1 2013-09-21 02:01:39 <Neozonz> anyone very knowledgable about eloipool
2 2013-09-21 04:23:59 <warren> would inserting my own compile errors using pragmas be non-portable?
3 2013-09-21 04:24:31 <Luke-Jr> I'm not sure we pay much attention to compiler portability anymore..
4 2013-09-21 04:24:35 <Luke-Jr> as long as it works with GCC and LLVM
5 2013-09-21 04:25:06 <maaku> and clang. and msvc...
6 2013-09-21 04:25:45 <Luke-Jr> clang is LLVM
7 2013-09-21 04:25:52 <Luke-Jr> and I doubt we compile under MSVC today
8 2013-09-21 04:26:38 <maaku> nitpicky but no clang and llvm are different things; many people who compile bitcoin themselves use msvc
9 2013-09-21 04:26:52 <maaku> i don't, but i see it used a lot
10 2013-09-21 04:27:32 <Luke-Jr> maaku: I would seriously be surprised if MSVC could build it. I've never heard of anyone doing such a thing in over a year.
11 2013-09-21 04:27:44 <Luke-Jr> even people building natively on Windows are few
12 2013-09-21 04:28:10 <gmaxwell> >many people who compile bitcoin themselves use msvc < I don't believe thats correct.
13 2013-09-21 04:28:31 <gmaxwell> I'm all for not being gratitiously incompatible, portablity has many benefits. But I don't believe our code currently complies in MSVC.
14 2013-09-21 04:29:07 <Luke-Jr> MSVC still doesn't support C99
15 2013-09-21 05:06:40 <warren> is it MSVC that disallows inline asm, or is it something in windows that forces you to use intrin.h's function instead and inline asm cpuid is being denied?
16 2013-09-21 05:23:28 <jgarzik> maaku, a distinction without difference. clang is part of the LLVM suite.
17 2013-09-21 05:24:10 <maaku> jgarzik: http://dragonegg.llvm.org/
18 2013-09-21 05:26:47 <warren> I'm seriously annoyed at windows...
19 2013-09-21 05:26:58 <warren> I guess I've never done any dev on windows before.
20 2013-09-21 05:29:43 <warren> the gitian build is crashing and there's no useful information wy
21 2013-09-21 05:29:44 <warren> why
22 2013-09-21 05:30:05 <warren> if I make a non-stripped build will windows show me more about where it crashed?
23 2013-09-21 05:32:03 <warren> who are the windows devs here?
24 2013-09-21 05:41:15 <Luke-Jr> warren: Diapolo is pretty much the only Windows dev, and he's only on IRC rarely
25 2013-09-21 05:42:57 <warren> using i586-mingw32msvc-objdump -d it appears this old mingw32 is compiling it wrong.
26 2013-09-21 05:43:13 <warren> maybe I should just give up until gitian is upgraded
27 2013-09-21 05:53:29 <jgarzik> "Bitcoin Protocol as Law, and the Politics of a Stateless Currency" http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2294124
28 2013-09-21 05:54:01 <jgarzik> Doesn't break much new ground. Overstates Bitcoin Foundation a bit, IMO.
29 2013-09-21 05:54:22 <jgarzik> Even without breaking new ground, though, it's nice to see bitcoin research.
30 2013-09-21 05:54:25 <phantomcircuit> jgarzik, you mean it's not the iluminati???
31 2013-09-21 05:55:00 <jgarzik> phantomcircuit, Was Satoshi /with/ the Illuminati or against them, that is the question!
32 2013-09-21 05:55:13 <jgarzik> Satoshi is Alex Jones.
33 2013-09-21 05:56:24 <jgarzik> Gah. Need to get all these bitcoin mining speadsheets to the CPA, so that I can do my taxes, so that I know how much money I can actually spend.
34 2013-09-21 05:56:37 <jgarzik> ACTION has been looking at money he cannot touch for a couple months now
35 2013-09-21 05:57:01 <phantomcircuit> jgarzik, iirc you can pay the irs on a quarterly basis if you want to
36 2013-09-21 05:57:18 <phantomcircuit> you still file an annual return but at least the money isn't sitting there eyeing you
37 2013-09-21 06:00:26 <jgarzik> not optional :)
38 2013-09-21 06:00:26 <jgarzik> phantomcircuit, more like, I -must- file a quarterly return...
39 2013-09-21 06:00:33 <jgarzik> er, quarterly PAYMENT, sorry
40 2013-09-21 06:01:19 <jgarzik> CPA yells at me, and IRS charges penalties, if quarterly payments not made.
41 2013-09-21 06:01:45 <phantomcircuit> jgarzik, no i mean the amount is quasi optional
42 2013-09-21 06:01:54 <phantomcircuit> if you pay them too much they dont return it until the end of the year
43 2013-09-21 06:02:28 <phantomcircuit> otoh there is a very real opportunity cost to not having cash on hand even if it's earmarked for taxes
44 2013-09-21 06:05:02 <Luke-Jr> jgarzik: your CPA says you have to pay taxes on bitcoins? H&R claims the IRS told them I don't until it hits fiat. I'm not sure whether to believe them or demand a refund and find someone else. :/
45 2013-09-21 06:05:28 <phantomcircuit> Luke-Jr, have you tried asking the irs?
46 2013-09-21 06:05:36 <jgarzik> Luke-Jr, I think that's a nutters position
47 2013-09-21 06:05:37 <Luke-Jr> phantomcircuit: the IRS talks to people?
48 2013-09-21 06:05:45 <phantomcircuit> Luke-Jr, yeah they do
49 2013-09-21 06:06:03 <Luke-Jr> ACTION wishes the IRS would just send him a bill.
50 2013-09-21 06:06:03 <phantomcircuit> jgarzik, H&R block is staffed by part time employees during tax season
51 2013-09-21 06:06:14 <jgarzik> Luke-Jr, advice I've gotten says to book bitcoins received in USD value at the time received
52 2013-09-21 06:06:19 <phantomcircuit> Luke-Jr, how could they send you a bill for bitcoins :)
53 2013-09-21 06:06:40 <phantomcircuit> i would be super surprised if the IRS came back to you and said you didn't have to pay
54 2013-09-21 06:07:04 <jgarzik> If I receive a bitcoin today, I report $125 in income. If I received a bitcoin a year ago, I report $2 in income (or whatever the price was, then).
55 2013-09-21 06:07:16 <phantomcircuit> jgarzik, i suspect the final conclusion will be some complicated combination of income booked at the value when you received them plus capital gains between that time and when you sell them
56 2013-09-21 06:07:29 <jgarzik> phantomcircuit, yep, was just about to write that :)
57 2013-09-21 06:07:53 <jgarzik> phantomcircuit, It is my _personal_ supposition that cap gains are also due, but I've not seen any CPA really pushing that.
58 2013-09-21 06:07:53 <phantomcircuit> which reminds me i should finish adding my fake security under quickbooks
59 2013-09-21 06:08:16 <jgarzik> and I really don't want to calc that :) :)
60 2013-09-21 06:08:22 <phantomcircuit> jgarzik, that's not the kind of thing a CPA is going to push for until the IRS says you have to pay
61 2013-09-21 06:08:38 <phantomcircuit> and i suspect it's too complicated for the IRS to really blame individuals for failing to do it
62 2013-09-21 06:08:44 <jgarzik> +1
63 2013-09-21 06:08:53 <jgarzik> it's quite complicated to figure out
64 2013-09-21 06:09:03 <jgarzik> so I am happy to avoid it until there is very clear guidance
65 2013-09-21 06:09:04 <phantomcircuit> well and actually the IRS it turns out is pretty reasonable
66 2013-09-21 06:09:07 <Luke-Jr> phantomcircuit: well, that could probably be said for the whole tax law in general :P
67 2013-09-21 06:09:11 <phantomcircuit> Luke-Jr, also yes the IRS will really talk to you
68 2013-09-21 06:09:31 <jgarzik> Luke-Jr, but yeah, that sounds like bad advice. Most other businesses book at the time of receipt.
69 2013-09-21 06:09:35 <jgarzik> from what I've seen
70 2013-09-21 06:09:43 <phantomcircuit> Luke-Jr, indeed you can probably get an appointment in person to ask the question, although they'll probably tell you they need to ask someone else in writing
71 2013-09-21 06:10:13 <phantomcircuit> now here's a real question
72 2013-09-21 06:10:27 <jgarzik> The GAO report covered this uncertainty (while not adding any certainty themselves...)
73 2013-09-21 06:10:32 <phantomcircuit> you're receiving bitcoin payments for a digital service from anonymous persons
74 2013-09-21 06:10:36 <phantomcircuit> do you collect sales tax?
75 2013-09-21 06:10:49 <phantomcircuit> notice that you have no idea which state or even country they are in
76 2013-09-21 06:11:01 <jgarzik> phantomcircuit, state-by-state answer, at present
77 2013-09-21 06:11:33 <jgarzik> pretty much the same policy as Amazon.com takes
78 2013-09-21 06:11:45 <jgarzik> (for the moment)
79 2013-09-21 06:11:50 <phantomcircuit> jgarzik, right except amazon knows which state the buyer actually is in
80 2013-09-21 06:12:03 <phantomcircuit> (ok well they really use the billing address which might be different but close enough)
81 2013-09-21 06:14:03 <phantomcircuit> jgarzik, which is even more complicated in a state like california where the sales tax changes by county and even special tax zones within counties
82 2013-09-21 06:25:51 <jgarzik> phantomcircuit, I was referring to what Amazon does /today/
83 2013-09-21 06:26:09 <jgarzik> phantomcircuit, with the upcoming regime, Amazon must know per-county and per-state regs and charge accordingly
84 2013-09-21 06:26:41 <phantomcircuit> jgarzik, right i know, but you still need to know which city (it's actually per city in CA not county) the person you're dealing with is
85 2013-09-21 06:26:45 <jgarzik> phantomcircuit, right now, as with other mail order companies, sales tax is only charged for a couple states /if you are located in that state/
86 2013-09-21 06:26:50 <petertodd> jgarzik: ...just the sort of thing Amazon would be smart to encourage...
87 2013-09-21 06:26:52 <phantomcircuit> but i dont because those people are 100% anonymous
88 2013-09-21 06:27:06 <petertodd> phantomcircuit: ha
89 2013-09-21 06:27:08 <jgarzik> phantomcircuit, thus, some laws dictate based on biz location not customer location
90 2013-09-21 06:27:11 <phantomcircuit> i guess i could start asking people which city they're in
91 2013-09-21 06:27:39 <jgarzik> phantomcircuit, it's what LLBean is in Maine and Amazon is in Seattle. ME and WA do not have such silly laws.
92 2013-09-21 06:28:06 <phantomcircuit> jgarzik, neither does the state of california
93 2013-09-21 06:28:11 <jgarzik> phantomcircuit, heh, or just create an offshore operating entity, and just import profits into the US
94 2013-09-21 06:28:46 <phantomcircuit> offshore operating entity and then wait for another tax holiday
95 2013-09-21 06:28:49 <phantomcircuit> like the big boys
96 2013-09-21 06:28:51 <phantomcircuit> heh
97 2013-09-21 06:28:54 <phantomcircuit> (inb4jail)
98 2013-09-21 06:29:39 <phantomcircuit> jgarzik, for a real mind fuck, if you purchase something in city a and bring it into city b you are required to pay the additional sales tax
99 2013-09-21 06:29:48 <phantomcircuit> afaik the law there doesn't have any time limit
100 2013-09-21 06:29:55 <petertodd> â½
101 2013-09-21 06:29:58 <Neozonz> any methods to debugging why a block was rejected at upstream?
102 2013-09-21 06:30:18 <petertodd> Neozonz: do you have the rejected block?
103 2013-09-21 06:30:20 <jgarzik> phantomcircuit, North Carolina has a "sales & use tax" where citizens must collect sales tax on purchases, if Internet sales places like Amazon do not.
104 2013-09-21 06:30:24 <jgarzik> ACTION rolls eyes
105 2013-09-21 06:30:30 <jgarzik> compliance is low
106 2013-09-21 06:30:31 <phantomcircuit> jgarzik, yeah so does ca
107 2013-09-21 06:30:50 <phantomcircuit> petertodd, the sales tax rate in various cities in ca vary by fairly large amounts
108 2013-09-21 06:31:04 <Neozonz> petertodd, yes
109 2013-09-21 06:31:04 <petertodd> phantomcircuit: I've heard - insane system
110 2013-09-21 06:31:09 <phantomcircuit> sf is like 11% or something and some are more like 7%
111 2013-09-21 06:31:19 <Neozonz> i have the block hash, ect from testnet
112 2013-09-21 06:31:24 <jgarzik> Judging by the NC Revenue Department, only 97% of citizens do not buy things online ;-) ;-)
113 2013-09-21 06:31:28 <jgarzik> Internet, what is that?
114 2013-09-21 06:31:35 <petertodd> Neozonz: but do you have the full block data?
115 2013-09-21 06:31:40 <phantomcircuit> petertodd, if i buy something in a city with sales tax of 7% and i bring it to sf, im required by law to pay the 4% to sf on the difference
116 2013-09-21 06:31:52 <Neozonz> petertodd, yup
117 2013-09-21 06:32:10 <phantomcircuit> petertodd, however there isn't apparently a time limit in law for that, so if i live in one city and move to another
118 2013-09-21 06:32:20 <phantomcircuit> technically you have to pay sales tax on everything you own
119 2013-09-21 06:32:24 <phantomcircuit> (or at least the difference)
120 2013-09-21 06:32:38 <petertodd> Neozonz: bit of a pain, but in theory you could re-sync from scratch, and have your node halt syncing when you think the block was submitted and debug it based on that
121 2013-09-21 06:32:49 <petertodd> (I'm assuming looking through debug.log didn't help here)
122 2013-09-21 06:33:08 <Neozonz> sorry wrong key
123 2013-09-21 06:33:26 <petertodd> phantomcircuit: sales tax based on what dates though? :/ like, if the sales tax rate changes...
124 2013-09-21 06:33:38 <Neozonz> do i look on the bitcoind for debug.log?
125 2013-09-21 06:33:47 <petertodd> Neozonz: ~/.bitcoin/debug.log
126 2013-09-21 06:33:56 <Neozonz> anythin specific i should look for
127 2013-09-21 06:34:01 <phantomcircuit> petertodd, excellent question, something im 100% sure the people who wrote the law didn't even think about
128 2013-09-21 06:34:49 <petertodd> phantomcircuit: Like, suppose you move from Alphaville to Betaville, and then Alphaville doubles their sales tax...be bizzare if you had to pay the double, when just being there the whole time didn't
129 2013-09-21 06:35:16 <phantomcircuit> petertodd, the biggest offenders there were car dealerships, so of course there are specific requirements for them to report based on the address the car is registered to
130 2013-09-21 06:35:22 <petertodd> Neozonz: just grep for the block hash and read the messages; if you used -debug the messages are somewhat informative
131 2013-09-21 06:35:34 <phantomcircuit> since of course the general use tax stuff that theoretically covers everything is broken to the point of not working
132 2013-09-21 06:35:48 <petertodd> phantomcircuit: ah, that makes a lot of sense... cars are the second biggest ticket item most people ever buy
133 2013-09-21 06:36:01 <petertodd> phantomcircuit: houses are the first, but usually that law doesn't apply... usually
134 2013-09-21 06:36:10 <phantomcircuit> petertodd, i guess but you could still buy from a dealer out of state
135 2013-09-21 06:36:46 <petertodd> phantomcircuit: heh, with the usual boundry condition weirdness at the borders...
136 2013-09-21 06:37:14 <phantomcircuit> smart people in ca buy their cars in oregon and pay zero sales tax
137 2013-09-21 06:37:18 <phantomcircuit> it's a lovely drive too
138 2013-09-21 06:37:32 <phantomcircuit> (inb4audit)
139 2013-09-21 06:37:53 <Neozonz> petertodd, wow thank u very much
140 2013-09-21 06:37:59 <Neozonz> i never new there was so much information in this log
141 2013-09-21 06:38:18 <petertodd> Neozonz: ha, yeah it's pretty good. I leave debug=1 in my bitcoin.conf so I'll never miss out
142 2013-09-21 06:38:37 <Neozonz> ERROR: ProcessBlock() : already have block xxxxx
143 2013-09-21 06:38:47 <Neozonz> that's what i'm getting each time
144 2013-09-21 06:38:57 <petertodd> Neozonz
145 2013-09-21 06:39:06 <petertodd> Neozonz: ah, well, is it in the best chain?
146 2013-09-21 06:41:50 <Neozonz> seems like... my client is submitblock -> setbestchain -> process block accepted -> error processblock() already have block
147 2013-09-21 06:42:39 <petertodd> well what's the height of the block? ie does getblockhash <height> return that block hash?
148 2013-09-21 06:42:50 <Neozonz> I can see a createblock before submitblock as well
149 2013-09-21 06:43:18 <petertodd> yeah, that'd be from your miner calling getblocktemplate
150 2013-09-21 06:44:40 <Neozonz> yup
151 2013-09-21 06:44:48 <Neozonz> getblockhash returns that block hash
152 2013-09-21 06:45:03 <petertodd> situation normal then
153 2013-09-21 06:45:13 <Neozonz> but it happens everytime i solve a block
154 2013-09-21 06:45:38 <petertodd> probably just means a peer is relaying it back to you then
155 2013-09-21 06:45:52 <petertodd> or soemthing like that
156 2013-09-21 06:46:12 <Neozonz> this is on testnet though
157 2013-09-21 06:46:27 <petertodd> so? you still have peers on testnet
158 2013-09-21 06:46:43 <Neozonz> so all this time i havent blocked? :o
159 2013-09-21 06:46:50 <petertodd> no, you have
160 2013-09-21 06:47:27 <Neozonz> but each time it is an orphan?
161 2013-09-21 06:47:52 <petertodd> processblock() saying you already have the block doesn't mean the block was killed, it means it's already been accepted
162 2013-09-21 06:48:16 <Neozonz> i see
163 2013-09-21 06:48:28 <Neozonz> how long till i see confirmation on my wallet for mining then?
164 2013-09-21 06:48:33 <petertodd> 120 blocks
165 2013-09-21 06:48:38 <Neozonz> or for it to show up on my wallet
166 2013-09-21 06:49:12 <petertodd> it should show up in bitcoin-qt immediately, but listed as a mining reward
167 2013-09-21 06:49:49 <Neozonz> thats the problem tho, i dont see it :(
168 2013-09-21 06:49:51 <petertodd> (though it depends - what software are you mining with?)
169 2013-09-21 06:49:57 <Neozonz> eloipool
170 2013-09-21 06:50:06 <petertodd> ah, yeah I have no idea how that works
171 2013-09-21 06:50:14 <petertodd> it may be going to a totally different address for all I know
172 2013-09-21 06:50:22 <Neozonz> me neither! lol
173 2013-09-21 06:50:34 <Neozonz> trying to figure it out while on testnet
174 2013-09-21 06:50:38 <petertodd> use the source (from) luke
175 2013-09-21 06:51:02 <Neozonz> Ahhhh
176 2013-09-21 06:51:03 <Neozonz> I see
177 2013-09-21 06:51:11 <Neozonz> It accepts the block always on the first 1
178 2013-09-21 06:51:26 <Neozonz> then 2nd time after it fails
179 2013-09-21 06:51:35 <petertodd> remember your node will only *ever* accept a block once, because the second time it already has it
180 2013-09-21 06:51:37 <Neozonz> seems eloipool just does it as reassurance
181 2013-09-21 06:51:55 <phantomcircuit> wait
182 2013-09-21 06:51:58 <phantomcircuit> is it friday?
183 2013-09-21 06:52:03 <phantomcircuit> lol it is
184 2013-09-21 06:52:07 <phantomcircuit> herp derp
185 2013-09-21 06:53:45 <petertodd> phantomcircuit: it must be a friday night because I'm in a dark room listening to pounding techno... er, on youtube
186 2013-09-21 06:54:25 <petertodd> http://www.youtube.com/watch?v=azAI2imsiRo
187 2013-09-21 07:05:20 <Neozonz> weird.... its sending the transactions to a ghost address
188 2013-09-21 07:05:27 <petertodd> huh
189 2013-09-21 07:05:33 <petertodd> ghost as in not in your wallet?
190 2013-09-21 07:05:35 <Neozonz> its not my wallet
191 2013-09-21 07:05:36 <Neozonz> lol
192 2013-09-21 07:05:41 <Neozonz> glad its testnet!
193 2013-09-21 07:05:45 <petertodd> indeed
194 2013-09-21 07:05:56 <Neozonz> but its sending it to a testnet address
195 2013-09-21 07:06:01 <Neozonz> which is stranger
196 2013-09-21 07:06:08 <petertodd> under the hood testnet and mainnet addresses are the same thing...
197 2013-09-21 07:31:22 <warren> OK, it seems upgrading the win32 gitian OS to something modern will fix this and gain the ability to build win64.
198 2013-09-21 07:31:34 <warren> maybe I should do that instead. There's no reason win32 needs the same ancient Ubuntu.
199 2013-09-21 07:32:14 <petertodd> warren: well, being the ancient ubuntu does have the advantage of making it unlikely someone backdoored it last week...
200 2013-09-21 07:32:33 <warren> petertodd: that's the real reason it was chosen? =
201 2013-09-21 07:32:34 <warren> =)
202 2013-09-21 07:32:36 <petertodd> warren: use whatever is the oldest that'll work within reason
203 2013-09-21 07:32:51 <warren> petertodd: well, that and I don't know ubuntu very well
204 2013-09-21 07:33:20 <petertodd> warren: you're gonna use redhat aren't you? :P
205 2013-09-21 07:33:35 <warren> if it has mingw, why not
206 2013-09-21 07:33:38 <petertodd> heh
207 2013-09-21 07:33:52 <warren> I trust them to make secure things more than ubuntu.
208 2013-09-21 07:34:10 <warren> The RHEL kernel team alone is several times the size of their entire company.
209 2013-09-21 07:34:35 <petertodd> heck, if you want to be really transparent when you choose this, pick, say, 2 or 4 options from different vendors and make the choice randomly based on a future block hash :/
210 2013-09-21 07:34:56 <petertodd> as the NSA I trust the RedHat people to listen to me more :P
211 2013-09-21 07:35:00 <warren> well, that might be OK, except you need a toolchain that actually works.
212 2013-09-21 07:35:16 <petertodd> warren: yeah, and evaluating even two toolchains is a pain
213 2013-09-21 07:35:27 <petertodd> warren: but if you find out two options work fine, but as well pick them randomly
214 2013-09-21 07:36:39 <warren> is mingw well maintained upstream? I've never used it before bitcoin.
215 2013-09-21 07:37:09 <Neozonz> petertodd, i feel like an idiot, the address was specified in the conf file, i was just looking at the wrong conf file ;p
216 2013-09-21 07:37:10 <petertodd> beats me
217 2013-09-21 07:37:14 <Neozonz> thanks for the help
218 2013-09-21 07:37:20 <petertodd> Neozonz: no problem
219 2013-09-21 07:39:53 <warren> it just occured to me that I don't need to install VM's with gitian to test mingw
220 2013-09-21 07:41:17 <petertodd> lol
221 2013-09-21 07:41:17 <wei_> is there a function to validate an address with bitcoinj?
222 2013-09-21 07:43:33 <warren> So Diapolo uses mingw on windows to build Bitcoin?
223 2013-09-21 07:46:53 <wei_> to answer my question above: I instantiate VersionedChecksummedBytes with the address and see if there's an error. wonder if there's a better way
224 2013-09-21 07:48:38 <warren> ARGH!
225 2013-09-21 07:49:09 <warren> mac lacks cpuid.h
226 2013-09-21 07:49:16 <warren> so I might as well use inline asm for all OS's
227 2013-09-21 07:55:01 <petertodd> wei_: by "validate" what do you want to know?
228 2013-09-21 07:55:11 <petertodd> wei_: just that the checksum is ok?
229 2013-09-21 07:55:33 <wei_> petertodd: yup.
230 2013-09-21 07:56:39 <petertodd> wei_: VersionedChecksummedBytes should be fine for that then - though you'll note that P2SH is a valid address byte but bitcoinj is still lacking support for sending to them for some reason
231 2013-09-21 07:58:35 <wei_> ideally i could call a static function instead of creating a new instance for each checksum (potentially a lot), but yes, this should work
232 2013-09-21 07:59:55 <petertodd> wei_: Yeah, in bitcoin-pythonlib there's a CBitcoinAddress class that just takes in the text string and can spit out the scriptPubKey, but IIRC bitcoinj's implementation only has the concept that an address is a pubkey or pubkey hash :/
233 2013-09-21 08:00:33 <wei_> so how would i pass in whether I want to check a testnet or prodnet address?
234 2013-09-21 08:01:05 <petertodd> wei_: in bitcoinj? dunno, it may have a global 'testnet' flag like the ref client has
235 2013-09-21 08:01:13 <petertodd> wei_: as for python-bitcoinlib, I haven't decided yet
236 2013-09-21 08:01:40 <wei_> k, reading through the code now. i'll report back :)
237 2013-09-21 08:09:21 <wei_> alright, so com.google.bitcoin.core.Address takes a network (prod or testnet) and a string to checksum. might be overkill, but that'll work for now
238 2013-09-21 08:09:37 <wei_> thanks for your help petertodd
239 2013-09-21 08:10:03 <petertodd> wei_: np
240 2013-09-21 08:53:20 <warren> petertodd: pretty much all the documented ways of detecting and using SSE2 is failing in gitian
241 2013-09-21 08:53:46 <petertodd> warren: in the VM?
242 2013-09-21 08:54:03 <warren> during build or during runtime on the target OS
243 2013-09-21 08:54:24 <petertodd> right, so running in a VM?
244 2013-09-21 08:54:32 <warren> and the ancient toolchain used to build MacOS 10.5 compatible bitcoin doesn't like the inline asm approach that works on linux and mac.
245 2013-09-21 08:54:36 <warren> I mean linux and win32.
246 2013-09-21 08:54:47 <petertodd> could be the VM is screwing it up
247 2013-09-21 08:54:50 <warren> no
248 2013-09-21 08:54:53 <petertodd> or maybe the old ubuntu is just too old
249 2013-09-21 08:54:59 <warren> probably
250 2013-09-21 08:55:10 <petertodd> odd though, because sse2 is older than ubuntu 10
251 2013-09-21 08:55:31 <warren> mingw32 lacks cpuid intrinsics entirely
252 2013-09-21 08:55:49 <warren> so I wouldn't be surprised if screws up the build
253 2013-09-21 08:55:52 <petertodd> hmm
254 2013-09-21 08:56:02 <warren> it's crashing in the SSE2 part of the code, and I don't know how to debug on windows
255 2013-09-21 08:56:10 <petertodd> me neither
256 2013-09-21 08:56:30 <wumpus> yea the gitian builder should really be upgraded to some newer ubuntu version, at least the one that cross compiles windows
257 2013-09-21 08:56:41 <wumpus> some of the hardening features are disabled too, at the moment
258 2013-09-21 08:56:53 <warren> wumpus: yeah, it would be capable of mingw64 that way too
259 2013-09-21 08:56:58 <wumpus> yes
260 2013-09-21 08:56:59 <petertodd> well, what's the oldest version that still has what we want?
261 2013-09-21 08:57:19 <wumpus> no need to use an old version for the windows cross compile at all
262 2013-09-21 08:57:21 <warren> petertodd: "oldest version" is too paranoid and counter-productive
263 2013-09-21 08:57:32 <wumpus> the only reason we use an old version for the *linux* compile is compatibility
264 2013-09-21 08:57:38 <warren> petertodd: we're suffering NOW with gitian linux compiler bugs
265 2013-09-21 08:57:53 <warren> wumpus: what OS do you dev on?
266 2013-09-21 08:58:03 <wumpus> but for the cross compile we could use 13.10 if we wanted
267 2013-09-21 08:58:09 <wumpus> 12.04 usually, sometimes 13.04
268 2013-09-21 08:58:18 <petertodd> warren: well, go ahead and use what you want then - one way of looking at it is chances are bitcoin-reference isn't what protects the vast majority of coins anyway sadly :(
269 2013-09-21 08:58:52 <wumpus> 12.04 is a long-time supported release so would be a natural choice
270 2013-09-21 09:00:03 <warren> RHEL7 might be a good choice, that will be a VERY long term supported release.
271 2013-09-21 09:00:25 <wumpus> yes but let's stick with ubuntu for now
272 2013-09-21 09:00:36 <petertodd> wumpus: agreed, lets not change two things at once
273 2013-09-21 09:00:39 <wumpus> there is no problem in switching every few years
274 2013-09-21 09:00:57 <wumpus> it's nice to have some of the newer features
275 2013-09-21 09:01:14 <warren> 12.04 is LTS?
276 2013-09-21 09:01:17 <wumpus> yes
277 2013-09-21 09:01:20 <warren> I don't know Ubuntu
278 2013-09-21 09:02:22 <warren> how long is 13.10 supported? Aside from the "omg it might be backdoored if newer" issue, I don't see a problem with everyone using the same OS for a particular gitian.sigs.
279 2013-09-21 09:03:14 <wumpus> releases between LTS are only supported very shortly
280 2013-09-21 09:03:17 <petertodd> 13.10 to 2015
281 2013-09-21 09:03:29 <petertodd> it'll be 14.04 to 2019.5
282 2013-09-21 09:03:36 <warren> wumpus: you have a newer mingw32 or mingw64 installed? could you please check if it has cpuid.h or intrin.h?
283 2013-09-21 09:03:37 <petertodd> 12.04 to 2018
284 2013-09-21 09:03:38 <wumpus> I'd really recommend going to a LTS release
285 2013-09-21 09:04:01 <warren> wumpus: some of the bugs I'd like fixed are definitely not fixed in 12.04
286 2013-09-21 09:04:15 <wumpus> warren: on newer ubuntu it's better to use mingw64, even for the 32 bit build
287 2013-09-21 09:04:21 <wumpus> warren: let's see
288 2013-09-21 09:05:07 <warren> petertodd: sounds like 13.10 could be used until 14.04 is released, which would be the next LTS?
289 2013-09-21 09:05:43 <wumpus> "i686-w64-mingw32" is the target name now (yes, it's confusing, but its's windows 32 bit :-)
290 2013-09-21 09:06:19 <petertodd> warren: well, 13.10 isn't a LTS version, on the other hand for the purposes of a gitian image we don't actually need security updates (modulo dependencies)
291 2013-09-21 09:06:24 <wumpus> lib/gcc/x86_64-w64-mingw32/4.6/include/cpuid.h
292 2013-09-21 09:06:34 <wumpus> lib/gcc/i686-w64-mingw32/4.6/include/cpuid.h
293 2013-09-21 09:06:37 <wumpus> yes
294 2013-09-21 09:06:39 <warren> intrin.h?
295 2013-09-21 09:06:40 <wumpus> it has cpuid.h
296 2013-09-21 09:07:01 <wumpus> and also intrin.h
297 2013-09-21 09:07:15 <warren> ok, #ifdef WIN32 should just use intrin.h then
298 2013-09-21 09:07:21 <warren> to match MSVC
299 2013-09-21 09:07:29 <warren> (not like MSVC actually works....)
300 2013-09-21 09:07:34 <wumpus> so 12.04 should be good enough
301 2013-09-21 09:07:38 <warren> oh?
302 2013-09-21 09:07:39 <warren> hm
303 2013-09-21 09:08:01 <warren> wumpus: do you know what changes are needed to install 12.04 for gitian? I'll test it for my problem
304 2013-09-21 09:08:18 <wumpus> very little changes, there are people that already did it without problems
305 2013-09-21 09:08:35 <wumpus> I can't help you with the details though
306 2013-09-21 09:08:37 <warren> gcc-4.6 doesn't fix pragma C++ bugs that I'd like, but that would be non-portable code anyway.
307 2013-09-21 09:09:09 <wumpus> let's avoid bleeding edge stuff anyway
308 2013-09-21 09:09:47 <warren> doing the linux build there would be incompatible due to the newer glibc?
309 2013-09-21 09:09:57 <wumpus> yes
310 2013-09-21 09:10:04 <wumpus> it would no longer run on older linux versions
311 2013-09-21 09:10:17 <petertodd> wumpus: that by itself is a very good reason to stick with 12.04
312 2013-09-21 09:10:40 <petertodd> s/stick with/upgrade to no later than/
313 2013-09-21 09:11:02 <warren> wumpus: would bitcoin-0.9 accept a gitian 12.04 upgrade for win32?
314 2013-09-21 09:11:04 <wumpus> indeed
315 2013-09-21 09:11:26 <wumpus> warren: I would, can't talk for the rest of the dev team tho
316 2013-09-21 09:11:38 <wumpus> but I think everyone is waiting for someone to do it :)
317 2013-09-21 09:11:53 <petertodd> warren: it's fine if litecoin does gitian builds on 12.04 and bitcoin does something a little different anyway
318 2013-09-21 09:12:34 <warren> petertodd: if I go through this effort for one thing, might as well try it for the other
319 2013-09-21 09:13:11 <petertodd> warren: sure, I'm just saying it's not a big deal if they decide they don't like your efforts, at least for litecoin, just a waste on their part
320 2013-09-21 09:13:56 <warren> I don't see any drawback to upgrading the win32 build environment.
321 2013-09-21 09:13:59 <wumpus> as I said, some hardening flags are also broken on the current mingw32 version, so I think upgrading the ubuntu version to 12.04 would be very welcome
322 2013-09-21 09:14:13 <wumpus> it'd increase security!
323 2013-09-21 09:14:22 <petertodd> +1 on that
324 2013-09-21 09:14:37 <warren> and probably majority of the users are using win32 ...
325 2013-09-21 09:14:50 <petertodd> and 12.04 strikes me as an uncontroversial choice - it's gotta be one of the most deployed linux's out there
326 2013-09-21 09:14:54 <warren> wumpus: one of our largest pools actually makes their own custom win64 build. that was unexpected.
327 2013-09-21 09:15:23 <petertodd> warren: I wouldn't be so sure about that; I'd expect a decent % of the windows users to just use other wallets rather than run full nodes
328 2013-09-21 09:15:45 <warren> petertodd: how's your client download statistics?
329 2013-09-21 09:17:01 <petertodd> warren: blockchain.info >>>> anythin else
330 2013-09-21 09:17:19 <petertodd> warren: I don't have stats on linux-vs-win32, and anyway a big chunk of linux people will be compiling from source
331 2013-09-21 09:24:41 <petertodd> ok, rough estimate: IIRC windows and mac boxen basically never are setup to respond to pings, while Linux boxen will modulo their firewalls. I got 31 peers that didn't respond to pings, vs. 55 that did
332 2013-09-21 09:24:52 <petertodd> dunno what consumer routers do mind you, but interesting result anyway
333 2013-09-21 09:27:46 <warren> consumer routers dominate, you almost never find end-user home computer with a public IP anymore
334 2013-09-21 09:30:03 <petertodd> right, but do those routers usually respond to pings?
335 2013-09-21 09:33:59 <petertodd> Oh, and eyeballing it, looks like the majority of peers are not on consumer ISPs either.
336 2013-09-21 09:35:13 <petertodd> for example, one of my peers is 68.168.104.10, which hosts alboappliance.com (!)
337 2013-09-21 09:38:17 <petertodd> Hmm... yeah, so I went through all my peers, and it's only 25% on residental connections, the other 75% are business/colo/vps.
338 2013-09-21 09:38:56 <petertodd> (not including tor hidden service peers)
339 2013-09-21 09:51:31 <warren> petertodd: consumer routers respond to ping by default
340 2013-09-21 09:51:36 <warren> generally
341 2013-09-21 09:52:57 <petertodd> warren: well in any case the ip addresse reverse-dns and whois records pretty clearly shows consumers are running a minority of peers connected to my node
342 2013-09-21 09:53:27 <petertodd> FWIW very few non-Satoshi peers either, just two Bitcoinj peers
343 2013-09-21 10:17:31 <warren> wumpus: petertodd: my particular SSE2 issue that crashes with 10.04 mingw32 works fine with fedora's mingw64
344 2013-09-21 10:17:44 <warren> miscompile or something
345 2013-09-21 10:18:34 <wumpus> yet again proves that the 10.04 ming32 is buggy
346 2013-09-21 11:25:52 <berndj> does bitcoind make any attempt to open the berzerkely db files with DB_DIRECT? (which ultimately becomes open(..., ...|O_DIRECT) doesn't it?)
347 2013-09-21 12:27:51 <lclc> has somebody a list of valid and invalid Bitcoin addresses to test a regex?
348 2013-09-21 12:54:58 <berndj> lclc, what level of "valid"
349 2013-09-21 12:55:36 <berndj> do you consider 1BitcoinEaterAddressDontSendf59kuE to be valid? how about 1BitcoinEaterAddressDontSendf59kuF?
350 2013-09-21 12:57:25 <lclc> they look both valid to me
351 2013-09-21 12:58:19 <berndj> i'm pretty sure the latter isn't valid (you can't send coin to it - bad checksum)
352 2013-09-21 12:59:11 <lclc> mh ok, that's why I'm asking for a list
353 2013-09-21 13:00:47 <berndj> i'm not sure you could distinguish between valid and invalid on that level ^^ unless maybe you write a Rfc822.pm-like regex
354 2013-09-21 13:01:47 <lclc> yes looks like I can't do it with just one regex. I'll regex first and then validate with checksum
355 2013-09-21 13:29:03 <jgarzik> *facepalm* *facepalm*
356 2013-09-21 13:29:24 <jgarzik> Bitcoin Wallet android app activates itself
357 2013-09-21 13:29:32 <jgarzik> i.e. connecting to the network etc
358 2013-09-21 13:29:40 <jgarzik> no way to turn this off, without uninstalling the app
359 2013-09-21 13:30:15 <jgarzik> no way to achieve the normal "run this app only when the user explicitly loads the app"
360 2013-09-21 13:32:17 <phantomcircuit> jgarzik, lol
361 2013-09-21 13:32:29 <phantomcircuit> jgarzik, im not really surpised
362 2013-09-21 13:32:41 <phantomcircuit> it seems like bitcoinj and derivatives are
363 2013-09-21 13:32:43 <phantomcircuit> uh
364 2013-09-21 13:32:45 <phantomcircuit> quirky
365 2013-09-21 13:32:58 <phantomcircuit> no offense to TD
366 2013-09-21 13:33:26 <phantomcircuit> holy god it's 6:30am
367 2013-09-21 13:33:35 <phantomcircuit> i should be asleep o.o
368 2013-09-21 13:35:45 <Tril> jgarzik: well, you could set android wallet to use only a trusted peer of localhost. then it only connects if you SSH somewhere and forward port 8333.
369 2013-09-21 13:36:45 <Tril> it used to have an option to only synch blocks when on AC power, was this removed?
370 2013-09-21 13:37:02 <Tril> (or when open)
371 2013-09-21 14:17:54 <jgarzik> Tril, that option was removed, yes :(
372 2013-09-21 14:18:10 <jgarzik> in favor of the "developers know better than user when the app should be run" model.
373 2013-09-21 14:23:32 <jgarzik> *facepalm*
374 2013-09-21 14:23:52 <jgarzik> "The android app lifecycle is install to uninstall" (meaning you cannot turn it off, without uninstalling)
375 2013-09-21 14:24:05 <jgarzik> We must re-learn the lessons of other OS's, it seems.
376 2013-09-21 14:28:03 <jgarzik> Having phones connect to the bitcoin network, whenever they see a network, is just wrong.
377 2013-09-21 14:28:23 <jgarzik> Really crappy privacy behavior
378 2013-09-21 14:43:25 <yrashk> I am having a situation where my local bitcoind never sends my tx references in invs. I do regular getblocks (from scratch to the end), not sure if it affects
379 2013-09-21 14:43:31 <yrashk> sends me*
380 2013-09-21 14:43:38 <yrashk> it only sends blocks
381 2013-09-21 14:43:54 <yrashk> am I missing something?
382 2013-09-21 14:44:53 <gmaxwell> "tx references"?
383 2013-09-21 14:45:34 <yrashk> tx hashes
384 2013-09-21 14:46:15 <yrashk> I get invs but no MSG_TXs there
385 2013-09-21 14:46:21 <gmaxwell> Please step back and stay what you're trying to accomplish. I can't decode whats wrong from the level of detail that you're providing.
386 2013-09-21 14:47:32 <gmaxwell> s/stay/say/
387 2013-09-21 14:47:50 <yrashk> I connect to bitcoind, request getblocks from the beginning, and keep asking getblocks continuosly from the genesis, continuously updating locator in those requests
388 2013-09-21 14:48:02 <yrashk> according to the wiki, I should also receive unsolicited invs
389 2013-09-21 14:48:22 <yrashk> and I assumed it will be invs with both MSG_TX and MSG_BLOCK inventory
390 2013-09-21 14:48:25 <yrashk> makes sense so far?
391 2013-09-21 14:49:07 <gmaxwell> You have not described what you are trying to accomplish.
392 2013-09-21 14:50:07 <yrashk> I am listening for blocks and transactions, requesting them using getdata upon receipt of inv
393 2013-09-21 14:54:41 <yrashk> I assumed if my local bitcoind receives an inv with a MSG_TX inside it, it will send an inv to its peer (my client in this case) advertising availability of this tx
394 2013-09-21 14:56:53 <yrashk> ok, I figured it out
395 2013-09-21 14:57:02 <yrashk> I needed to set version.relay to 1
396 2013-09-21 14:58:45 <gmaxwell> In the future you will recieve better help if you first begin with what you are trying to accomplish, instead of limiting yourself to protocol minutia and forcing people to attempt to decode what you want.
397 2013-09-21 14:59:20 <gmaxwell> If you had said "I am trying to build a piece of software which connects to a node and watches for announcement of a transaction." I would have responded to you about the flags directly.
398 2013-09-21 15:00:11 <yrashk> I apologize
399 2013-09-21 15:00:19 <gmaxwell> But with what you were saying I couldn't figure out if you expected nodes to INV things they'd never fetched and validated, or if you were confused about INV vs GET or if you were watching for particular transactions which weren't being relayed due to relay rules or validity, or whatever.
400 2013-09-21 15:00:23 <yrashk> I've been up for almost 24h, pulling an all-nighter
401 2013-09-21 15:00:30 <gmaxwell> No need to apologize.
402 2013-09-21 15:09:33 <yrashk> is there any way to more or less reliably distinguish invs sent in response to getblocks (with the limit of 500 items) and unsolicited invs? I am trying to avoid sending "continuation" getblocks in response to unsolicited invs
403 2013-09-21 15:14:42 <yrashk> for example, would the first tx in the first block of the inv array have prev_block referencing exactly my locator hash (I only send one at a time)?
404 2013-09-21 15:18:39 <yrashk> I am sorry, just the prev_block of the first block
405 2013-09-21 15:22:28 <yrashk> looks like I can https://github.com/bitcoin/bitcoin/blob/e0d6dd11e3d9246b510a78118497d81b6d94ee14/src/main.cpp#L3658
406 2013-09-21 15:55:00 <aoeu> Hi guys, check out my new Bitcoin wallet app for Windows Phone http://www.windowsphone.com/en-us/store/app/bitcoin/ca65fc5b-14f0-4da4-8e39-e2d4b702b2ea
407 2013-09-21 15:56:25 <tgs3> aoeu: is it open source?
408 2013-09-21 15:56:44 <aoeu> tgs3: It will be. I got to clean up the code up bit.
409 2013-09-21 15:56:58 <tgs3> anyway windows and security... heh
410 2013-09-21 15:58:08 <aoeu> tgs3: Why is that?
411 2013-09-21 15:58:21 <aoeu> tgs3: It's not Windows, it's Windows Phone. It's just as secure as iOS, and more secure than Android.
412 2013-09-21 16:01:08 <tgs3> aoeu: microsoft is well known to aid NSA to hack it's users
413 2013-09-21 16:01:36 <tgs3> they even install NSA keys to allow NSA to install malware in windows even signed as legitimate windows drivers - the _NSAKEY scandal
414 2013-09-21 16:02:09 <tgs3> and NSA is well known in mass hacking everyone
415 2013-09-21 16:02:43 <aoeu> So are Apple and Google.
416 2013-09-21 16:02:54 <aoeu> tgs3: The NSA doesn't care about your Bitcoins...
417 2013-09-21 16:03:14 <aoeu> tgs3: But if you want to wear the tinfoil hat, go for it.
418 2013-09-21 16:05:23 <tgs3> aoeu: just saying winows phone is totall shit for security
419 2013-09-21 16:05:47 <tgs3> still, I guess it's fine that even the people sadly using it can be enabled to participate in bitcoin
420 2013-09-21 16:06:31 <aoeu> tgs3: You're a bigot. You have no idea what you're talking about, you don't know anything about Windows Phone.
421 2013-09-21 16:06:43 <aoeu> Windows Phone is not less secure than other OSes.
422 2013-09-21 16:07:03 <tgs3> aoeu: I know it is made by company that was captured willingly allowing 3rd party to backdoor own system
423 2013-09-21 16:07:19 <tgs3> for me that is a huge minus, maybe you are o.k. with it ;)
424 2013-09-21 16:07:42 <aoeu> tgs3: It's a mistake to use Bitcoin Qt on my PC I guess. So is using an online wallet on Chrome.
425 2013-09-21 16:07:57 <aoeu> tgs3: You don't know if Apple does that too.
426 2013-09-21 16:08:01 <tgs3> aoeu: why would it be a mistake to use Qt on PC?
427 2013-09-21 16:08:13 <tgs3> I know
428 2013-09-21 16:08:19 <tgs3> of course Apple also is in bed with feds
429 2013-09-21 16:08:35 <tgs3> that is why (of reasons) why we have open source
430 2013-09-21 16:08:49 <tgs3> *one of reasons
431 2013-09-21 16:10:10 <tgs3> aoeu: did you just implied that PC === Windows, and only alternative is Mac? :o
432 2013-09-21 16:10:37 <aoeu> tgs3: Because "Windows is evil and the NSA".
433 2013-09-21 16:11:01 <aoeu> tgs3: I did not. I would guess that there's no NSA backdoor in Linux.
434 2013-09-21 16:11:53 <tgs3> if you followed news, there where attempts by "someone" to backdoor Linux, it seems they failed, and with open source we are possitioned better to capture it
435 2013-09-21 16:12:29 <tgs3> there is a reason why bitcoin is FOSS and why we work on verificable binary builds of it. It would be silly to use bitcoin if it would be closed-source don't you think?
436 2013-09-21 16:14:30 <tgs3> *were
437 2013-09-21 16:15:25 <aoeu> But Bitcoin Qt does run on Windows.
438 2013-09-21 16:15:43 <aoeu> Can't they still track what you do? Most likely.
439 2013-09-21 16:16:07 <aoeu> Heck, my Windows Phone app is open source. I compile it myself and send it to the store. It's not different from Qt.
440 2013-09-21 16:16:16 <aoeu> The only difference is that you have to trust me.
441 2013-09-21 16:16:57 <tgs3> aoeu: Bitcoin Qt runs on all 3 os
442 2013-09-21 16:17:18 <aoeu> So?
443 2013-09-21 16:17:36 <aoeu> What if my app was made using Phonegap, which runs on all mobile OSes?
444 2013-09-21 16:18:10 <tgs3> I simply ment that your app will be insecure because the underlying os is not to be trusted
445 2013-09-21 16:18:37 <tgs3> and I already said, that you are right that NSA does not (currently) care to go after all bitcoiners, so it's not a huge issue for casual users
446 2013-09-21 16:21:15 <tgs3> but sure the people who did choose to be on windows phones, will be better off with access to a wallet, so yey :) aoeu \
447 2013-09-21 19:54:05 <TD> BlueMatt: ping
448 2013-09-21 20:19:25 <Luke-Jr> ACTION peers at the rapid sending of large amounts of BTC
449 2013-09-21 20:41:17 <jgarzik> TD, discovered a Bitcoin Wallet issue -- it seems you cannot turn it off
450 2013-09-21 20:41:26 <TD> turn it off?
451 2013-09-21 20:42:18 <TD> just disable the connection bars notification. that's off by default anyway, so i guess you switched it on?
452 2013-09-21 20:42:45 <warren> TD: connection bars notification means the wallet randomly turns itself on ?
453 2013-09-21 20:43:12 <TD> no. the wallet always syncs in the backgound, with backoff if you don't really use it
454 2013-09-21 20:43:17 <TD> the connection bars just make it obvious
455 2013-09-21 20:43:37 <TD> apps syncing in the background is normal on android. just switch off the bars so the notification doesn't bother you
456 2013-09-21 20:43:41 <warren> no options to NEVER sync in the background/
457 2013-09-21 20:44:56 <TD> nope. there's a thread about that in the forums, but nobody could come up with a good reason for that behaviour to be settable beyond "i'm a power user who likes to feel in control", which isn't really a reason when you're trying to make software simple. so andreas doesn't want to add it.
458 2013-09-21 20:45:04 <TD> if anything the connection bars thing shouldn't be an option either
459 2013-09-21 20:46:48 <TD> if you don't open the app then it eventually backs off to syncing only once per day
460 2013-09-21 20:46:52 <jgarzik> TD, That's the problem. It does not offer the user /any/ option to turn off background, even if (say) you don't want the power drain, you don't ever use the wallet, you don't want the privacy intrusion, etc.
461 2013-09-21 20:47:11 <jgarzik> It's pretty easy to watch a Bitcoin Wallet wander around the city, if the keys (bloom filter) doesn't change
462 2013-09-21 20:47:33 <jgarzik> And it's just plain Basic Right that a user should be able to turn off an app on their own phone.
463 2013-09-21 20:47:34 <TD> if you don't ever use the wallet, move the coins out and uninstall it. seeing the wallet wander around a city - how, exactly?
464 2013-09-21 20:47:43 <Tril> ACTION agrees with jgarzik. No security awareness, sad. It's easy to spoof a GSM tower, monitor port 8333, then find wallets to possibly steal..
465 2013-09-21 20:48:03 <TD> no, not on android. all kinds of things on android sync in the background. just do an "adb logcat" sometime. your phone is very often doing things without you seeing it directly
466 2013-09-21 20:48:04 <jgarzik> TD, unencrypted protocol, unchanging bloom filter makes a great fingerprint
467 2013-09-21 20:48:16 <TD> fingerprint for who?\
468 2013-09-21 20:48:23 <jgarzik> This is _money_
469 2013-09-21 20:48:25 <chmod755> it's also easy to disable mobile internet
470 2013-09-21 20:48:28 <jgarzik> Not just some random process
471 2013-09-21 20:49:56 <Luke-Jr> jgarzik: can you just add a dummy hash to the bloom filter to make it unique each time, or would it still be comparable to other ones?
472 2013-09-21 20:50:08 <TD> if you can demonstrate to me or andreas following users around a city by, er, spoofing GSM towers and mapping their bloom filters to their real identities, i'm sure that would give him reason to reconsider
473 2013-09-21 20:50:16 <TD> but i bet you can't
474 2013-09-21 20:50:20 <Luke-Jr> and I agree with jgarzik, if for no reason other than controlling battery life
475 2013-09-21 20:50:38 <Luke-Jr> but I also see TD's point that this is evil-Google-Android where you have no privacy or control <.<
476 2013-09-21 20:50:54 <TD> as i said, if you don't use the app then it backs off to once per day, at which point battery impact is negligible (<1%)
477 2013-09-21 20:51:16 <Luke-Jr> TD: what if I use it, but don't want battery drain?
478 2013-09-21 20:51:19 <TD> if you see significant battery impact even when you don't use the app, then that's a bug
479 2013-09-21 20:51:39 <TD> if you use the app then it makes no difference. total work done is the same. what it's doing is catching up with the chain in the background
480 2013-09-21 20:51:44 <warren> I uninstalled this app because it didn't give user control.
481 2013-09-21 20:51:52 <TD> whether that happens in spurts in the background or all at once, the energy usage is equal
482 2013-09-21 20:52:05 <TD> ACTION shrugs
483 2013-09-21 20:52:10 <Luke-Jr> TD: spurts can be on AC power
484 2013-09-21 20:52:20 <Luke-Jr> err
485 2013-09-21 20:52:24 <Luke-Jr> all-at-once*
486 2013-09-21 20:52:35 <Luke-Jr> and wifi, so it doesn't use cellular data
487 2013-09-21 20:52:45 <Luke-Jr> there are a TON of reasons to not do it in the background always
488 2013-09-21 20:52:45 <TD> yeah. and i think it will do the daily catchup at night for exactly that reason
489 2013-09-21 20:52:49 <edcba> isn't the app a service that can be stopped ?
490 2013-09-21 20:52:50 <TD> it certainly used to sync on power
491 2013-09-21 20:53:21 <michagogo> cloud|Wait, what? Bitcoin Wallet on Android is always-on? O_o
492 2013-09-21 20:53:22 <Luke-Jr> ACTION wonders why Nexus 7 requires recharging every week or so
493 2013-09-21 20:53:32 <TD> warren: it's andreas' app so you could take it up with him, but lots of people already did. "i want control" is *not* a reason that will convince him. he's making this app for ordinary people, he doesn't want it to end up looking like Lotus Notes
494 2013-09-21 20:53:34 <Luke-Jr> my smaller handheld can go a month in the same conditions
495 2013-09-21 20:53:52 <TD> michagogo|cloud: no. it is started by the OS once a day, if you don't use it much
496 2013-09-21 20:53:57 <TD> for a minute or two usually
497 2013-09-21 20:54:32 <Luke-Jr> jgarzik: Qt 5 supports Android and iOS now I hear
498 2013-09-21 20:54:40 <Luke-Jr> I wonder if we can get Bitcoin-Qt going on it
499 2013-09-21 20:55:09 <TD> ACTION would love to see what bitcoin-qt would do to battery life ....
500 2013-09-21 20:55:30 <chmod755> hmmm... does bitcoind work on android?
501 2013-09-21 20:55:32 <Luke-Jr> nothing, as long as you don't run it while on batteries..
502 2013-09-21 20:55:39 <gmaxwell> I was already thinking of doing an intermittent mode for bitcoind for people who aren't listening.
503 2013-09-21 20:55:53 <gmaxwell> (not that I think that running bitcoin-qt on android is a good idea. :) )
504 2013-09-21 20:55:55 <TD> chmod755: no. android limits the amount of memory a single app can use and there's no swap
505 2013-09-21 20:56:05 <Luke-Jr> gmaxwell: someone has a pullreq up for a network activity toggle
506 2013-09-21 20:56:09 <TD> chmod755: maybe on the very latest tablets you can use hundreds of megs of memory, but it's really not advisable
507 2013-09-21 20:56:32 <gmaxwell> Luke-Jr: yea, I'm aware, I was thinking of basically a switch to make that automatic.
508 2013-09-21 20:56:36 <TD> what problem with bitcoin-qt on android/iOS solve though?
509 2013-09-21 20:56:45 <Luke-Jr> TD: lack of a sensible wallet on Android
510 2013-09-21 20:56:50 <TD> â¦.
511 2013-09-21 20:56:55 <edcba> lol
512 2013-09-21 20:57:05 <gmaxwell> TD: only luke was saying that. Don't think I was. :P
513 2013-09-21 20:57:35 <Luke-Jr> gmaxwell: I'm thinking after sipa's headers-only stuff is merged
514 2013-09-21 20:57:37 <TD> so let me get this straight - you are upset about an app that syncs once a day for less than a minute usually, and which has (for me at least) no impact on battery life at all
515 2013-09-21 20:57:44 <TD> and your fix is to run a *fully validating node*
516 2013-09-21 20:58:10 <edcba> there are android tablets though
517 2013-09-21 20:58:17 <edcba> bitcoin could run on it
518 2013-09-21 20:58:24 <gmaxwell> TD: sounds about as reasonable as 13:49 < TD> if you can demonstrate to me or andreas following users around a city by, :P
519 2013-09-21 20:58:33 <Luke-Jr> XD
520 2013-09-21 20:58:57 <gmaxwell> TD: but hey, as you note bitcoin wallet for android is not a power user thing.
521 2013-09-21 20:59:12 <gmaxwell> (and again, I am not really arguing that running bitcoin-qt on android is a great plan)
522 2013-09-21 20:59:18 <Luke-Jr> ACTION can accept "wrong audience", but that infers another client for your own audience is desirable
523 2013-09-21 20:59:28 <edcba> if ppl wants an off switch, fork the thing and be happy
524 2013-09-21 20:59:38 <TD> what options do you want from bitcoin-qt that aren't in the android app? other than an off switch?
525 2013-09-21 21:00:11 <TD> which i don't even get, because if you don't let bitcoin-qt sync in the background constantly it'll fall behind and when you actually want an up to date balance it'll take ages to catch up
526 2013-09-21 21:00:20 <TD> (which is one reason why it's painful on laptops)
527 2013-09-21 21:00:39 <Luke-Jr> TD: sipa's newer stuff should be nearly instant for that
528 2013-09-21 21:00:52 <TD> i think you misunderstood what sipa has implemented
529 2013-09-21 21:00:56 <TD> he has not implemented SPV mode
530 2013-09-21 21:01:09 <TD> even with his stuff merged in, you still download and validate full blocks
531 2013-09-21 21:01:14 <Luke-Jr> for now
532 2013-09-21 21:01:18 <Luke-Jr> but it works before you do
533 2013-09-21 21:01:39 <Luke-Jr> I think?
534 2013-09-21 21:01:49 <gmaxwell> Decode that?
535 2013-09-21 21:01:58 <TD> what works? no, it's exactly the same. until you downloaded all the blocks you will see a stale balance
536 2013-09-21 21:02:03 <chmod755> i want an option to send my bitcoins to my home-wallet when someone steals my phone
537 2013-09-21 21:02:20 <TD> the point of doing headers first is to allow for parallelization of the download process
538 2013-09-21 21:02:24 <gmaxwell> chmod755: keep a copy of your portable wallet.
539 2013-09-21 21:02:29 <michagogo> cloud|chmod755: that's called a backup :-P
540 2013-09-21 21:02:31 <TD> it makes a big difference when you're network bottlenecked
541 2013-09-21 21:02:36 <michagogo> cloud|gmaxwelled
542 2013-09-21 21:03:00 <gmaxwell> TD: as a side effect it also removes a bunch of delays from slow/misbehaving peers.
543 2013-09-21 21:03:07 <chmod755> gmaxwell, ok, but then i want it to shred the wallet on my phone
544 2013-09-21 21:03:09 <TD> right
545 2013-09-21 21:03:10 <chmod755> :D
546 2013-09-21 21:03:22 <TD> chmod755: send all the money in your backup to a fresh address
547 2013-09-21 21:03:28 <TD> "shredded"
548 2013-09-21 21:03:35 <gmaxwell> chmod755: when you're robbed just take your backup and move all the coins out of it.. the wallet on the phone is then worthless.
549 2013-09-21 21:03:55 <edcba> but something to automatically do it would be better
550 2013-09-21 21:03:55 <gmaxwell> (hey, maybe the theif will deposit more coins in it and you can take those too⦠:P)
551 2013-09-21 21:04:11 <Luke-Jr> when my device is stolen, I want it to turn on the camera and GPS, and save all the data to a server
552 2013-09-21 21:04:13 <Luke-Jr> <.<
553 2013-09-21 21:04:15 <chmod755> gmaxwell, lol
554 2013-09-21 21:04:45 <TD> anyway, i did look at running bitcoin-qt on android years ago. before i started on bitcoinj
555 2013-09-21 21:04:53 <gmaxwell> "Well, I lost my phone, but the theif deposited enough for me to steal back to pay for it"
556 2013-09-21 21:05:07 <TD> it would have required a lot of somewhat tricky changes, which is one reason i didn't want to do it. i was scared i'd break something
557 2013-09-21 21:05:13 <TD> for instance, you can't really hold all the block headers in memory
558 2013-09-21 21:05:38 <TD> but that assumption is pretty heavily baked into the code
559 2013-09-21 21:05:42 <gmaxwell> TD: sometimes silly questions like "bitcoind on android" can inspire useful thinking in any case. E.g. if someone wants to go reduce our memory usage to make it possible, then that may be an independant good.
560 2013-09-21 21:06:01 <TD> well, yes, although complex changes have risk
561 2013-09-21 21:06:06 <TD> so there'd better be a good reason :)
562 2013-09-21 21:06:17 <chmod755> bitcoind on openwrt!
563 2013-09-21 21:06:24 <gmaxwell> Sure, though the headers themselves are under 20mbytes now. Our static usage is nearly 300mbytes.
564 2013-09-21 21:06:31 <TD> yeah
565 2013-09-21 21:06:39 <TD> when i started, 16mb was a not uncommon hard limit â¦.
566 2013-09-21 21:06:43 <Luke-Jr> C++ kinda breaks mmap, doesn't it? :/
567 2013-09-21 21:06:48 <TD> i don't know what the current distribution of memory classes is
568 2013-09-21 21:06:54 <TD> better. but still not amazing.
569 2013-09-21 21:07:15 <TD> the top crasher for the android wallet app is out of memory failures
570 2013-09-21 21:07:22 <michagogo> cloud|How much RAM does an RPi have?
571 2013-09-21 21:07:32 <Luke-Jr> michagogo|cloud: 256-512 IIRC
572 2013-09-21 21:07:36 <TD> Luke-Jr: breaks mmap?
573 2013-09-21 21:07:37 <gmaxwell> In any case, running bitcoin-qt on some kind of high power android tablet should be possible. Not sure that you'd want to, but it should be possible.
574 2013-09-21 21:07:49 <Luke-Jr> TD: no way to say "this object is allocated in a mmap'd file"?
575 2013-09-21 21:07:57 <gmaxwell> Luke-Jr: you can use stl containers with whatever allocators you want.
576 2013-09-21 21:07:58 <TD> c++ has placement new
577 2013-09-21 21:08:05 <TD> so yeah, you can do that
578 2013-09-21 21:08:08 <Luke-Jr> oh?
579 2013-09-21 21:08:50 <gmaxwell> It's easy to make a mess if all the allocations aren't in one place. E.g. you have objects in allocator a who have references to objects allocated in allocator b.
580 2013-09-21 21:08:54 <Luke-Jr> hm
581 2013-09-21 21:08:59 <gmaxwell> But its certantly possible.
582 2013-09-21 21:09:09 <gmaxwell> There is very little that isn't possible in C++, for better or worse. :)
583 2013-09-21 21:09:53 <TD> AFAIK the largest heap size any android device will let you use is 256mb
584 2013-09-21 21:10:00 <TD> so bitcoind would need some optimisation to even stand a chance
585 2013-09-21 21:10:12 <gmaxwell> michagogo|cloud: if you're even asking that question I recommend: http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G135341370451
586 2013-09-21 21:10:49 <Luke-Jr> interesting
587 2013-09-21 21:10:53 <gmaxwell> TD: right but there really is no reason that we shouldn't fit comfortably in that. We don't currently... but there isn't anything interesting that we're doing which should preclude that.
588 2013-09-21 21:11:01 <TD> yeah
589 2013-09-21 21:11:24 <michagogo> cloud|gmaxwell: Just wondering if it'll run a bitcoind
590 2013-09-21 21:11:43 <gmaxwell> michagogo|cloud: there are people running it on rpi, but I don't generally recommend it. an odroid is a lot better. :P
591 2013-09-21 21:11:46 <michagogo> cloud|I don't have one and am not planning on trying to run bitcoind on one
592 2013-09-21 21:11:59 <michagogo> cloud|Just wondering...
593 2013-09-21 21:13:16 <Luke-Jr> michagogo|cloud: you need to push that commit w the same branch name
594 2013-09-21 21:13:34 <michagogo> cloud|I think I did.
595 2013-09-21 21:13:45 <michagogo> cloud|pgp, both times
596 2013-09-21 21:14:14 <michagogo> cloud|(Though for some reason git checkout can't find my commit from 2 months ago)
597 2013-09-21 21:14:28 <Luke-Jr> michagogo|cloud: did the push succeed?
598 2013-09-21 21:14:49 <michagogo> cloud|Yes -- as you can tell by the fact that the new commit is github-link able
599 2013-09-21 21:15:01 <gmaxwell> at some point in the not too distant future we're likely to start having problems with insufficient listening public nodes, getting the memory usage down should help. Looks like we've lost maybe ~30% of them in the last 6 months.
600 2013-09-21 21:15:06 <michagogo> cloud|(Also, I couldn't seem to find the original branch...)
601 2013-09-21 21:15:45 <michagogo> cloud|I recreated it.
602 2013-09-21 21:15:56 <michagogo> cloud|(Would that cause problems?)
603 2013-09-21 21:16:08 <michagogo> cloud|ACTION is a git noob
604 2013-09-21 21:16:42 <Luke-Jr> gmaxwell: I'm sure Google will run them for us! >_<
605 2013-09-21 21:18:51 <gmaxwell> Hm. getaddr.bitnodes.io shows successful getversion from 6072 on 2013-06-14 to 4119 on 2013-09-20. Which I think is a bit more drop than Luke and Sipa's spiders have found.
606 2013-09-21 21:20:47 <Luke-Jr> gmaxwell: I see over 74,000 0.8.x nodes alone :o
607 2013-09-21 21:21:30 <gmaxwell> Luke-Jr: thats ever connect, not actually reachable at any time.
608 2013-09-21 21:21:46 <Luke-Jr> gmaxwell: I'm not sure it is?
609 2013-09-21 21:21:49 <gmaxwell> Sipa says his current reachable count is about 3k.
610 2013-09-21 21:22:20 <Luke-Jr> gmaxwell: my non-seeds.txt data is filtered
611 2013-09-21 21:22:22 <gmaxwell> which is basically consistent with the bitnodes numbers for getversion (esp once you consider sipa's are filtered by e.g. version number and height while the bitnodes numbers are not)
612 2013-09-21 21:22:55 <Luke-Jr> http://codepad.org/1AFRurf0
613 2013-09-21 21:23:09 <gmaxwell> Luke-Jr: well, go try.. shuf -n 100 bigfile.txt and see how many of them you can getversion.
614 2013-09-21 22:43:14 <warren> huh, only 3k listening nodes?
615 2013-09-21 22:59:24 <gmaxwell> warren: after varrious healthy and reachability exclusions.