012-07-04 21:05:07 <freewil> they just got their facts a little mixed up
554 2012-07-04 21:06:01 <BlueMatt> I have no problem with people protesting people committing fraud, I do have a problem with people prtesting when they have no clue wtf they are talking about
555 2012-07-04 21:07:38 <freewil> i can agree with that
556 2012-07-04 21:08:20 <freewil> another consequence of fradualent interest rates, social security trust fund will peak 6 years earlier than projected
557 2012-07-04 21:08:23 <freewil> http://www.zerohedge.com/contributed/2012-07-04/bernanke-%E2%80%93-my-goal-wreck-social-security
558 2012-07-04 21:09:39 <freewil> dont read it too literally
559 2012-07-04 21:11:23 <BlueMatt> nor of the fed's mandate and purpose
560 2012-07-04 21:15:49 <freewil> not sure why you say that
561 2012-07-04 21:16:08 <BlueMatt> clang stripped out the whole serialize mess where we go in and serialize CTxIns, CTxOuts, etc...and pulls it into a series of calls to serialize vector of unsigned chars
562 2012-07-04 21:16:12 <BlueMatt> now thats well done
563 2012-07-04 21:16:48 <BlueMatt> freewil: pointing out that op twist had that effect is interesting, saying that the fed made a mistake or that the fed should be blamed for that effect is stupid
564 2012-07-04 21:18:05 <freewil> it's a critique of the fed's ZIRP policy
565 2012-07-04 21:18:18 <gmaxwell> WELCOME TO #FED
566 2012-07-04 21:18:22 <gmaxwell> oh this isn't #FED
567 2012-07-04 21:18:23 <gmaxwell> ?
568 2012-07-04 21:18:38 <BlueMatt> oh, its more general about setting low interest rates, nope nevermind its just plain stupid then
569 2012-07-04 21:18:48 <freewil> ...
570 2012-07-04 21:19:00 <freewil> hard to debate with someone who doesnt make any real objections
571 2012-07-04 21:19:25 <BlueMatt> lets go to -otc, because that is where financial debates always (used) to happen
572 2012-07-04 21:19:58 <BlueMatt> freewil: #bitcoin-otc
573 2012-07-04 21:37:54 <jrmithdobbs> sipa: its looking like this is something esoteric in how bitcoin is handling socks
574 2012-07-04 21:38:24 <jrmithdobbs> sipa: the .2.3 branch (which will be stable soon-ish) has a lot stricter checking on all the socks stuff
575 2012-07-04 21:38:48 <galambo> is there a channel #fedcoin-dev ;)
576 2012-07-04 22:47:04 <jrmithdobbs> sipa: pretty sure this is bitcoin side, -proxy=127.0.0.1:1080 -socks=5 -connect=v4ip; gives error reading response errors when the socks proxy on 127.0.0.1:1080 is ssh -D
577 2012-07-04 22:47:27 <sipa> jrmithdobbs: could be
578 2012-07-04 22:48:31 <jrmithdobbs> i think their seperation and more thorough error checking to keep circuits/users/protocols separate fixed something that broke it
579 2012-07-04 23:06:22 <freewil> nice
580 2012-07-04 23:08:48 <jrmithdobbs> sipa: fixed it, it's a bug on both ends
581 2012-07-04 23:09:06 <BlueMatt> sipa: ultraprune, or...?
582 2012-07-04 23:09:06 <jrmithdobbs> sipa: we're sending an extra 0x00 at the end of the initial auth handshake
583 2012-07-04 23:09:13 <sipa> BlueMatt: yeah
584 2012-07-04 23:09:22 <jrmithdobbs> sipa: tor doesn't parse that extra 0 so doesn't throw an error on the first packet
585 2012-07-04 23:09:54 <jrmithdobbs> it's still in the receive buffer when it goes to read the next packet, sees that 0x00 at the front instead of 0x05 and throws a general failure in the wrong place
586 2012-07-04 23:10:07 <jrmithdobbs> it should be throwing the failure on the first packet
587 2012-07-04 23:10:11 <gmaxwell> TWO BUGS FOR ONE PRICE
588 2012-07-04 23:10:22 <jrmithdobbs> our fix is nice and easy
589 2012-07-04 23:11:01 <jrmithdobbs> http://pastebin.com/335T4iL2
590 2012-07-04 23:11:32 <jrmithdobbs> my bitcoin repo is clobbered if someone wants to put in a pull for that or fix the string prettier
591 2012-07-04 23:12:49 <sipa> jrmithdobbs: so only send 0x05 0x01 ??
592 2012-07-04 23:12:56 <sipa> instead of 0x05 0x01 0x00 ?
593 2012-07-04 23:13:25 <jrmithdobbs> sipa: no it needs to send 0x05 0x01 0x00, it's currently sending 0x05 0x01 0x00 0x00
594 2012-07-04 23:13:45 <jrmithdobbs> sipa: nSize includes the trailing \0 of the cstr the way it's set, i don't know how to set it right
595 2012-07-04 23:14:12 <sipa> jrmithdobbs: right!
596 2012-07-04 23:14:23 <jrmithdobbs> OBO
597 2012-07-04 23:14:25 <jrmithdobbs> good times
598 2012-07-04 23:14:30 <BlueMatt> is there a gavinandresen around?
599 2012-07-04 23:15:02 <sipa> or the initializer must be changed to {0x05,0x01,0x00} instead of "51\0"
600 2012-07-04 23:15:13 <sipa> jrmithdobbs: thanks for the bughunt!
601 2012-07-04 23:15:15 <jrmithdobbs> ya however you think is best
602 2012-07-04 23:15:33 <BlueMatt> in any case, there should be thought put into list of pulls for 0.7 and start the rc process IMHO
603 2012-07-04 23:15:33 <jrmithdobbs> sipa: was an interesting problem that just wasn't making sense, glad to do it ;p
604 2012-07-04 23:15:51 <BlueMatt> (as in do some insane number of pulls over a few days and then rc)
605 2012-07-04 23:16:47 <Ferroh> Is there a known issue with the satoshi client,
606 2012-07-04 23:16:55 <BlueMatt> yes, many ;)
607 2012-07-04 23:16:57 <Ferroh> where a valid address is reported as being invalid?
608 2012-07-04 23:17:13 <Ferroh> and then this is fixed in the current build?
609 2012-07-04 23:17:16 <sipa> no such issue is known
610 2012-07-04 23:17:20 <Ferroh> because I have a lot of clients reporting this
611 2012-07-04 23:17:25 <Ferroh> and I have to issue them a new address
612 2012-07-04 23:17:42 <Ferroh> they have this problem with blockchain.info as well, but blockchain.info might very well be using the satoshi client to validate addresses
613 2012-07-04 23:17:45 <gmaxwell> Ferroh: people were reporting that _blockchain.info_ was doing that.
614 2012-07-04 23:17:53 <gmaxwell> that seems unlikely.
615 2012-07-04 23:17:56 <Ferroh> well I have someone telling me that he was using satoshi
616 2012-07-04 23:18:09 <Ferroh> ("bitcoin-qt" is his exact description)
617 2012-07-04 23:18:17 <gmaxwell> Ferroh: get the version and we can try exactly that.
618 2012-07-04 23:18:39 <gmaxwell> Could be some crazy locale thing, find out (1) the address in question, and (2) his keyboard layout / language.
619 2012-07-04 23:19:20 <Ferroh> what is the easiest way for him to get the version number?
620 2012-07-04 23:19:28 <BlueMatt> help -> about
621 2012-07-04 23:19:30 <Ferroh> (remember I am dealing with a client so i want to bug him minimally)
622 2012-07-04 23:19:33 <BlueMatt> eh, about bitcoin-qt
623 2012-07-04 23:19:53 <sipa> Help -> About Bitcoin-Qt
624 2012-07-04 23:20:05 <Ferroh> sadly i dont see this on ubuntu
625 2012-07-04 23:20:09 <Ferroh> ok I will try asking him
626 2012-07-04 23:20:24 <sipa> you don't see this on ubuntu?
627 2012-07-04 23:20:31 <sipa> it should be there...
628 2012-07-04 23:21:02 <Ferroh> there is no help menu that I can see
629 2012-07-04 23:21:15 <Ferroh> i dont know ubuntu well, especially this fancy new ui crap
630 2012-07-04 23:21:40 <Ferroh> but no, i see no such option
631 2012-07-04 23:21:46 <Ferroh> version 0.6.2
632 2012-07-04 23:21:57 <BlueMatt> oh, thats an unrelated bug
633 2012-07-04 23:22:15 <BlueMatt> (and unsolved)
634 2012-07-04 23:22:31 <Ferroh> ok thats not a big deal anyway
635 2012-07-04 23:22:48 <jrmithdobbs> sipa: that change also fixes compat with ssh's -D socks proxy in v5 mode
636 2012-07-04 23:22:49 <Ferroh> the client is unresponsive, so I'm going to stop pestering him about technical stuff
637 2012-07-04 23:22:59 <Ferroh> (by the "client" i mean the person, not the software)
638 2012-07-04 23:23:01 <jrmithdobbs> sipa: looks like from initial testing, anyways
639 2012-07-04 23:23:11 <jrmithdobbs> (ssh throws the error after the first packet correctly, why i figured it out actually)
640 2012-07-04 23:23:23 <sipa> ic
641 2012-07-04 23:23:46 <sipa> jrmithdobbs: i'll fix it; poke me or file an issue if i forget
642 2012-07-04 23:27:47 <jrmithdobbs> sipa: all good, my node's working for now with that patch so whenever you have time
643 2012-07-04 23:31:26 <BlueMatt> no one in the dev team cares about 0.7 planning
644 2012-07-04 23:31:31 <BlueMatt> like sipa?
645 2012-07-04 23:35:47 <jrmithdobbs> BlueMatt: it's a holiday in the us, noone's paying attention i'm sure
646 2012-07-04 23:36:24 <BlueMatt> heh, how did I forget that
647 2012-07-04 23:36:35 <BlueMatt> Ill wait till monday