1 2012-02-10 00:00:31 <phantomcircuit> BlueMatt, i suspect they're fairly legit, but have no idea what the fuck they're doing
  2 2012-02-10 00:00:45 <BlueMatt> i wouldnt say that, I have no idea
  3 2012-02-10 00:00:56 <BlueMatt> they look legit, nfc if they know what they are doing
  4 2012-02-10 00:01:00 <BlueMatt> but they need a new designer
  5 2012-02-10 00:01:08 <BlueMatt> (afaict they even had a booth at ces)
  6 2012-02-10 00:01:25 <phantomcircuit> i looked into them a bit
  7 2012-02-10 00:01:34 <phantomcircuit> everything about them screams legit
  8 2012-02-10 00:01:44 <BlueMatt> nice
  9 2012-02-10 00:01:55 <BlueMatt> what gave you the impression they didnt know what they were doing?
 10 2012-02-10 00:02:03 <phantomcircuit> they are however doing something that i believe is regulated
 11 2012-02-10 00:02:09 <phantomcircuit> but quite possibly isn't
 12 2012-02-10 00:02:28 <BlueMatt> ???
 13 2012-02-10 00:02:46 <phantomcircuit> BlueMatt, i dont see any indication of whether they have figured out the regulatory compliant stuff for accepting payments on behalf of someone else in the us
 14 2012-02-10 00:02:47 <BlueMatt> so they are legit, but not following regulations, or so you think?
 15 2012-02-10 00:02:54 <BlueMatt> mmm
 16 2012-02-10 00:03:03 <BlueMatt> yea, I was looking for that too
 17 2012-02-10 00:03:17 <phantomcircuit> which is actually not that surprising since the rules are hilariously convoluted
 18 2012-02-10 00:03:45 <BlueMatt> the "No hassle of PCI Compliance" quote on their site looks sketch considering they dont bother to push their own compliance
 19 2012-02-10 00:03:51 <phantomcircuit> afaikt bitcoins do not qualify as a security commodity or currency and as such nobody dealing in them qualifies as a financial institution
 20 2012-02-10 00:04:10 <phantomcircuit> meh PCI compliance is a nuisance and 100% worthless
 21 2012-02-10 00:04:12 <BlueMatt> so in other words, just have to wait for them to get sued
 22 2012-02-10 00:04:25 <BlueMatt> yes it is, but its one of those compliance things that you have to do
 23 2012-02-10 00:04:35 <phantomcircuit> unfortunately yeah i suspect the only way to figure out for sure is for someone to get sued
 24 2012-02-10 00:05:00 <BlueMatt> even then it wont be for sure until it hits federal appellate court level
 25 2012-02-10 00:05:04 <phantomcircuit> PCI compliance is a private agreement between you cand visa or mastercard
 26 2012-02-10 00:05:06 <phantomcircuit> it is not the law
 27 2012-02-10 00:05:12 <BlueMatt> (unless the supreme court picks up a bitcoin case, which I doubt)
 28 2012-02-10 00:05:20 <midnightmagic> It is not worthless, because it is required.
 29 2012-02-10 00:05:22 <phantomcircuit> s/cand/and/
 30 2012-02-10 00:05:44 <BlueMatt> yea, its not the law, but anyone who wants to be a payment processor better damn well be accepting ccs in addition to bitcoin
 31 2012-02-10 00:05:47 <BlueMatt> and that means pci
 32 2012-02-10 00:05:48 <phantomcircuit> i seriously doubt even PCI DSS compliance has ever prevented any theft
 33 2012-02-10 00:06:10 <BlueMatt> Im sure it has, but usually it doesnt
 34 2012-02-10 00:06:19 <midnightmagic> Citation needed.
 35 2012-02-10 00:06:27 <phantomcircuit> ok maybe that was hyperbole
 36 2012-02-10 00:06:30 <BlueMatt> (because companies that are so up tight about pci to give a shit about it, dont have any actual security just do pci)
 37 2012-02-10 00:06:30 <phantomcircuit> but still
 38 2012-02-10 00:07:53 <Idiot___> pci?
 39 2012-02-10 00:09:05 <phantomcircuit> Idiot___, google
 40 2012-02-10 00:09:26 <Idiot___> google?
 41 2012-02-10 00:09:40 <phantomcircuit> BlueMatt, can we just ban him
 42 2012-02-10 00:09:43 <phantomcircuit> this is ridiculous
 43 2012-02-10 00:10:12 <Idiot___> is it cos iz black?
 44 2012-02-10 00:10:30 <Joric> yes
 45 2012-02-10 00:10:35 <Idiot___> :(
 46 2012-02-10 00:10:38 <phantomcircuit> i knew someone would say yes
 47 2012-02-10 00:12:39 <cjd> hasn't everyone switched to pci-express by now?
 48 2012-02-10 00:12:48 <Idiot___> thats what i thought cjd
 49 2012-02-10 00:13:02 <Idiot___> or the even better one agp
 50 2012-02-10 00:15:03 <phantomcircuit> cjd, dont feed the trolls
 51 2012-02-10 00:15:35 <Joric> Idiot___, payment card industry
 52 2012-02-10 00:15:59 <Idiot___> Joric, no thank you
 53 2012-02-10 00:28:34 <luke-jr> BlueMatt: and I'm guessing you'll need to rescan if you received anything on a compressed pubkey with the old client?
 54 2012-02-10 00:29:37 <sipa> luke-jr: won't work
 55 2012-02-10 00:30:01 <sipa> you just have two wallets really... old clients will use the non-compressed keys, new clients will use the compressed ones
 56 2012-02-10 00:30:21 <sipa> with a shared history
 57 2012-02-10 00:30:58 <helo> how are keys compressed? i thought they were essentially random
 58 2012-02-10 00:31:18 <BlueMatt> that way old clients will just die
 59 2012-02-10 00:32:03 <sipa> helo: public keys are points on an EC curve, given by X and Y coordinates
 60 2012-02-10 00:32:19 <sipa> helo: they can be compressed by only storing the X coordinate, and calculating Y from the curve equation
 61 2012-02-10 00:32:26 <cjd> s/die/get sybiled/
 62 2012-02-10 00:32:35 <helo> hmmmm *wikis*
 63 2012-02-10 00:32:40 <sipa> cjd: no, die
 64 2012-02-10 00:34:11 <Idiot___> oih
 65 2012-02-10 00:34:18 <Idiot___> is the private key like the Z coordinate?
 66 2012-02-10 00:34:29 <sipa> no
 67 2012-02-10 00:34:55 <sipa> http://en.wikipedia.org/wiki/Elliptic_curve_cryptography
 68 2012-02-10 00:36:41 <sipa> Idiot___: in very short: EC defines a "multiplication" between numbers and points, and there is one special point determined in advance, G; a private key is a number x, and x*G is the public key
 69 2012-02-10 00:36:53 <sipa> and there is no fast way to compute the division of two points
 70 2012-02-10 00:44:08 <Idiot___> oh thx sipa
 71 2012-02-10 00:44:31 <Idiot___> how do you get enough points on the curve to give you 2^256 keys tho?
 72 2012-02-10 00:44:52 <sipa> because there are 2^256 points on the curve
 73 2012-02-10 00:44:57 <phantomcircuit> you use a big fucking number for the x/y
 74 2012-02-10 00:52:45 <helo> there are some cool youtube videos about ecc
 75 2012-02-10 00:53:32 <Idiot___> i donno it all seems above my skill level
 76 2012-02-10 00:53:38 <Idiot___> i find it difficult multiplying normal numbers
 77 2012-02-10 00:53:43 <Diablo-D3> cracking ecc: slightly harder than landing a man on the moon, but only slightly
 78 2012-02-10 00:54:35 <BlueMatt> so its only slightly not doable with years-old technology
 79 2012-02-10 00:54:59 <sipa> i doubt spending all money and man-hours on ECC cracking that were spent on putting a man on the moon would suffice :)
 80 2012-02-10 00:56:23 <riush> b
 81 2012-02-10 00:56:30 <BlueMatt> c
 82 2012-02-10 00:56:35 <riush> oops :)
 83 2012-02-10 00:57:40 <sipa> boops?
 84 2012-02-10 00:57:50 <BlueMatt> coops
 85 2012-02-10 00:58:18 <Idiot___> why are eliptic curves so hard to go from private key to public key?
 86 2012-02-10 00:58:26 <Idiot___> cant you just use a bit of sketching paper
 87 2012-02-10 00:58:38 <Idiot___> and flip it over, then figure it out that way
 88 2012-02-10 00:58:51 <BlueMatt> ...
 89 2012-02-10 00:58:58 <BlueMatt> ok, thats really getting old
 90 2012-02-10 00:59:20 <sipa> Idiot___: they're not the kind of curves you can draw on paper
 91 2012-02-10 00:59:25 <sipa> it's all just abstract numbers
 92 2012-02-10 00:59:44 <sipa> and there is no division operator, so you can't go back to the number G was multiplied with
 93 2012-02-10 00:59:59 <Idiot___> oh
 94 2012-02-10 01:00:03 <Diablo-D3> says who
 95 2012-02-10 01:00:04 <Idiot___> hey rabbito
 96 2012-02-10 01:00:29 <Diablo-D3> it slices, it dices, it had sex with your mother last night
 97 2012-02-10 01:01:33 <sipa> Diablo-D3: please
 98 2012-02-10 01:01:53 <Diablo-D3> dude, Im taking a much needed break
 99 2012-02-10 01:02:08 <Diablo-D3> I have enlisted perl to help me finish writing this new kernel faster
100 2012-02-10 01:08:52 <NxTitle> hmm... I should really change my ident from root so SuprTiggr stops barking at me
101 2012-02-10 01:10:50 <nanotube> you should bark back. :)
102 2012-02-10 01:13:34 <BlueMatt> damn, I only just barely used 2g of data last month, I need to step it up
103 2012-02-10 01:13:41 <BlueMatt> moar tethering!!!!
104 2012-02-10 01:13:52 <sipa> 2g mobile data?
105 2012-02-10 01:14:03 <BlueMatt> 3g+4g
106 2012-02-10 01:14:16 <sipa> 2 GiB i mean
107 2012-02-10 01:14:22 <BlueMatt> oh, yea
108 2012-02-10 01:14:36 <BlueMatt> In past months Ive used between 4 and 6 :)
109 2012-02-10 01:14:45 <sipa> how do you do that? i barely get to 500 MiB (and i'm allowed to use 2 GiB)
110 2012-02-10 01:14:55 <BlueMatt> sprint unlimited 4g :)
111 2012-02-10 01:15:10 <BlueMatt> nfc where all the data comes from, but I dont care
112 2012-02-10 01:15:31 <cjd> maybe tv channels which now gone?
113 2012-02-10 01:15:38 <BlueMatt> ?
114 2012-02-10 01:16:07 <cjd> parts of the uhf band which they gutted when they switched to dtv
115 2012-02-10 01:16:21 <BlueMatt> how would that effect my bw usage?
116 2012-02-10 01:16:38 <cjd> they would have lots of available/cheap b/w
117 2012-02-10 01:16:43 <BlueMatt> also, is anyone actually using those bands yet?
118 2012-02-10 01:17:08 <cjd> auction is done do I kind of assumed so
119 2012-02-10 01:17:10 <BlueMatt> isnt that the one where google got the fcc to require them to be used for "open" things?
120 2012-02-10 01:17:27 <cjd> Big G lost that battle IIRC
121 2012-02-10 01:17:49 <BlueMatt> iirc they never wanted to win (or so most people think) they just wanted restrictions on whoever won
122 2012-02-10 01:18:01 <BlueMatt> (didnt they bet numerical constants?)
123 2012-02-10 01:18:08 <BlueMatt> and iirc they got them
124 2012-02-10 01:18:26 <cjd> sounds like you know better than I
125 2012-02-10 01:18:45 <BlueMatt> meh, I dont remember that well...
126 2012-02-10 01:19:08 <cjd> run bitcoin over 4g?
127 2012-02-10 01:19:20 <cjd> oh do they let you open ports?
128 2012-02-10 01:19:23 <BlueMatt> heh, that would do it, but I never do
129 2012-02-10 01:19:26 <BlueMatt> I dont think so
130 2012-02-10 01:19:30 <cjd> ahh
131 2012-02-10 01:19:31 <gmaxwell> cjd: it could run over evdo.
132 2012-02-10 01:19:41 <gmaxwell> (and certantly 4g)
133 2012-02-10 01:19:56 <BlueMatt> bitcoin doesnt eat much bw, but if you let it sit there 24/7 it will sip quite a bit
134 2012-02-10 01:19:57 <cjd> so you really need a vpn or cheapo vps
135 2012-02-10 01:20:13 <spawn-> who needed vpn?
136 2012-02-10 01:20:22 <spawn-> im offer vps/vpn/ircbouncer
137 2012-02-10 01:20:33 <gmaxwell> BlueMatt: uses hardly any if you're not listening. The maximum datarate of the blockchain itself is just 14kbit/sec over the long term.
138 2012-02-10 01:21:35 <BlueMatt> gmaxwell: +addr + a bunch of crap... it adds up
139 2012-02-10 01:23:32 <sipa> you'll probably see each transaction twice... once in the blockchain and once on its own
140 2012-02-10 01:26:14 <Kez_Maefele> negative balance..  - https://bitcointalk.org/index.php?topic=63308.msg740670#msg740670
141 2012-02-10 01:26:59 <BlueMatt> its been seen
142 2012-02-10 01:27:42 <cjd> what's funny is setting open file limit to like 1024 and making dpi boxes cry
143 2012-02-10 01:28:46 <BlueMatt> Kez_Maefele: bbe missed the duplicate coinbase
144 2012-02-10 01:29:36 <sipa> but it is an actual double spend
145 2012-02-10 01:30:16 <sipa> ah, maybe there was a duplicate coinbase in between that is not reported by BBE
146 2012-02-10 01:30:21 <sipa> that's possible
147 2012-02-10 01:30:36 <BlueMatt> isnt that what someone said earlier?
148 2012-02-10 01:30:44 <BlueMatt> (I didnt check, just thought someone said that)
149 2012-02-10 01:30:45 <sipa> oh, must have missed that
150 2012-02-10 01:30:58 <BlueMatt> I thought you said that...
151 2012-02-10 01:31:40 <sipa> i said something about duplicate coinbases, but not what i meant afterwards
152 2012-02-10 01:32:08 <Kez_Maefele> is it specific for BBE only, or to bitcoin client also ?
153 2012-02-10 01:32:19 <BlueMatt> well apparently its something else
154 2012-02-10 01:33:51 <sipa> http://blockexplorer.com/testnet/block/0000000013aa9f67da178005f9ced61c7064dd6e8464b35f6a8ca8fabc1ca2cf
155 2012-02-10 01:33:56 <sipa> bingo
156 2012-02-10 01:34:21 <sipa> it's not true that only one can be redeemed
157 2012-02-10 01:34:39 <sipa> coinbase-spend-coinbase-spend is possible
158 2012-02-10 01:34:48 <sipa> coinbase-coinbase-spend-spend isn't
159 2012-02-10 01:59:25 <gmaxwell> Houston we have a (hopefully minor) problem
160 2012-02-10 02:00:14 <sipa> gmaxwell: how so?
161 2012-02-10 02:00:18 <BlueMatt> continue...
162 2012-02-10 02:00:31 <gmaxwell> I was pointing forrestv to the Feb15 version message change and he pointed out that he had uncovered some NATs noticing the IP addresses in header fields and changing them.
163 2012-02-10 02:01:03 <BlueMatt> nats do dpi to change ips?
164 2012-02-10 02:01:04 <BlueMatt> ???
165 2012-02-10 02:01:28 <gmaxwell> BlueMatt: duh. But they should sanely only do it to protocols they understand!
166 2012-02-10 02:01:58 <sipa> which means invalid checksums?
167 2012-02-10 02:02:07 <BlueMatt> I wasnt aware of that, and it is pretty damn dumb
168 2012-02-10 02:02:07 <gmaxwell> He only observed two p2pool users with this behavior, but obviously they gave up on using it quickly.
169 2012-02-10 02:02:36 <k9quaint> which NAT Is doing it?
170 2012-02-10 02:02:46 <k9quaint> I can't picture iptables doing that :|
171 2012-02-10 02:03:25 <sipa> in iptables you need to load specific modules for various protocols, afaik
172 2012-02-10 02:04:37 <BlueMatt> so...what use upnp to get the ip to encode in version msgs?
173 2012-02-10 02:04:37 <gmaxwell> I'm having a hard timing coming up with useful search terms for this.
174 2012-02-10 02:05:03 <k9quaint> gmaxwell: try searching for justin beiber
175 2012-02-10 02:05:04 <gmaxwell> BlueMatt: ah, that would improve survival yes.
176 2012-02-10 02:05:22 <BlueMatt> (also means we have a target for 0.6...) feb 15
177 2012-02-10 02:05:30 <BlueMatt> which is probably much, much too fast
178 2012-02-10 02:05:38 <sipa> huh?
179 2012-02-10 02:05:43 <k9quaint> make a distributed VPN
180 2012-02-10 02:06:01 <k9quaint> then put a hyphen in your name and make youtube vids
181 2012-02-10 02:06:01 <sipa> 0.6 won't be released before bip16 makes 55% support
182 2012-02-10 02:06:03 <BlueMatt> sipa: gotta have a version out for people who's connections break on the 15th
183 2012-02-10 02:06:06 <gmaxwell> well because     CAddress addrMe = (fUseProxy ? CAddress(CService("0.0.0.0",0)) : addrLocalHost);
184 2012-02-10 02:06:25 <gmaxwell> I assume that addrLocalHost gets filled out by UPNP too right?
185 2012-02-10 02:06:28 <sipa> BlueMatt: oh, right
186 2012-02-10 02:06:30 <gmaxwell> so those hosts might already be okay.
187 2012-02-10 02:06:34 <BlueMatt> so...0.5.3?
188 2012-02-10 02:06:38 <sipa> BlueMatt: possibly a 0.5.3 then
189 2012-02-10 02:06:41 <BlueMatt> gmaxwell: no, it currently does not
190 2012-02-10 02:06:55 <BlueMatt> only irc and static remote server
191 2012-02-10 02:07:03 <gmaxwell> IIRC the address it gets from irc/httpcrap yea...
192 2012-02-10 02:07:09 <doublec> what is the "feb 15 version message change"?
193 2012-02-10 02:07:10 <gmaxwell> Okay, so that may make it less bad.
194 2012-02-10 02:07:15 <BlueMatt> also, no way to know if "external" ip from upnp is really external
195 2012-02-10 02:07:25 <gmaxwell> BlueMatt: no, but we can't solve everything.
196 2012-02-10 02:08:14 <BlueMatt> doublec: on feb 15th version messages (and other iirc) get checksums, where they previously didnt have any
197 2012-02-10 02:08:27 <BlueMatt> s/other/a few others/
198 2012-02-10 02:08:38 <sipa> do we have an estimate how often this happens?
199 2012-02-10 02:08:47 <sipa> i have rarely see wrong checksums on messages
200 2012-02-10 02:08:50 <sipa> seen
201 2012-02-10 02:09:05 <sipa> oh, right, version is the only one with your own address
202 2012-02-10 02:09:08 <gmaxwell> Yep.
203 2012-02-10 02:09:10 <sipa> afaik even addr never includes it
204 2012-02-10 02:09:34 <gmaxwell> well addr will when you announce yourself but I wouldn't be surprised if they didn't look that deep into the stream.
205 2012-02-10 02:09:38 <BlueMatt> that just seems...wrong for a nat to be blindly replacing 4 random bytes that happen to be your internal ip with the external one...
206 2012-02-10 02:09:45 <gmaxwell> Otherwise they'd corrupt random things like jpegs. :)
207 2012-02-10 02:09:48 <sipa> can we just make version ignore the checksum?
208 2012-02-10 02:10:04 <gmaxwell> We could, but there are lots and lots of nodes that won't.
209 2012-02-10 02:10:06 <BlueMatt> wed have to get huge network upgrade in a matter of days
210 2012-02-10 02:10:14 <gmaxwell> Better to get change the behavior of the host inside the bad nat.
211 2012-02-10 02:10:16 <BlueMatt> (I really wish we could do that...)
212 2012-02-10 02:11:03 <gmaxwell> s/get //
213 2012-02-10 02:12:35 <BlueMatt> so...best thing I see to do is: add addrExternalFromUPnP and send that in version if it seems reasonable
214 2012-02-10 02:12:54 <sipa> ack
215 2012-02-10 02:13:02 <BlueMatt> and note in the release notes that if you are having problems connecting as of feb 15, make sure to turn on upnp
216 2012-02-10 02:13:19 <sipa> but a few stable network nodes that ignore wrong checksums in version might help a bit too
217 2012-02-10 02:13:28 <BlueMatt> yep
218 2012-02-10 02:13:28 <gmaxwell> Yes, at least stuff people can addnode.
219 2012-02-10 02:13:49 <gmaxwell> not just ignore checksums but also accept the old format. The connecting party inititates, so it could either-or.
220 2012-02-10 02:15:30 <sipa> it's not *that* trivial to do so, i think
221 2012-02-10 02:20:43 <sipa> you'd need to try processing the message twice
222 2012-02-10 02:20:52 <sipa> once with and once without checksum
223 2012-02-10 02:21:01 <sipa> ignoring the checksum for version is trivial though
224 2012-02-10 02:38:00 <gribble> New news from bitcoinrss: sipa opened pull request 816 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/816>
225 2012-02-10 02:40:33 <BlueMatt> why do we not have a cs_addrLocalHost its read and written from like 3 threads
226 2012-02-10 02:54:57 <XMPPwocky> sipa: ooh ooh
227 2012-02-10 02:55:34 <XMPPwocky> i can write a tamper_proxy mangler for that
228 2012-02-10 02:56:36 <XMPPwocky> at he very least, relaying an alert
229 2012-02-10 03:02:58 <gmaxwell> I'm trying not to be mean but I am tempted to respond to grarpamp with something like "was your message truncated? it appears to be missing the info as promised in your subject line"
230 2012-02-10 03:41:40 <BlueMatt> gmaxwell: we need to get him in irc
231 2012-02-10 03:42:32 <sipa> hmm his message is clear to me, though a bit confusing
232 2012-02-10 03:43:25 <gmaxwell> sipa: if only he actually included info on how much was saved!
233 2012-02-10 03:46:08 <gmaxwell> reduced my wallet size by 3.3%
234 2012-02-10 03:47:15 <BlueMatt> what about blkindex?
235 2012-02-10 03:47:26 <gmaxwell> trying now..
236 2012-02-10 03:48:02 <BlueMatt> anyone feel like looking into why addrLocalHost has no cs'
237 2012-02-10 03:48:11 <gmaxwell> BlueMatt: is it just an integer?
238 2012-02-10 03:48:48 <gmaxwell> so.. hm. on a blkindex.dat from a fairly recently bootstrapped node.. 366628864 -> 269881344
239 2012-02-10 03:48:56 <BlueMatt> no its a CAddress
240 2012-02-10 03:49:13 <sipa> ... i suppose it used to be just an integer
241 2012-02-10 03:49:16 <sipa> before netbase
242 2012-02-10 03:49:30 <BlueMatt> I actually dont think so, but its possible
243 2012-02-10 03:49:42 <gmaxwell> so.. 26% reduction.
244 2012-02-10 03:49:54 <gmaxwell> and I bet that will also make processing faster.
245 2012-02-10 03:50:13 <BlueMatt> (also, Im looking at 0.5.2 code, so it doesnt have netbase
246 2012-02-10 03:50:23 <sipa> indeed, it was a CAddress all along
247 2012-02-10 03:50:31 <sipa> now, it's all fixed memory
248 2012-02-10 03:50:42 <sipa> worst case is you read a partially updated value
249 2012-02-10 03:51:01 <BlueMatt> so, the next question is, can we reload db from libdb eg once we catch up?
250 2012-02-10 03:51:29 <gmaxwell> some addr.dat 9015296 -> 6389760
251 2012-02-10 03:51:41 <gmaxwell> BlueMatt: I'm guessing the compact call will do about the same thing.
252 2012-02-10 03:52:31 <gmaxwell> of course... the crappy behavior that leaves all that garbage in it contributes to our poor performance.
253 2012-02-10 03:54:23 <BlueMatt> yep...
254 2012-02-10 03:54:52 <sipa> we have CDB::Rewrite
255 2012-02-10 03:55:22 <BlueMatt> dont know if we want to use that for blkindex though...
256 2012-02-10 03:55:47 <gmaxwell> well thats what the dump/load did. Took a bit though.
257 2012-02-10 03:58:02 <BlueMatt> I suppose we can do it after a close if its been a while
258 2012-02-10 03:58:21 <gmaxwell> we don't want to make close any slower.
259 2012-02-10 03:58:25 <BlueMatt> check free memory first though
260 2012-02-10 03:58:36 <BlueMatt> so, what, do it after chain finishes loading?
261 2012-02-10 03:59:02 <sipa> i think you want this at most as an optional action
262 2012-02-10 03:59:08 <sipa> like "compact databases"
263 2012-02-10 03:59:14 <gmaxwell> Well, the rewrite .. will need a bunch of free space.
264 2012-02-10 03:59:34 <BlueMatt> sipa: well at that point you can just tell people to fire up db_dump/load
265 2012-02-10 04:01:32 <sipa> BlueMatt: for bitcoind users, yes; for bitcoin-qt users, no
266 2012-02-10 04:02:49 <BlueMatt> well, yea, but my point is if it makes a huge deal, we should probably try to do it at some opportune time
267 2012-02-10 04:03:00 <BlueMatt> s/deal/perf increase/
268 2012-02-10 04:03:09 <BlueMatt> or a decent one
269 2012-02-10 04:03:27 <sipa> i doubt it will improve performance much?
270 2012-02-10 04:04:08 <gmaxwell> no clue.
271 2012-02-10 04:04:27 <BlueMatt> maybe not then
272 2012-02-10 04:04:28 <BlueMatt> <gmaxwell> of course... the crappy behavior that leaves all that garbage in it contributes to our poor performance.
273 2012-02-10 04:05:48 <BlueMatt> anyone want to sanity-check https://github.com/TheBlueMatt/bitcoin/commit/fe599861c6df9ac948228d4fd8e54ce24ea64727
274 2012-02-10 04:06:42 <gmaxwell> I'd expect ~some better performance, just by making better cache usage if nothing else. Doubt it would be earth shaking.
275 2012-02-10 04:07:27 <sipa> BlueMatt: any reason not to put that value in addrLocalHost itself, if it makes sense?
276 2012-02-10 04:07:45 <sipa> ... which ought to be a vector
277 2012-02-10 04:07:49 <gmaxwell> BlueMatt: is the byte order right?
278 2012-02-10 04:08:06 <BlueMatt> sipa: I prefer not to because of horror stories about 6-layer nats in odd corners of the globe
279 2012-02-10 04:08:25 <sipa> how is that relevant?
280 2012-02-10 04:08:32 <BlueMatt> sipa: and I prefer to have the externalish ip separate from addrLocalHost for the version
281 2012-02-10 04:08:34 <gmaxwell> BlueMatt: meh. run isroutable on the upnp returned address.
282 2012-02-10 04:08:44 <BlueMatt> gmaxwell: it pulls as a string, there is no byte order
283 2012-02-10 04:08:51 <BlueMatt> (ish)
284 2012-02-10 04:09:01 <gmaxwell> besides except for this version message issue, if you are behind multilayer nat your IP isn't going to matter anyways.
285 2012-02-10 04:09:09 <gmaxwell> (because inbound won't work)
286 2012-02-10 04:09:21 <sipa> i think the real solution (especially with ipv6 coming up), is keep a vector of localhosts
287 2012-02-10 04:09:34 <sipa> and maybe keep a counter on them based on what others tell you in addrMe
288 2012-02-10 04:09:39 <gmaxwell> sipa: we also get told our address by inbound peers.
289 2012-02-10 04:09:43 <gmaxwell> ah yes.
290 2012-02-10 04:10:17 <BlueMatt> I dont like putting upnp-derived external ip in addrLocalHost, I dont know why but I just never trust upnp to do anything right
291 2012-02-10 04:10:33 <sipa> no reason to trust it less than whatismyip.com
292 2012-02-10 04:10:34 <BlueMatt> there are probably a few commonish routers that put garbage in the external ip field
293 2012-02-10 04:10:38 <sipa> well, maybe there is
294 2012-02-10 04:10:42 <sipa> but that doesn't matter
295 2012-02-10 04:10:57 <sipa> if you do sanity checks, and not let it destroy other potention local addresses
296 2012-02-10 04:11:20 <gmaxwell> BlueMatt: meh, okay, only update it if the current localaddress isn't routable.
297 2012-02-10 04:12:06 <gmaxwell> whatever the heck you get is better than some non-routable value. :)  (unless it's also non-routable, then its no worse)
298 2012-02-10 04:12:13 <BlueMatt> ok, Im too stoned to be arguing anything technical, Ill just go change it
299 2012-02-10 04:15:27 <sipa> BlueMatt: you probably want to ignore the upnp function call's result if it returns failure
300 2012-02-10 04:16:39 <BlueMatt> heh, thats what I get for copying miniupnpc's sample code...
301 2012-02-10 04:17:14 <BlueMatt> ok https://github.com/TheBlueMatt/bitcoin/commit/32b4d9c935fd5b3ef03593f2c023b6217d69e2aa
302 2012-02-10 04:17:21 <sipa> hmm, maybe that implies nothing will ever be in the result vector if it returns failure
303 2012-02-10 04:17:37 <BlueMatt> I would assume so, anyway doesnt matter
304 2012-02-10 04:18:28 <sipa> heh does that compile? :o
305 2012-02-10 04:18:47 <BlueMatt> oops, did I miss a bracket or smth?
306 2012-02-10 04:19:19 <sipa> no, CAddress has no such constructor
307 2012-02-10 04:19:40 <sipa> you probably want it to be a CNetAddr, and use addrLocalHost.SetIP()
308 2012-02-10 04:19:44 <BlueMatt> compiles for me as-is, are you merging it onto master?
309 2012-02-10 04:20:03 <BlueMatt> (im on pre-netbase)
310 2012-02-10 04:20:10 <sipa> oh
311 2012-02-10 04:20:30 <sipa> nvm then
312 2012-02-10 04:20:49 <BlueMatt> Ill port it to master in a sec, just wanted one for 0.5.X first
313 2012-02-10 04:21:01 <BlueMatt> luke-jr: ping
314 2012-02-10 04:21:31 <luke-jr> ?
315 2012-02-10 04:22:25 <BlueMatt> luke-jr: just wanted to ping you.  pull-request for 0.5.X: https://github.com/TheBlueMatt/bitcoin/commit/32b4d9c935fd5b3ef03593f2c023b6217d69e2aa
316 2012-02-10 04:22:38 <BlueMatt> (Im about to do a 0.6 version)
317 2012-02-10 04:22:57 <luke-jr> BlueMatt: all stable commits need to be merged to master first
318 2012-02-10 04:23:07 <luke-jr> in some form or another
319 2012-02-10 04:23:21 <sipa> that implies branching master
320 2012-02-10 04:23:40 <sipa> i have no problem with that, but i've heard other object to it before (jgarzik in particular)
321 2012-02-10 04:23:51 <luke-jr> sipa: ?
322 2012-02-10 04:24:04 <BlueMatt> sipa: we branched master a long time ago
323 2012-02-10 04:24:32 <luke-jr> how is it different from any other pullreq?
324 2012-02-10 04:24:36 <sipa> what am i thinking
325 2012-02-10 04:24:50 <sipa> ok fine, just wait till this is in master
326 2012-02-10 04:24:53 <luke-jr> O.o
327 2012-02-10 04:25:49 <sipa> no, somehow i thought we couldn't merge this in the 0.6 line, but as it's a bugfix and quite trivial... i guess it should be anyway
328 2012-02-10 04:26:42 <BlueMatt> luke-jr: I wanted to ping you as I wrote a 0.5.X version instead of you having to backport
329 2012-02-10 04:26:58 <BlueMatt> luke-jr: (as this is something that a build should exist of before feb 15)
330 2012-02-10 04:27:14 <BlueMatt> https://github.com/bitcoin/bitcoin/pull/817
331 2012-02-10 04:27:15 <luke-jr> BlueMatt: thanks, if you can link it in the final pullreq I appreciate it
332 2012-02-10 04:27:23 <BlueMatt> already done :)
333 2012-02-10 04:27:51 <luke-jr> O.o
334 2012-02-10 04:27:55 <luke-jr> NATs are editing packets?
335 2012-02-10 04:28:00 <BlueMatt> apparently
336 2012-02-10 04:28:01 <sipa> so it would seem
337 2012-02-10 04:28:28 <luke-jr> BlueMatt: maybe replace with 0.0.0.0 if no UPnP then?
338 2012-02-10 04:28:33 <luke-jr> assuming it's non-routable anyway
339 2012-02-10 04:29:11 <BlueMatt> if theres no upnp, chances are we dont have to worry about crazy nats
340 2012-02-10 04:29:26 <BlueMatt> (at least not much)
341 2012-02-10 04:29:38 <luke-jr> in any case, I agree this is a real concern. I imagine sipa can pull it into master ASAP?
342 2012-02-10 04:30:01 <BlueMatt> it can wait a day or two
343 2012-02-10 04:30:04 <gribble> New news from bitcoinrss: TheBlueMatt opened pull request 817 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/817>
344 2012-02-10 04:30:10 <luke-jr> I have much bigger problems to deal with right now unfortunately
345 2012-02-10 04:30:17 <BlueMatt> but, yea as stated, it would be nice to have builds of this available in 0.5.X before feb 15
346 2012-02-10 04:30:44 <luke-jr> my wife is in the hospital, and this DDoS jerk just took out my email
347 2012-02-10 04:30:57 <BlueMatt> damn...
348 2012-02-10 04:31:35 <luke-jr> and phones too, so the hospital can't call me back -.-
349 2012-02-10 04:31:53 <BlueMatt> ddosing your phones???
350 2012-02-10 04:32:01 <gmaxwell> VoIP
351 2012-02-10 04:32:02 <luke-jr> phones are on the same server as email
352 2012-02-10 04:32:07 <BlueMatt> ouch...
353 2012-02-10 04:32:37 <BlueMatt> can I ask who runs their home phones on VoIP? anyway, go deal with that, this can wait a few days
354 2012-02-10 04:33:04 <gmaxwell> BlueMatt: I have  every time for the last decade that I've bothered having a home phone.
355 2012-02-10 04:33:35 <olp> asterisk?
356 2012-02-10 04:33:41 <tcatm> Has this NAT problem been verified? I.e. is it known which routers cause problems?
357 2012-02-10 04:33:56 <BlueMatt> routers - no, but it has caused issues with p2pool
358 2012-02-10 04:34:23 <BlueMatt> gmaxwell: hmm, for some reason I didnt know it was that easy to deal with
359 2012-02-10 04:34:41 <sipa> even cisco does it: https://supportforums.cisco.com/docs/DOC-5229
360 2012-02-10 04:35:26 <gmaxwell> tcatm: it was observed on two differnet p2pool users, which uses essentially the same protocol, but with the checksum.
361 2012-02-10 04:35:35 <gmaxwell> (and, I've seen it elsewhere in the past)
362 2012-02-10 04:36:37 <BlueMatt> sipa: thats ridiculous
363 2012-02-10 04:36:55 <gmaxwell> tcatm: but as I think I said on the pull, I don't have a good feel for how bad a problem it is.
364 2012-02-10 04:37:14 <luke-jr> BlueMatt: if we don't know which routers, how do we know they support UPnP?
365 2012-02-10 04:37:18 <gmaxwell> Maybe 'just' a couple percent of users will be hit.
366 2012-02-10 04:38:19 <tcatm> I'm strongly against working around buggy NATs. They should never even look at the payload.
367 2012-02-10 04:38:52 <sipa> well, you can argue that checksumming shouldn't be part of the application-level networking either
368 2012-02-10 04:39:06 <midnightmagic> I run my home phone on voip.
369 2012-02-10 04:39:20 <gmaxwell> sipa: another possible workaround here is to just send 0.0.0.0 in that field.
370 2012-02-10 04:39:23 <BlueMatt> luke-jr: I prefer to send useful data in addrMe, if for no reason but for sanity's sake
371 2012-02-10 04:39:40 <sipa> well, for now addrMe is completely ignored by the receiver, afaik
372 2012-02-10 04:39:44 <luke-jr> BlueMatt: non-routable IPs are never useful here
373 2012-02-10 04:39:54 <sipa> but there is certainly no need to send non-routable IPs there
374 2012-02-10 04:39:58 <BlueMatt> luke-jr: the point is it should always be routable
375 2012-02-10 04:40:11 <luke-jr> &
376 2012-02-10 04:40:13 <gmaxwell> sipa: I can't think of what it would really be useful for off the top of my head.
377 2012-02-10 04:40:28 <tcatm> Well, if they do s/$LOCALIP/$PUBLICIP/g that would cause problems with messages including a number that looks like $LOCALIP...
378 2012-02-10 04:40:38 <gmaxwell> Yea.. if it's not routable just send 0.0.0.0.. makes sense to me.
379 2012-02-10 04:40:50 <sipa> let's see
380 2012-02-10 04:40:53 <BlueMatt> tcatm: I would assume its only in the first n bytes of the tcp session
381 2012-02-10 04:40:54 <sipa> i have UPnP here
382 2012-02-10 04:41:06 <gmaxwell> tcatm: they might be a bit smarter .. e.g. looking for $REMOTEIP$LOCALIP  and or only looking at the first N bytes of the session.
383 2012-02-10 04:41:58 <gmaxwell> (and we have the checksums to protect everything else)
384 2012-02-10 04:42:13 <BlueMatt> luke-jr: oh, you are saying if its not routable, send 0.0.0.0, instead of if !upnp
385 2012-02-10 04:42:16 <BlueMatt> luke-jr: thats fine by me
386 2012-02-10 04:42:33 <gmaxwell> unfortunately this isn't a soft roll. if x% of routers break this, bitcoin will stop working for x% overnight.
387 2012-02-10 04:42:42 <tcatm> Why do we need checksums over TCP?
388 2012-02-10 04:42:52 <luke-jr> tcatm: because nobody implements TCP checksums :/
389 2012-02-10 04:43:33 <midnightmagic> tcatm: Even we here require additional protection above tcp checksumming.. lots of software/devices/etc get it wrong. :(
390 2012-02-10 04:43:37 <tcatm> With v6 they are mandatory.
391 2012-02-10 04:43:41 <gmaxwell> tcatm: I don't know what the actual motivation was, but tcp checkums are pretty inadequate. They provide far less than 16 bits worth of protection on most streams.
392 2012-02-10 04:44:19 <gmaxwell> (because they're just a ones compliment sum of bytes basically, and most frames aren't actually big enough to cover the space well)
393 2012-02-10 04:44:51 <tcatm> Lower layers have checksums and error correction, too.
394 2012-02-10 04:44:53 <BlueMatt> tcatm: http://www.ir.bbn.com/documents/articles/crc-sigcomm00.ps
395 2012-02-10 04:45:00 <BlueMatt> (originally linked by gmaxwell)
396 2012-02-10 04:45:05 <gmaxwell> I've personally wittnessed tcp passing corrupted data quite a few times.
397 2012-02-10 04:45:26 <gmaxwell> tcatm: yea, see that paper wrt ethernet crc too.
398 2012-02-10 04:45:36 <midnightmagic> tcatm: It doesn't seem to help. I've seen the same thing gmaxwell has. Enough data goes by, the data corrupts..
399 2012-02-10 04:46:04 <gmaxwell> tcatm: the lower layer stuff isn't end to end and packets are often unprotected on devices busses and memories while in transit.
400 2012-02-10 04:46:26 <midnightmagic> I've even seen it on my gE home network, but then I have a good 50TB of storage that churns.. maybe a little more than I'd like.
401 2012-02-10 04:46:51 <gmaxwell> (I had a OC192 interface that managed to set a particualr bit to one about 4000 bytes into every other packet once.. that was fun to troubleshoot)
402 2012-02-10 04:47:26 <FROTUSCI> did you put it in your ear gmaxwell?
403 2012-02-10 04:47:29 <BlueMatt> networking horror stories...
404 2012-02-10 04:47:54 <gmaxwell> (and did this above the link layer CRC, of course)
405 2012-02-10 04:49:23 <gmaxwell> tcatm: in any case, we've 'got 'em' like em or not.
406 2012-02-10 04:50:14 <tcatm> Can we remove the IP from the version message and get it from the socket instead?
407 2012-02-10 04:51:00 <gmaxwell> tcatm: We don't even use that field for anything as far as I can tell, but we can't remove the field especially not in the enormous base of deployed clients.
408 2012-02-10 04:51:31 <gmaxwell> BlueMatt: I am really liking that "replace with 0.0.0.0" if the localaddress is non-routable.
409 2012-02-10 04:51:39 <tcatm> Well, set it to 0.0.0.0 for the next few years before it is removed?
410 2012-02-10 04:51:44 <BlueMatt> gmaxwell: sounds good to me
411 2012-02-10 04:51:55 <luke-jr> rather, would :: be better?
412 2012-02-10 04:52:00 <gmaxwell> BlueMatt: we already send zeros there if the proxy is set... so we know it's safe.
413 2012-02-10 04:52:13 <BlueMatt> yep
414 2012-02-10 04:52:26 <gmaxwell> luke-jr: Whatever it does the proxy behavior should be identical.
415 2012-02-10 04:52:38 <sipa> of course it is safe... it is ignored completely
416 2012-02-10 04:53:03 <gmaxwell> sipa: well even more safe  e.g. what if some old version we're forgetting doesn't ignore it?   :)
417 2012-02-10 04:53:37 <sipa> ... or some alternate implementation
418 2012-02-10 04:54:09 <gmaxwell> meh. this sucks that it's going to be a hard cut. We won't have a chance to send an alert if it looks like its going to be bad, because we won't know until too late.
419 2012-02-10 04:54:39 <luke-jr> could send out an advisory alert
420 2012-02-10 04:54:45 <gmaxwell> ugh.
421 2012-02-10 04:54:54 <BlueMatt> gmaxwell: updated: https://github.com/bitcoin/bitcoin/pull/817
422 2012-02-10 04:55:08 <luke-jr> "There is a major protocol change due tomorrow, that was scheduled 2 years ago. If your Bitcoin client stops working, please upgrade and contact us on IRC"
423 2012-02-10 04:55:09 <gmaxwell> I really don't want to use an alert for something that might only hit 1:1000 users.
424 2012-02-10 04:55:17 <BlueMatt> release 0.5.3 and make it well-known that if you have connection issues on feb 15, you need to upgrade
425 2012-02-10 04:55:20 <gmaxwell> hm. luke-jr thats not so bad.
426 2012-02-10 04:55:23 <luke-jr> gmaxwell: well, there's always the possibility of other unknowns here
427 2012-02-10 04:56:08 <gmaxwell> Can we also backport the checksum ignoring for 0.5.3? That way an installed base of 0.5.3 listners will be protective.
428 2012-02-10 04:56:49 <gmaxwell> BlueMatt: I think you forgot an -a in your amend?
429 2012-02-10 04:56:57 <luke-jr> (fUseProxy || !addrLocalHost.IsRoutable() ? CAddress(CService("0.0.0.0",0)) : addrLocalHost)
430 2012-02-10 04:57:01 <gmaxwell> oh there it is.
431 2012-02-10 04:57:02 <luke-jr> ^ is that the right precedence order?
432 2012-02-10 04:57:48 <gmaxwell> good spotting.
433 2012-02-10 04:57:50 <sipa> gmaxwell: one issue (or advantage) with checksum ignoring is that it will make clients with wrong date after februari 15th still connect
434 2012-02-10 04:57:59 <sipa> but with severe protocol misunderstandings
435 2012-02-10 04:58:21 <gmaxwell> sipa: whats the severe protocol misunderstandings?
436 2012-02-10 04:58:30 <sipa> only a few messages make it through
437 2012-02-10 04:58:46 <sipa> the version message is misinterpreted completely
438 2012-02-10 04:58:51 <sipa> so the protocol version is garbage
439 2012-02-10 04:58:52 <BlueMatt> gmaxwell: ???
440 2012-02-10 04:58:54 <gmaxwell> luke-jr: ? is super weak only assignment and comma are weaker.
441 2012-02-10 04:58:55 <luke-jr> could be smart about it
442 2012-02-10 04:59:11 <luke-jr> only ignore the checksum if the address in question is IPv4-like
443 2012-02-10 04:59:39 <sipa> or maybe better: only ignore the checksum if the version number is still sane
444 2012-02-10 04:59:50 <sipa> but what is 'sane' ?
445 2012-02-10 05:00:07 <luke-jr> meh
446 2012-02-10 05:00:10 <gmaxwell> bleh.
447 2012-02-10 05:00:16 <luke-jr> wrong-date clients will reject our version anyway
448 2012-02-10 05:00:40 <BlueMatt> luke-jr: new version of that pull also posted for 0.5.X
449 2012-02-10 05:00:42 <gmaxwell> far side inititates, we could downgrade.. but sipa said that wasn't as trivial as I thought..
450 2012-02-10 05:01:31 <gmaxwell> sipa: we'll still hang up on the wrong date nodes pretty quickly.
451 2012-02-10 05:02:31 <sipa> gmaxwell: how so?
452 2012-02-10 05:02:58 <sipa> i changed my date, and ran a client with my patch
453 2012-02-10 05:03:14 <gmaxwell> I guess not then. bleh.
454 2012-02-10 05:04:50 <gmaxwell> so, contingency plan if the nat issue is a big issue: we spin up a bunch of nodes with your patch, set to join all the IRC channels, make sure they get into dnsseed.
455 2012-02-10 05:05:10 <BlueMatt> and have a 0.5.3 on hand to tell people to use
456 2012-02-10 05:05:14 <gmaxwell> We should also make the dnsseed nodes check shortly after the cutover to evict broken nodes.
457 2012-02-10 05:05:30 <gmaxwell> yes, and have 0.5.3 already released which has the BlueMatt patch.
458 2012-02-10 05:05:49 <BlueMatt> +sipa's nochecksum?
459 2012-02-10 05:05:58 <gmaxwell> and ... I think we should do the alert like luke suggested 24 hours before the switchover.
460 2012-02-10 05:06:07 <BlueMatt> thats fair
461 2012-02-10 05:06:15 <gmaxwell> No, can't do the no checksum in a release for the reason sipa gives. :(
462 2012-02-10 05:06:40 <gmaxwell> The alert shouldn't tell people to upgrade so much as advise about the change and tell them where to go to get the latest information.
463 2012-02-10 05:07:15 <gmaxwell> (as we won't know if 0.5.3 is sufficient for all changeover related problems until its likely too late for an alert)
464 2012-02-10 05:07:21 <BlueMatt> gmaxwell: mmm, ok missed that
465 2012-02-10 05:07:37 <BlueMatt> yea, a temp page on bitcoin.org or smth
466 2012-02-10 05:07:45 <sipa> well, is completely messed up version messages such a problem?
467 2012-02-10 05:07:48 <sipa> it's dirty for sure
468 2012-02-10 05:08:23 <sipa> but it will only occur for nodes that a) have a wrong date very close to the switchover point b) wouldn't get a connection anyway either
469 2012-02-10 05:09:05 <gmaxwell> sipa: hm. perhaps .. kinda sucks that this is going to break our clock notices for a while.
470 2012-02-10 05:09:17 <gmaxwell> then again, it sounded like they weren't working right already.
471 2012-02-10 05:09:58 <gmaxwell> sipa: hm. I suppose it's not that bad.
472 2012-02-10 05:10:26 <gmaxwell> well.. they're going to misparse ours and not get the time from us either.
473 2012-02-10 05:12:20 <sipa> it seems i'm behind a double-nat
474 2012-02-10 05:12:24 <sipa> and UPnP works
475 2012-02-10 05:12:26 <BlueMatt> ...
476 2012-02-10 05:12:30 <BlueMatt> ouch
477 2012-02-10 05:12:31 <sipa> and doesn't change payload data
478 2012-02-10 05:13:42 <gmaxwell> sipa: works as in enables you to listen?
479 2012-02-10 05:13:50 <sipa> yes, and receive connections
480 2012-02-10 05:14:06 <gmaxwell> crazy. cool.
481 2012-02-10 05:14:19 <gmaxwell> yea, we really need someone that gets mangled to test this with.
482 2012-02-10 05:14:24 <sipa> so the inner nat must somehow forward the upnp message upsteam
483 2012-02-10 05:14:49 <BlueMatt> yuck
484 2012-02-10 05:15:37 <sipa> or there is just one nat, but its idea of the WAM ip is wrong
485 2012-02-10 05:15:42 <sipa> WAN
486 2012-02-10 05:15:51 <BlueMatt> wouldnt be at all surprised about that one
487 2012-02-10 05:15:53 <gmaxwell> oh so you're learning the wrong wan IP via upnp?
488 2012-02-10 05:16:01 <gmaxwell> that kinda surprises me.
489 2012-02-10 05:16:01 <sipa> indeed
490 2012-02-10 05:16:11 <gmaxwell> is it giving you another 1918 address?
491 2012-02-10 05:16:14 <BlueMatt> its upnp, its a mess
492 2012-02-10 05:16:16 <gmaxwell> Are we reading the right response?
493 2012-02-10 05:16:24 <sipa> the problem is not upnp-related
494 2012-02-10 05:16:38 <sipa> the router's web config page also shows the wrong wan ip
495 2012-02-10 05:16:43 <sipa> (it's a 10.x one)
496 2012-02-10 05:16:56 <BlueMatt> oh, yea maybe you are behind 2 nats
497 2012-02-10 05:17:06 <gmaxwell> then damn, sure sounds like double nat. Cool.
498 2012-02-10 05:17:39 <BlueMatt> cool? thats gross
499 2012-02-10 05:17:59 <sipa> cool that it works
500 2012-02-10 05:18:02 <sipa> gross that it exists
501 2012-02-10 05:18:13 <BlueMatt> yea
502 2012-02-10 05:18:46 <sipa> i mean... teredo is also an incredible hack, but so nice that it works
503 2012-02-10 05:19:41 <BlueMatt> hacks are ugly and far too common, but they do tend to work
504 2012-02-10 05:20:10 <gribble> The operation succeeded.
505 2012-02-10 05:20:10 <sipa> ;;later tell theymos it seems BBE doesn't recognize compressed pubkeys
506 2012-02-10 05:20:17 <gmaxwell> BlueMatt: it's going to get worse before it gets better.
507 2012-02-10 05:20:29 <BlueMatt> gmaxwell: yep, very, very sadly
508 2012-02-10 05:44:33 <finway> ;;seen roconor
509 2012-02-10 05:44:33 <gribble> I have not seen roconor.
510 2012-02-10 05:44:43 <finway> seen roconnor
511 2012-02-10 05:44:49 <finway> ;;seen roconnor
512 2012-02-10 05:44:49 <gribble> roconnor was last seen in #bitcoin-dev 4 days, 15 hours, 59 minutes, and 15 seconds ago: <roconnor> sipa, gmaxwell: I think I know how to turn 1000 confirmation coins back into 1 confirmation coins using duplicate transactions.
513 2012-02-10 05:45:58 <finway> someone's spreading FUD on the forum
514 2012-02-10 05:49:14 <sipa> finway: where?
515 2012-02-10 05:49:27 <finway> https://bitcointalk.org/index.php?topic=63308.0
516 2012-02-10 05:51:01 <finway> is _coinbase_double_mining&spending really an serious  issue when reorg happens?
517 2012-02-10 05:51:12 <sipa> it could cause a block chain split
518 2012-02-10 05:51:41 <finway> but split happens everyday right?
519 2012-02-10 05:51:56 <finway> s/split/reorg/
520 2012-02-10 05:52:04 <sipa> no, i mean a real split
521 2012-02-10 05:52:10 <sipa> one that doesn't get resolved
522 2012-02-10 05:52:28 <sipa> anyway, we plan to get a fix in 0.6 for that still
523 2012-02-10 05:52:36 <sipa> it's a known issue
524 2012-02-10 06:05:15 <nazguls> gmaxwell is banning everyone from #bitcoin
525 2012-02-10 06:05:23 <nazguls> dude is mad
526 2012-02-10 06:06:56 <midnightmagic> jiles- deserved it.
527 2012-02-10 06:10:42 <finway> where can i learn about REORG ?
528 2012-02-10 06:12:32 <sipa> https://github.com/bitcoin/bitcoin/blob/master/src/main.cpp#L1306
529 2012-02-10 06:13:42 <finway> I can think of some serious atcck scenario:
530 2012-02-10 06:15:19 <finway> no i can't
531 2012-02-10 06:15:52 <finway> mineA spend A(scatter to every living bitcoin address)    mineDUPA spend A
532 2012-02-10 06:15:56 <finway> reorg,   boom.
533 2012-02-10 06:16:59 <finway> i thought pirate40 have received unknown 0.001 bitcoins ?
534 2012-02-10 06:18:03 <finway> get everyone's address dirty.
535 2012-02-10 06:19:22 <finway> sipa: can that happen?
536 2012-02-10 06:22:28 <FROTUSCI> "Fifty jurisdictions, including Connecticut, Puerto Rico, the Virgin Islands, and Washington, D.C. regulate secondhand jewelers and coin dealers, in some way, such as licensing, anti-fraud, counterfeiting, or regulation of certain activities. Many of the states require secondhand jewelry and coin dealers to obtain a permit through the municipality in which the dealer does business."
537 2012-02-10 06:22:30 <FROTUSCI> " The municipalities are authorized to set permit fees, but the state maintains requirements regarding recordkeeping, state and local police accessibility, and holding periods.""
538 2012-02-10 06:22:43 <FROTUSCI> gmaxwell is on crack oblivious to legal reality
539 2012-02-10 06:24:19 <FROTUSCI> handy chart of each state's gold dealer's regulatory requirements: http://www.cga.ct.gov/2009/rpt/2009-R-0395.htm
540 2012-02-10 06:24:40 <gmaxwell> FROTUSCI: I think you've gotten your sock accounts mixed up.
541 2012-02-10 06:25:16 <FROTUSCI> gmaxwell: im just curious why you would deny something so obvious. are you affiliated with that coin site?
542 2012-02-10 06:25:35 <k9quaint> gmaxwell: are you suffering from an excess of free time?
543 2012-02-10 06:25:53 <gmaxwell> k9quaint: I'm actually having a wonderful day!
544 2012-02-10 06:27:02 <finway> All of the blocks in the old chain that are not in the new one become orphan blocks, and their generations are invalidated. Transactions that use the newly-invalid generated coins also become invalid
545 2012-02-10 06:27:28 <gmaxwell> finway: thats why bitcoin has the maturity delay.
546 2012-02-10 06:30:02 <finway> gmaxwell: well, if someguy uses coinbase_spend_coinbase_spend make an reorg
547 2012-02-10 06:30:36 <gmaxwell> finway: what about it. A 100 block reorg?  It's not as simple as you're saying.
548 2012-02-10 06:30:46 <finway> say,   mineA spendA_to_all_570000_address_left_1_satoshi  mineDUPA spendA
549 2012-02-10 06:30:51 <finway> will that cause a reorg?
550 2012-02-10 06:31:02 <gmaxwell> ...
551 2012-02-10 06:31:30 <gmaxwell> No, thats entirely unrelated. There is nothing wrong with mine/spend/minedue/spend by itself. It's harmless and fine.
552 2012-02-10 06:31:41 <gmaxwell> I don't have time right now to explain the actual concern, unfortunately.
553 2012-02-10 06:31:55 <finway> gmaxwell,thanks.
554 2012-02-10 06:33:55 <FROTUSCI> to say that coin dealers dont need to be registered is ridiculous: http://www.cga.ct.gov/2009/rpt/2009-R-0395.htm   theyre required in all but 3 states
555 2012-02-10 06:42:11 <gribble> New news from bitcoinrss: laanwj opened pull request 818 on bitcoin/bitcoin <https://github.com/bitcoin/bitcoin/pull/818>
556 2012-02-10 06:51:38 <FROTUSCI> California code: No person shall act as a commodity merchant unless the person is registered with the CFTC.  "Commodity" means any metal or mineral (including a precious metal set forth in Section 29515). "Precious metal" means the following in either coin, bullion, or other form: Silver, Gold, etc.
557 2012-02-10 06:51:39 <FROTUSCI> when you apply for a business license to deal i gold coins, they do a background check and reject you if you've had certain convictions in the past
558 2012-02-10 06:52:12 <AAA_awright_> BCBot shouldn't be messaging users without being talked to
559 2012-02-10 06:55:51 <gmaxwell> AAA_awright: it only messages you to inform you the channel is being logged.
560 2012-02-10 06:56:02 <FROTUSCI> in general if you deliver the coins quickly, within 28 days, you are exempt from CFTC requirements. but if you delay shipment, you are treated as a futures dealer
561 2012-02-10 06:56:05 <gmaxwell> AAA_awright: this is a purpose that couldn't be accomplished if it waited for you to ask.
562 2012-02-10 06:56:44 <AAA_awright> That's something that needs to go in the /topic, it's not good form for a bot to do that
563 2012-02-10 07:00:14 <FROTUSCI> sup tux
564 2012-02-10 07:49:06 <wump> I think it's good form to notify people they are being logged
565 2012-02-10 07:50:51 <sipa_> i think i'm missing conversations here?
566 2012-02-10 07:59:55 <FROTUSCI> interview with coinabul guy jay shore: http://www.youtube.com/watch?v=d46dFmAmnuc
567 2012-02-10 08:00:06 <FROTUSCI> he doesn't like being tracked down apparently
568 2012-02-10 08:00:38 <cjd> if it's the one I think it is, that an "old friend" so to speak
569 2012-02-10 08:00:45 <cjd> prior bad blood etc etc
570 2012-02-10 08:03:45 <FROTUSCI> cjd: reminds be of the bbc video where they track down a troll in cardiff
571 2012-02-10 08:07:02 <wump> sipa_: someone was complaining  about the bot
572 2012-02-10 09:20:46 <Hunterbunter> is anyone here a super scary expert on c++, python, mining, pools, openCL, etc, and could create a c++ miner in their sleep? I would like to talk to you
573 2012-02-10 09:21:31 <Hunterbunter> I don't want "just another miner"...I just functionality I want to include in my c++ app
574 2012-02-10 09:21:42 <Hunterbunter> my assumption is that if someone can write one in their sleep, they know it inside out and can help me with what I need (it's more complicated than just a miner).
575 2012-02-10 09:21:43 <sipa_> what functionality?
576 2012-02-10 09:22:21 <Hunterbunter> mostly on demand throttling
577 2012-02-10 09:22:34 <Hunterbunter> as well as autoconnected to a pool
578 2012-02-10 09:22:39 <Hunterbunter> *autoconnecting
579 2012-02-10 09:23:09 <Hunterbunter> by on demand I suppose I really mean dynamic throttling
580 2012-02-10 09:23:30 <sipa> looking at cgminer is probably a good way to start
581 2012-02-10 09:24:12 <Hunterbunter> I did have a look at the code, but I'm truly lost in it - and I have my focus elsewhere at the moment
582 2012-02-10 09:24:22 <Joric> i got best results on poclbm
583 2012-02-10 09:24:53 <Hunterbunter> I'm looking for someone to help me integrate it into my own existing app, for a significant bounty
584 2012-02-10 09:25:17 <Hunterbunter> I dont' really want to describe the whole thing on here, but if anyone is willing and able, I'm happy to chat more in private
585 2012-02-10 09:26:33 <coingenuity> Hunterbunter: i know someone who can help
586 2012-02-10 09:26:36 <coingenuity> how much is the bounty?
587 2012-02-10 09:28:16 <Hunterbunter> I'm thinking a share of profit up to 10k btc
588 2012-02-10 09:28:26 <Hunterbunter> Unfortunately I can't offer much up front
589 2012-02-10 09:29:27 <coingenuity> huh...i can point you towards the author of cgminer but he probably won't mess with it without an upfront of some sort
590 2012-02-10 09:29:42 <coingenuity> Hunterbunter: #ozcoin and chat with conman and Diablo-D3 a bit
591 2012-02-10 09:29:56 <coingenuity> one of them may be willing to help
592 2012-02-10 09:30:00 <Hunterbunter> sure it's worth a shot
593 2012-02-10 09:30:16 <coingenuity> :)
594 2012-02-10 10:12:14 <FROTUSCI> coingenuity complained to #freenode about the youtube vid. haha
595 2012-02-10 10:13:34 <FROTUSCI> the guy lost his shit at a starbuckss
596 2012-02-10 13:14:11 <TD> luke-jr: is your dns seed down ?
597 2012-02-10 13:16:00 <luke-jr> TD: everything of mine is down
598 2012-02-10 13:16:05 <TD> :(
599 2012-02-10 13:16:07 <TD> dos?
600 2012-02-10 13:17:08 <luke-jr> TD: 20 Gbit DDoS
601 2012-02-10 13:17:20 <TD> somebody wants to outrun the chain or ... ?
602 2012-02-10 13:17:22 <luke-jr> TD: Google want to take em on? ;)
603 2012-02-10 13:17:33 <luke-jr> TD: no idea what their goal is
604 2012-02-10 13:17:42 <TD> ugh
605 2012-02-10 13:18:06 <luke-jr> now that Eligius has protection, they've moved on to my personal stuff
606 2012-02-10 13:18:17 <lianj> yikes
607 2012-02-10 13:18:18 <luke-jr> apparently they DDoS'd some freenode servers too
608 2012-02-10 13:19:34 <TD> throw your stuff behind appengine and i guess you'll get the regular DoS protection for free
609 2012-02-10 13:19:41 <TD> it depends what type of attack it is. won't do much for DNS i guess
610 2012-02-10 13:19:55 <TD> and i mean he knows your provider now
611 2012-02-10 13:26:29 <Idiot___> im just wondering
612 2012-02-10 13:26:35 <Idiot___> what if all the pools are ddos'ed at the same time?
613 2012-02-10 13:31:12 <gmaxwell> Idiot___: Then it's a pretty good time for solo miners and p2pool users.
614 2012-02-10 13:43:13 <luke-jr> gmaxwell: p2pool can be ddos'd too :p
615 2012-02-10 13:44:20 <gmaxwell> luke-jr: it becomes solomining if you (well, ??? software bugs)
616 2012-02-10 13:44:59 <Idiot___> so if all the pools including p2p can that mean double spending takes place?
617 2012-02-10 13:45:26 <gmaxwell> Idiot___: No.
618 2012-02-10 13:45:37 <gmaxwell> Idiot___: the attacker still has to create blocks at the current difficulty.
619 2012-02-10 13:48:14 <helo> someone said that the p2pool mining difficulty is 4x higher... that can't be true can it?
620 2012-02-10 13:48:52 <gmaxwell> 4x higher than what?
621 2012-02-10 13:49:06 <helo> higher than mining solo, i assume
622 2012-02-10 13:49:19 <gmaxwell> Thats nonsense.
623 2012-02-10 13:49:31 <helo> good... that would be pretty broken :)
624 2012-02-10 13:49:36 <gmaxwell> helo: where is that someone?
625 2012-02-10 13:49:49 <Idiot___> i heard p2p has its own blockchain
626 2012-02-10 13:49:53 <helo> hmmmmm i'll try to find them
627 2012-02-10 13:50:18 <gmaxwell> helo: don't let them know I'm coming. Stabbing is more fun when its a surprise.
628 2012-02-10 13:50:51 <helo> heh
629 2012-02-10 13:53:12 <Idiot___> gmaxwell: dont let who know your coming?
630 2012-02-10 14:03:41 <TD> weird
631 2012-02-10 14:03:46 <TD> sendtoaddress got a lot slower for me in 0.5.2
632 2012-02-10 14:03:52 <TD> i see it hang for seconds before completing
633 2012-02-10 14:05:51 <gmaxwell> TD: has your wallet been flooded with 1e-8 inputs? thats the symptom
634 2012-02-10 14:06:12 <TD> no. it's a testnet in a box that has mined for a while
635 2012-02-10 14:06:22 <TD> so i guess it's got on the order of 10,000 transactions or so
636 2012-02-10 14:07:18 <Idiot___> wow urve got moneyz