1 2013-02-07 00:13:55 <HM2> hmm
2 2013-02-07 00:24:25 <[7]> hm... so someone sneaked a bug into the sha256 reference implementation on wikipedia!
3 2013-02-07 00:24:25 <[7]> http://en.wikipedia.org/w/index.php?title=SHA-2&diff=534550697&oldid=533885753
4 2013-02-07 00:24:25 <[7]> went unnoticed for weeks apparently
5 2013-02-07 00:38:01 <jgarzik> MC1984: no new bootstrap (yet)
6 2013-02-07 00:38:10 <jgarzik> MC1984: though since 0.8rc has been tagged, it's time (finally)
7 2013-02-07 00:38:40 <MC1984> make it and lets seed
8 2013-02-07 00:50:34 <HM2> 0.8-rc1 up and running here
9 2013-02-07 00:50:56 <HM2> gave up trying to get over the bdb4 vs bdb5 issue and just binned the blockchain
10 2013-02-07 00:55:10 <jgarzik> 2013-02-07 01:54:54 ERROR: CTxMemPool::accept() : inputs already spent
11 2013-02-07 00:55:10 <jgarzik> 2013-02-07 01:54:54 Misbehaving: 74.96.240.91:8333 (0 -> 0)
12 2013-02-07 00:57:56 <MC1984> on xp the progress bar is kinda square and not ovid to fill the bar border.....
13 2013-02-07 00:58:02 <MC1984> ovoid
14 2013-02-07 00:58:33 <MC1984> soit looks a bit like an ovoid with a orange rectangle over it than a bar
15 2013-02-07 00:59:20 <MC1984> #firstwoldproblems
16 2013-02-07 01:33:57 <Luke-Jr> jgarzik: pushed some docs and simplifications
17 2013-02-07 01:34:30 <Luke-Jr> jgarzik: that shouldn't be considered misbehaviour O.o
18 2013-02-07 01:48:08 <Luke-Jr> Eligius proposal node upgraded to 0.8.0rc1
19 2013-02-07 01:48:29 <Luke-Jr> I'll discuss making it the GBT node with wizkid057
20 2013-02-07 02:10:41 <jgarzik> did 0.8.0rc1 get released, or merely tagged?
21 2013-02-07 02:14:59 <gavinandresen> jgarzik: tagged and built, but we need more gitian builders / signatures before I announce
22 2013-02-07 02:15:35 <gavinandresen> (I did upload binaries to sourceforge, too, which was probably a mistake)
23 2013-02-07 03:17:19 <BlueMatt> ;;later tell gavinandresen your 0.8.0rc1 builds are bad, they dont match mine, and specifically are compiled using gcc "4.4.2-1ubuntu4", but the latest on lucid (which should have been used) is 4.4.3-1ubuntu1
24 2013-02-07 03:17:23 <gribble> The operation succeeded.
25 2013-02-07 03:21:55 <Luke-Jr> O.o
26 2013-02-07 03:22:07 <Luke-Jr> BlueMatt: post yours? :P
27 2013-02-07 03:22:10 <BlueMatt> gavin's trying to steal our coins!
28 2013-02-07 03:22:21 <BlueMatt> Luke-Jr: I pushed the sigs, Ive already overwriten the builds....
29 2013-02-07 03:23:05 <Luke-Jr> oh, right
30 2013-02-07 03:31:59 <BlueMatt> my win32s also dont match, but I find nothing that indicates fault (and mine have been failing to match for a while now, I blame kvm vs lxc, but Ill wait for more builds to see whats up)
31 2013-02-07 03:41:32 <n9ne> Can anyone help me, or point me in the right direction for some help? I need to connect to port 25 using telnet, but I am unable and have tried many strategies, to no avail. :(
32 2013-02-07 03:42:37 <randy-waterhouse> hey anybody here familiar with the script hash stuff?
33 2013-02-07 03:49:48 <kjj> n9ne: what isn't working?
34 2013-02-07 03:50:07 <kjj> randy-waterhouse: I'm a little familiar. what do you need?
35 2013-02-07 03:50:23 <randy-waterhouse> just a few pointers I think
36 2013-02-07 03:50:26 <n9ne> @kjj it just says it keeps trying until it times out. every time.
37 2013-02-07 03:50:45 <n9ne> @kjj and my computer (a mac) used to do it just fine. Now it won't connect to any server on port 25
38 2013-02-07 03:51:06 <kjj> your ISP started blocking outgoing SMTP traffic, most likely
39 2013-02-07 03:51:23 <n9ne> Can I get around it?
40 2013-02-07 03:51:33 <kjj> try 587 instead
41 2013-02-07 03:52:04 <kjj> randy-waterhouse: pointers to what?
42 2013-02-07 03:52:14 <randy-waterhouse> kjj: e.g. at this link here http://explorer.dot-bit.org/tx/1235804 the scriptSig scriptPubKey for a tx are shown ...
43 2013-02-07 03:52:47 <randy-waterhouse> which number in the scriptSig is the <sig> and which is the <pubKey> specifically
44 2013-02-07 03:52:49 <randy-waterhouse> ?
45 2013-02-07 03:55:15 <randy-waterhouse> so assuming it is a valid tx i tried to check using cli "openssl dgst" that it was valid but to no avail ... are there a few tips to watch out for?
46 2013-02-07 03:55:33 <n9ne> @kjj 587 constantly times out as well
47 2013-02-07 03:56:35 <gmaxwell> randy-waterhouse: uh, you can't really use openssl for anything useful here??? it's not like it understands bitcoin transactions. :)
48 2013-02-07 03:56:53 <randy-waterhouse> gmaxwell: well that is not exactly true
49 2013-02-07 03:57:05 <randy-waterhouse> since botg.sh generates addresses quite correctly
50 2013-02-07 03:57:13 <randy-waterhouse> and uses openssl dgst throughout
51 2013-02-07 03:57:19 <kjj> n9ne: do you have nmap?
52 2013-02-07 03:57:55 <randy-waterhouse> i'm not doing a tx ... just hashing the pubKey to address
53 2013-02-07 03:58:21 <kjj> randy-waterhouse: I'm not sure how the namecoin explorer parses transactions, so I don't know how many layers have been unwrapped
54 2013-02-07 03:58:42 <randy-waterhouse> kjj: ok thanks
55 2013-02-07 03:59:07 <kjj> but, in the link you gave, the txout pubkey hash is http://explorer.dot-bit.org/tx/1235804
56 2013-02-07 03:59:08 <gmaxwell> randy-waterhouse: uh... a seperate program (even if its in a lame language like shell) can do things the component progras will not.
57 2013-02-07 03:59:10 <kjj> bf16fabe466f021ffbf00aeebd41e588d363579e
58 2013-02-07 03:59:34 <n9ne> @kjj No I don't. I just know that the mx server is 7577287963395599715.mail.outlook.com
59 2013-02-07 03:59:36 <randy-waterhouse> kjj: yes I figured that
60 2013-02-07 03:59:37 <gmaxwell> but hey, if you'd rather try to debate things with me than ask useful questions and accept the answers you get??? be my guest.
61 2013-02-07 03:59:46 <randy-waterhouse> fu
62 2013-02-07 04:00:05 <randy-waterhouse> i'm not debating anything ...
63 2013-02-07 04:00:14 <randy-waterhouse> which number in the scriptSig is the <sig> and which is the <pubKey> specifically?
64 2013-02-07 04:00:21 <randy-waterhouse> kjj: e.g. at this link here http://explorer.dot-bit.org/tx/1235804 the scriptSig scriptPubKey for a tx are shown ...
65 2013-02-07 04:00:24 <gmaxwell> But no matter how much you argue you're not going to get openssl to validate a bitcoin signature.
66 2013-02-07 04:00:28 <n9ne> @randy-waterhouse gmaxwell knows what he's talking about. He's literally a computer genius.
67 2013-02-07 04:00:32 <randy-waterhouse> gmaxwell: : e.g. at this link here http://explorer.dot-bit.org/tx/1235804 the scriptSig scriptPubKey for a tx are shown ...
68 2013-02-07 04:00:46 <randy-waterhouse> which number in the scriptSig is the <sig> and which is the <pubKey> specifically?
69 2013-02-07 04:00:57 <randy-waterhouse> thnx?
70 2013-02-07 04:01:16 <kjj> n9ne: 587 isn't open on that box, but 25 is, but your ISP is almost certainly filtering it. why do you need to connect to it?
71 2013-02-07 04:01:55 <kjj> randy-waterhouse: https://en.bitcoin.it/w/images/en/e/e1/TxBinaryMap.png
72 2013-02-07 04:02:05 <n9ne> @kjj it's a computer project of mine to spoof an e-mail to my university e-mail address
73 2013-02-07 04:02:18 <n9ne> using telnet
74 2013-02-07 04:02:50 <n9ne> I've spoofed before, but I have no idea how to do it if I can't connect to the mx server using telnet :(
75 2013-02-07 04:03:20 <kjj> heh. that is why ISPs block outgoing 25. try it from your university network
76 2013-02-07 04:03:38 <randy-waterhouse> kjj: so https://en.bitcoin.it/wiki/Script#Standard_Transaction_to_Bitcoin_address has another layer of encoding that you just showed I see ... thnx
77 2013-02-07 04:04:06 <gmaxwell> that explorer doesn't seem to have any way to get the actual raw transaction. And it looks like it's somewhat mangling the scriptsigs.
78 2013-02-07 04:04:07 <n9ne> @kjj I'm on my university network :(
79 2013-02-07 04:04:37 <n9ne> I can't use a university computer because they all have not enabled telnet , and it requires admin privs to do so. I am using my laptop on their network
80 2013-02-07 04:05:03 <randy-waterhouse> gmaxwell: thanks ... that's what I was wondering ... are there any tools to pull raw tx from the blockchain?
81 2013-02-07 04:05:23 <kjj> in bitcoin, yes. getrawtransaction RPC call will do it
82 2013-02-07 04:05:53 <kjj> but namecoin runs on an old fork and doesn't have it. I only point that out because your link was to a namecoin block
83 2013-02-07 04:06:18 <gmaxwell> randy-waterhouse: in bitcoin it's easy, run getrawtransaction ??? but thats namecoin and AFAIK no one maintains software for namecoin anymore.
84 2013-02-07 04:07:43 <randy-waterhouse> gmaxwell: thanks but i wasn't for a update about the status of namecoin
85 2013-02-07 04:07:51 <randy-waterhouse> s/asking
86 2013-02-07 04:08:15 <randy-waterhouse> kjj: thns again, most helpful; as usual
87 2013-02-07 04:08:25 <n9ne> Feisty D:
88 2013-02-07 04:08:36 <gmaxwell> What the hell.
89 2013-02-07 04:08:37 <kjj> heh. that guy's a bit of a dick.
90 2013-02-07 04:09:18 <gmaxwell> I think it's unfortunate that no one maintains software for namecoin anymore. I was just answering the question. :-/
91 2013-02-07 04:12:39 <n9ne> I'm trying to use "waterroof" to open a port on the network
92 2013-02-07 04:12:55 <n9ne> or to check if a port is open
93 2013-02-07 04:38:39 <gmaxwell> Some context to randy-waterhouse's credit. Apparently some confusion from who knows where had him having a prior expectation that I was some big namecoin hater. With the understanding that he was tinking that, I think his response makes a lot more sense to me now.
94 2013-02-07 04:48:49 <gjs278> namecoin is gay
95 2013-02-07 04:48:57 <gjs278> tell them to send the hate my way
96 2013-02-07 04:50:29 <wizkid057> there any way to get bitcoind to bind outbound connections to a specified ip/interface?
97 2013-02-07 04:51:28 <kjj> -listen
98 2013-02-07 04:51:55 <gmaxwell> kjj: outbound
99 2013-02-07 04:52:17 <gmaxwell> wizkid057: not AFAIK.
100 2013-02-07 04:53:54 <kjj> oh, outbound. that's rare
101 2013-02-07 04:55:58 <gmaxwell> easiest thing to do would be to use the proxy stuff for that, I guess.
102 2013-02-07 04:57:10 <wizkid057> i'll just hide it in an LXC
103 2013-02-07 04:57:46 <kjj> I think it should be a fairly easy patch if there is some need
104 2013-02-07 06:00:22 <jj_> hey does anyone know of any bitcoin mining programs written in plain c?
105 2013-02-07 06:01:23 <Luke-Jr> jj_: BFGMiner?
106 2013-02-07 06:02:14 <jj_> googling it
107 2013-02-07 06:03:05 <jj_> nice but maybe something a bit smallers
108 2013-02-07 06:03:44 <Luke-Jr> jj_: libblkmaker has a very short example, without networking
109 2013-02-07 06:10:12 <midnightmagic> jj_: .. smaller?
110 2013-02-07 06:10:49 <jj_> sorry yes smaller as in much less code to analys
111 2013-02-07 06:10:57 <midnightmagic> it's like 29MB, and 26MB of it is fpga bitstreams.
112 2013-02-07 06:11:23 <jj_> im trying to get a basic understanding of how bit coins work by reading the code
113 2013-02-07 06:11:36 <Luke-Jr> jj_: mining isn't bitcoins
114 2013-02-07 06:11:44 <Luke-Jr> mining is just SHA256 :P
115 2013-02-07 06:11:52 <midnightmagic> bitcoind has it all, including a CPU miner. reading the .cl code for the GPU miners won't tell you much.
116 2013-02-07 06:11:53 <kjj> reading the wiki might be a better use of your time
117 2013-02-07 06:12:19 <jj_> mining is the discovery of bit coin strings isnt it?
118 2013-02-07 06:12:21 <Luke-Jr> I'd recommend Satoshi's PDF first
119 2013-02-07 06:12:31 <Luke-Jr> jj_: no, there are no "bitcoin strings"
120 2013-02-07 06:12:37 <jj_> :(
121 2013-02-07 06:13:59 <jj_> i thought a block was similar to a character string
122 2013-02-07 06:15:18 <andytoshi> jj_: the PDF is at http://www.bitcoin.org/bitcoin.pdf
123 2013-02-07 06:15:23 <jj_> i mean i thought the chain of digital signatures was similar to a string
124 2013-02-07 06:15:26 <andytoshi> that will give you the high level picture and the vocabulary to ask the right questions
125 2013-02-07 06:15:38 <jj_> thanks lol
126 2013-02-07 06:18:01 <andytoshi> ;;bc,blocks
127 2013-02-07 06:18:02 <gribble> 220009
128 2013-02-07 08:48:24 <Diablo-D3> gmaxwell: programmers are assholes
129 2013-02-07 08:49:09 <Diablo-D3> the ONLY b(2,6) de brujin sequence Ive found on the intertubes gives incorrect answers
130 2013-02-07 08:49:37 <Diablo-D3> de brujin sequences are good at finding the most significant bit in numbers using very few ops
131 2013-02-07 08:49:56 <Diablo-D3> its usually faster than ops in the cpu depending on what you're doing
132 2013-02-07 13:29:12 <gavinandresen> BlueMatt: so what's the best way to update the compiler in my target-lucid-*.qcow VM images ? make_base_vm to create brand-new ones ?
133 2013-02-07 13:53:20 <gavinandresen> Any OSX folks here? I've got a really annoying problem building the 0.8.0rc1 OSX binaries
134 2013-02-07 13:54:42 <gavinandresen> Apparently, the version of std::vector in the MacOSX10.5.sdk is missing the .data() method???.
135 2013-02-07 13:57:02 <BlueMatt> gavinandresen: the weird thing is the build script starts by updating, so I dont know whats up
136 2013-02-07 13:57:09 <BlueMatt> gavinandresen: in any case, you can do it manually....
137 2013-02-07 13:57:22 <BlueMatt> export PATH=$PATH:libexec && on-target -u root
138 2013-02-07 13:57:28 <BlueMatt> then apt-get update && apt-get dist-upgrade
139 2013-02-07 13:58:14 <gavinandresen> BlueMatt: thanks
140 2013-02-07 13:59:32 <BlueMatt> oh, and if you update and your linux builds still dont match mine, go ahead and delete mine (gitian-builder seems to have issues where kvm builds dont match lxc)
141 2013-02-07 14:35:38 <comboy_> what's the best tool to get detailed info about the blockchain (like not-mine transaction details)? I'm reading that gavin's bitcointools "are becoming obsolete as we move away from using berkdb"
142 2013-02-07 14:37:02 <gavinandresen> comboy_: running 0.8.0 bitcoind with -txindex then using RPC methods like getrawtransaction/getblock will get you a long ways.
143 2013-02-07 14:37:17 <gavinandresen> comboy_: ???. depending on what information you need.
144 2013-02-07 14:37:57 <comboy_> gavinandresen: oh, this getrawtransaction sounds good, I'll give 0.8.0 a try then, thanks a lot
145 2013-02-07 14:38:04 <gavinandresen> comboy_: if you need fast access to things like "give me all transactions to/from this public key" then you'll probably want to maintain your own database customized for whatever queries you need to do
146 2013-02-07 14:38:30 <comboy_> gavinandresen: yes I'm thinking about own database with own indexes, I just want to get raw data
147 2013-02-07 14:39:11 <gavinandresen> if you just want to get the raw data in the current blockchain, getblock will do that. and -blocknotify will tell you when bitcoind gets new blocks....
148 2013-02-07 14:40:02 <comboy_> yes but on 0.7.2 I seem to be only able to get info about my own transactions, or am I wrong?
149 2013-02-07 14:41:16 <comboy_> ah, I should use getrawtransaction I suppose and I've been trying gettransaction, maybe I'll read some more docs before wasting your time, anyway thanks for hints :)
150 2013-02-07 14:41:26 <HM2> eesh
151 2013-02-07 14:41:33 <HM2> 0.8-rc1 is still a memory hog
152 2013-02-07 14:43:16 <HM2> not synced yet though hmm
153 2013-02-07 14:46:46 <gavinandresen> HM2: you're syncing on a VPS ?
154 2013-02-07 14:46:47 <btc123> anyone know who runs blockchain.info?
155 2013-02-07 14:47:06 <gavinandresen> btc123: Ben Reeves runs blockchain.info
156 2013-02-07 14:47:29 <Scrat> aka piuk on irc
157 2013-02-07 14:47:36 <btc123> thanks guys
158 2013-02-07 14:47:41 <Scrat> hasn't been online for quite some time
159 2013-02-07 14:48:50 <HM2> gavinandresen: correct. 480/500 MB ram used, plus 600 of 1GiB swap
160 2013-02-07 14:49:24 <btc123> looks like i lost my iphone blockchain wallet... when the service was down on tuesday, the wallet unpaired for the first time since i installed it (6-8 months?) and i can't remember the password.... i suppose i'm screwed?
161 2013-02-07 14:49:32 <HM2> I take it there's no heiristic to scale back any default buffers based on machine memory?
162 2013-02-07 14:49:52 <HM2> He-ur-istic
163 2013-02-07 14:50:43 <gavinandresen> HM2: no, I think all of the cache settings assume you have at least 2GB of memory.
164 2013-02-07 14:51:38 <gavinandresen> "patches welcome" if you want to futz around with making it run better on low-memory machines, but memory is pretty cheap these days.
165 2013-02-07 14:52:52 <btc123> i don't suppose anyone knows where the blockchain iphone app stores the password on the device? (it auto-connects when it loads)
166 2013-02-07 14:53:01 <gavinandresen> HM2: actually, a patch I've been thinking of writing is a RPC command to report on the size of all the various internal buffers (memory pool of unconfirmed transactions, orphan transactions, etc etc)
167 2013-02-07 14:53:15 <HM2> yeah on my machine, but VPSs tend to charge you for other things alongside ram :P
168 2013-02-07 14:53:18 <gavinandresen> HM2 : if you want a project, that might be a good one to start with
169 2013-02-07 14:53:34 <HM2> noted
170 2013-02-07 14:56:18 <Diablo-D3> everyone should upvote this: http://www.reddit.com/r/IAmA/comments/182fth/i_am_matt_taibbi_politics_writer_for_rolling/c8azhpr
171 2013-02-07 14:58:47 <sipa> gavinandresen: jeff's network message refactor would help a lot regarding memory usage
172 2013-02-07 15:00:21 <gavinandresen> sipa: great, we should get that debugged and into the 0.8+ release early
173 2013-02-07 15:01:00 <sipa> especially with many connections
174 2013-02-07 15:01:14 <HM2> 8 connections so far, been syncing for about 12 hours
175 2013-02-07 15:05:07 <HM2> fortunately it's just a little test box.
176 2013-02-07 15:21:00 <gmaxwell> 01:49 < Diablo-D3> the ONLY b(2,6) de brujin sequence Ive found on the intertubes gives incorrect answers
177 2013-02-07 15:21:56 <Diablo-D3> gmaxwell: Im writing an opencl program to mine de bruijns
178 2013-02-07 15:22:04 <gmaxwell> I wrote code for them in the opus unit tests, but it only does k,2 ones, IIRC.. though it could be modified.
179 2013-02-07 15:22:31 <Diablo-D3> I wrote a cpu one, but it takes like 30 seconds to do a million
180 2013-02-07 15:22:39 <gmaxwell> though modifying it would require reading combinatoric algebra code I wrote for one time use, good luck with that.
181 2013-02-07 15:23:02 <Diablo-D3> Im cheating, Im just checking power of two numbers
182 2013-02-07 15:23:15 <gmaxwell> http://git.xiph.org/?p=opus.git;a=blob;f=tests/test_opus_common.h;h=d1e6b452918ed0e29c8208fcb7682bf174973b49;hb=HEAD#l44
183 2013-02-07 15:23:17 <Diablo-D3> 64 attempts for every number =P
184 2013-02-07 15:23:55 <gmaxwell> (they're useful for generating test cases: e.g. if you want to test all the transitions between modes of something)
185 2013-02-07 15:29:06 <rdymac> if 0.8 use filters then none node will has the entire blockchain, so how can it be share with others if each node has only data related to its wallet?
186 2013-02-07 15:32:31 <gmaxwell> rdymac: I'd love to know what gave you the confusion that 0.8 used bloom filters.
187 2013-02-07 15:33:30 <Diablo-D3> it doesnt?
188 2013-02-07 15:33:31 <Diablo-D3> why not?
189 2013-02-07 15:33:53 <gmaxwell> Because its inapplicable to a full node.
190 2013-02-07 15:34:29 <Diablo-D3> yeah, but what about the cache code?
191 2013-02-07 15:35:04 <kuzetsa> bitcoin-qt showing a bad alloc? --> http://i48.tinypic.com/63u8te.png
192 2013-02-07 15:35:07 <rdymac> a forum post
193 2013-02-07 15:35:17 <kuzetsa> I've never seen this error before
194 2013-02-07 15:35:18 <rdymac> let me found it again
195 2013-02-07 15:36:23 <gavinandresen> kuzetsa: what version of bitcoin-qt is that?
196 2013-02-07 15:37:15 <kuzetsa> gavinandresen: 0.7.2
197 2013-02-07 15:37:18 <gavinandresen> kuzetsa: ??? and it is handling getwork or getblocktemplate requests?
198 2013-02-07 15:37:32 <kuzetsa> gavinandresen: no, it is not
199 2013-02-07 15:37:50 <gavinandresen> kuzetsa: hmmm. Any RPC requests at all ?
200 2013-02-07 15:38:11 <rdymac> http://bit.ly/UXsJ6T
201 2013-02-07 15:38:35 <gavinandresen> somebody else tracked down a bad_alloc to creating too many threads, and I believe the latest RPC code creates a thread for every RPC request is why I ask
202 2013-02-07 15:38:44 <kuzetsa> gavinandresen: not from my own LAN... could rpc requests be coming in from random bitcoin nodes?
203 2013-02-07 15:39:12 <gavinandresen> kuzetsa: no, unless you opened up your RPC port with the rpcallowip option
204 2013-02-07 15:39:28 <kuzetsa> 1 sec, checking the .conf file
205 2013-02-07 15:39:38 <sipa> Diablo-D3, gmaxwel: technically, LevelDB does use bloom codes internally
206 2013-02-07 15:39:51 <Diablo-D3> good enough for me then
207 2013-02-07 15:40:13 <kuzetsa> I don't appear to have any references to RPC anywhere in my .conf
208 2013-02-07 15:41:20 <rdymac> they talk about the 0.8 rc released binary by Gavin
209 2013-02-07 15:41:29 <gavinandresen> kuzetsa: ok, then I'm baffled. Let us know if it happens when 0.8 is out
210 2013-02-07 15:41:42 <gmaxwell> rdymac: I won't follow a bitly'd link in here. Please provide the actual forum link.
211 2013-02-07 15:41:44 <kuzetsa> ACTION nods
212 2013-02-07 15:42:22 <rdymac> sorry https://bitcointalk.org/index.php?topic=141543.0
213 2013-02-07 15:42:33 <kuzetsa> at or around june 2012 when I first used any version of the "satoshi" reference client to date, this is the first I've ever seen something like this
214 2013-02-07 15:42:36 <rdymac> manually copied from mobile
215 2013-02-07 15:43:19 <gmaxwell> rdymac: that post is about bitcoin wallet for android.
216 2013-02-07 15:43:40 <gmaxwell> The post by mike below elaborates.
217 2013-02-07 15:43:47 <rdymac> "Anyone know where there's a copy of bitcoin 0.8? Can't find it on Gavin's page anymore."
218 2013-02-07 15:44:26 <sipa> 0.8 isnot released yet
219 2013-02-07 15:44:28 <rdymac> ummm didn't read anything about Android
220 2013-02-07 15:44:31 <gmaxwell> sipa: and _you_ don't you feed the FUD. :P
221 2013-02-07 15:44:37 <jgarzik> 2013-02-07 11:01:19 StopNode()
222 2013-02-07 15:44:37 <jgarzik> 2013-02-07 11:01:19 ThreadMessageHandler exited
223 2013-02-07 15:44:37 <jgarzik> 2013-02-07 11:01:19 ThreadOpenConnections exited
224 2013-02-07 15:44:37 <jgarzik> 2013-02-07 11:01:30 ThreadDumpAddress exited
225 2013-02-07 15:44:37 <jgarzik> 2013-02-07 11:01:40 ThreadsRPCServer still running
226 2013-02-07 15:44:38 <jgarzik> 2013-02-07 11:02:01 ThreadOpenAddedConnections exited
227 2013-02-07 15:44:48 <jgarzik> git HEAD hits shutdown deadlock again
228 2013-02-07 15:44:56 <gmaxwell> rdymac: the message is posted in the android subforum because it's about the android wallet.
229 2013-02-07 15:45:03 <sipa> hmm, can you debug jgarzik?
230 2013-02-07 15:45:19 <sipa> find out which thread is waiting in which critsect?
231 2013-02-07 15:45:24 <rdymac> damn, sorry, didn't read that
232 2013-02-07 15:45:36 <gmaxwell> rdymac: s'okay.
233 2013-02-07 15:45:37 <rdymac> not a good title post
234 2013-02-07 15:45:38 <jgarzik> sipa: haven't found a reliable reproducer, so I'm limited to what I can debug without having starting gdb prior to execution
235 2013-02-07 15:45:44 <jgarzik> i.e. active introspection that's it
236 2013-02-07 15:45:56 <gmaxwell> jgarzik: you can attach gdb once it's locked and at least see where it is.
237 2013-02-07 15:45:57 <jgarzik> just have an active, stuck process
238 2013-02-07 15:46:05 <sipa> you can't attach gdb once it's stuck?
239 2013-02-07 15:46:25 <gmaxwell> ACTION bets its the one he found we could never explain
240 2013-02-07 15:46:39 <sipa> yeah, could be
241 2013-02-07 15:47:16 <jgarzik> literally never done a gdb attach before.. ok, attached, pastebin coming
242 2013-02-07 15:47:55 <jgarzik> backtrace, immediately following attach: http://pastebin.com/ENnXRVyJ
243 2013-02-07 15:48:31 <gmaxwell> that thread is not interesting, go find an interesting one.
244 2013-02-07 15:48:37 <sipa> jgarzik: info threads
245 2013-02-07 15:48:50 <jgarzik> sipa: hah, just finished that
246 2013-02-07 15:48:53 <jgarzik> info threads: http://pastebin.com/ZR8PnXQV
247 2013-02-07 15:49:25 <gmaxwell> get a bt on 5.
248 2013-02-07 15:50:03 <sipa> id e also looks suspicious
249 2013-02-07 15:50:12 <sipa> id 3
250 2013-02-07 15:50:19 <jgarzik> id 3 bt: http://pastebin.com/wDZpPLqw
251 2013-02-07 15:50:22 <Diablo-D3> what the hell gcc
252 2013-02-07 15:50:33 <Diablo-D3> format \\u2018%llu\\u2019 expects argument of type \\u2018long long unsigned int\\u2019, but argument 2 has type \\u2018uint64_t\\u2019 [-Wformat]
253 2013-02-07 15:50:52 <jgarzik> id 5 bt: http://pastebin.com/rDiFDgfC
254 2013-02-07 15:51:18 <gmaxwell> ACTION loses bet
255 2013-02-07 15:51:54 <jgarzik> id 4 bt: http://pastebin.com/SQKd6Rsq
256 2013-02-07 15:51:59 <sipa> ok, id5 is the lelveldb compaction thread - nothing to worry about
257 2013-02-07 15:53:10 <jgarzik> asic miner testing (getwork) definitely occurred during this runtime
258 2013-02-07 15:53:15 <sipa> but wth is rpc trying to receive?
259 2013-02-07 15:53:35 <Diablo-D3> WTF
260 2013-02-07 15:53:35 <jgarzik> so it is fair to say that 'getwork' was stressed quite hard
261 2013-02-07 15:53:36 <sipa> it should timeout and notice the fRequestShutdown
262 2013-02-07 15:53:52 <Diablo-D3> it lets me use %zu which is size_t aka unsigned long
263 2013-02-07 15:55:15 <sipa> ACTION afk
264 2013-02-07 15:55:24 <gavinandresen> bah, my gitian machine has suddenly decide it doesn't want to work??? "Connection timed out during banner exchange"
265 2013-02-07 16:00:54 <gavinandresen> All righty??? we need more people who can gitian-build releases. I'm thinking of trying bribery??? great idea or terrible idea?
266 2013-02-07 16:02:07 <jgarzik> as long as they are trusted :)
267 2013-02-07 16:02:35 <gavinandresen> you mean not to cheat?
268 2013-02-07 16:04:14 <jgarzik> gavinandresen: as long as there is a trusted person with a signature, compensated+untrusted backups are fine
269 2013-02-07 16:04:35 <gavinandresen> agreed.
270 2013-02-07 16:04:52 <jgarzik> gavinandresen: Just saying that 100% compensated+untrusted won't work, but maybe that is just stating the blindingly obvious ;p
271 2013-02-07 16:05:25 <gmaxwell> Encourging more work on gitian would be good... there is only so many hours of dorking around with a VM in an VM and having it not work people are willing to tolerate. :)
272 2013-02-07 16:05:46 <gmaxwell> esp becase when it fails its totally opaque.
273 2013-02-07 16:07:07 <jgarzik> sipa: any other hung-bitcoind introspection wanted?
274 2013-02-07 16:07:23 <jgarzik> I've only seen this behavior when pointing the asic miner at RPC
275 2013-02-07 16:07:47 <jgarzik> disappointing that I cannot directly solo mine though
276 2013-02-07 16:07:56 <jgarzik> ACTION had been looking forward to that
277 2013-02-07 16:08:05 <gavinandresen> gmaxwell: well, I was thinking of "1 BTC to the first 5 people to submit pull requests to gitian.sigs with valid kvm gitian signatures" -- but maybe it would be better to set a bounty for whoever makes gitian build process a lot easier???
278 2013-02-07 16:09:23 <jgarzik> why not do both?
279 2013-02-07 16:09:31 <gavinandresen> jgarzik: why not indeed.
280 2013-02-07 16:09:33 <gmaxwell> jgarzik: if the GBT support were not broken.
281 2013-02-07 16:09:36 <jgarzik> and can a sig be faked? can you just paste the last hash + GPG sign it?
282 2013-02-07 16:09:43 <gmaxwell> doh.
283 2013-02-07 16:10:10 <gavinandresen> jgarzik: I was just looking at that, if the .assert files match exactly you could cheat by copying somebody else's and CLAIMING that you built....
284 2013-02-07 16:10:20 <jgarzik> gmaxwell: ?
285 2013-02-07 16:10:49 <gmaxwell> both sounds fine except for that (that you could fake it much easier than just doing it)... but I think there naturally would be more if gitian just worked.
286 2013-02-07 16:11:08 <gmaxwell> (esp if it just worked while the host was running in KVM)
287 2013-02-07 16:12:17 <jgarzik> what if the published hash was binary output + some stirring factor (email, gpg signature fingerprint, ...)
288 2013-02-07 16:12:53 <gmaxwell> jgarzik: unfortunately cgminer doesn't really have working GBT based mining. Hopefully once code is out for the avalon luke will port the driver to bfgminer and solo mining will just work.
289 2013-02-07 16:14:05 <gmaxwell> jgarzik: I cant see a way to prevent the rebinding. Basically, you have the binaries??? thats the same product you get from building it yourself.
290 2013-02-07 16:14:34 <kinlo> hmmmz, I'm willing to look at the gitian build process, but I just need someone to hold my hand, as I do not have the time to figure everything out on my own....
291 2013-02-07 16:14:38 <gmaxwell> you could do a two-phase thing where people post H(signtures) then you cut off payouts then people post the signatures.
292 2013-02-07 16:14:55 <gmaxwell> kinlo: do you run ubuntu?
293 2013-02-07 16:15:06 <kinlo> depends
294 2013-02-07 16:15:16 <gmaxwell> I _think_ it mostly just works if you run ubuntu natively on your host and then follow the instructions.
295 2013-02-07 16:15:17 <kinlo> is your question: do you run ubuntu native?
296 2013-02-07 16:15:23 <kinlo> then no, but I can set up a vm
297 2013-02-07 16:15:32 <kinlo> but if you need me to run kvm inside ubuntu...
298 2013-02-07 16:15:59 <gmaxwell> Right. Yea, with the host in the VM this did not work for me either of the times I tried it.
299 2013-02-07 16:16:13 <HM> God i love kde
300 2013-02-07 16:16:25 <kinlo> so basicly, one needs spare hardware to be able to set up gitian
301 2013-02-07 16:16:36 <kinlo> That means reinstalling my linux desktop
302 2013-02-07 16:16:38 <HM> there're 3 bitcoin plasma widgets
303 2013-02-07 16:17:34 <sipa> jgarzik: i'm currehtly in a small ski village in the french alps, with a cell phone with mobile internet
304 2013-02-07 16:18:02 <sipa> debugging is easier after saturday :)
305 2013-02-07 16:18:08 <jgarzik> sipa: just start an Ubuntu VM on your phone then!
306 2013-02-07 16:18:09 <jgarzik> :)
307 2013-02-07 16:18:27 <kinlo> sipa: I hope without roaming :)
308 2013-02-07 16:18:28 <jgarzik> ACTION is reviewing a HUD Settlement Statement
309 2013-02-07 16:18:33 <sipa> ok, brb in a year
310 2013-02-07 16:18:44 <sipa> kinlo: roaming :)
311 2013-02-07 16:18:50 <kinlo> sipa: auch :)
312 2013-02-07 16:19:04 <kinlo> good thing ssh doesn't take too much bandwith, and you can compress it
313 2013-02-07 16:19:13 <HM> mosh!
314 2013-02-07 16:19:23 <sipa> indeed
315 2013-02-07 16:23:25 <gavinandresen> I'm almost motivated enough to take this pull https://github.com/bitcoin/bitcoin/pull/1597 , extract just the "use VirtualBox instead of kvm" part and apply it to gitian
316 2013-02-07 16:26:30 <kinlo> gavinandresen: so you are building everything on mac?
317 2013-02-07 16:26:33 <kinlo> interesting...
318 2013-02-07 16:26:58 <gavinandresen> kinlo: no, Linux and Windows are built on an Ubuntu machine
319 2013-02-07 16:27:17 <kinlo> ic
320 2013-02-07 16:27:25 <kinlo> hmmmz, maybe I can use an old laptop to install ubuntu
321 2013-02-07 16:27:48 <kinlo> and get everything going from there, the hardware is good enough to run a minimal kvm instance
322 2013-02-07 16:28:28 <gavinandresen> kinlo: that's essentially my setup, an old underpowered laptop???. but gitian is continuing to be flaky for me. sigh.
323 2013-02-07 16:29:29 <gavinandresen> ??? actually, two old, underpowered laptops (I use an old 32-bit macbook to compile the OSX binaries)
324 2013-02-07 16:30:12 <kinlo> I'm going home, I'll look at installing ubuntu on my old laptop tonight, I'll see how far I get tonight
325 2013-02-07 16:30:15 <kinlo> afk
326 2013-02-07 16:38:04 <gavinandresen> mmm??? actually I think it might not be too hard to get gitian working with VirtualBox, using VBoxManage commands???.
327 2013-02-07 16:39:50 <HM> what is gitian?
328 2013-02-07 16:40:05 <HM> hmm a software distro method, cool
329 2013-02-07 16:40:46 <gavinandresen> HM: yes, https://github.com/devrandom/gitian-builder
330 2013-02-07 16:48:00 <Luke-Jr> gavinandresen: meh, no need for bribery - I'll build it sometime today
331 2013-02-07 16:54:07 <Luke-Jr> Sleep(5000) should return in 5 seconds, right? :/
332 2013-02-07 16:55:29 <jgarzik> it's wonderfully named to invoke confusion with sleep()
333 2013-02-07 16:55:52 <Luke-Jr> jgarzik: the odd thing is, it's never returning in this case
334 2013-02-07 16:56:03 <jgarzik> maybe it's sleeping
335 2013-02-07 16:56:11 <Luke-Jr> it is :P
336 2013-02-07 16:56:15 <Luke-Jr> lots of nanosleep calls
337 2013-02-07 16:56:36 <Luke-Jr> I wonder if at all related to Linux's OOM killer malfunctioning badly
338 2013-02-07 16:57:55 <moarrr> OOM?
339 2013-02-07 16:58:25 <Luke-Jr> out-of-memory
340 2013-02-07 16:58:28 <kinlo> uhmz, quick question: which version of ubuntu should I install?
341 2013-02-07 16:58:31 <Luke-Jr> despite that it's never out-of-memory for real
342 2013-02-07 16:58:52 <kinlo> is latest lts 64bit ok?
343 2013-02-07 16:59:06 <Luke-Jr> kinlo: if I had to use Ubuntu, I'd for for the LTS
344 2013-02-07 16:59:42 <kinlo> Luke-Jr: I'm asking for running gitian
345 2013-02-07 17:00:01 <Luke-Jr> oh, you want 64-bit for that
346 2013-02-07 17:00:11 <kinlo> but version doesn't matter?
347 2013-02-07 17:00:23 <Luke-Jr> I don't *think* so
348 2013-02-07 17:00:26 <Luke-Jr> it shouldn't
349 2013-02-07 17:01:16 <kinlo> ok, downloading 12.10 lts desktop 64 bit then
350 2013-02-07 17:58:25 <Diablo-D3> heh
351 2013-02-07 17:58:31 <Diablo-D3> on the eve of asic destruction
352 2013-02-07 17:58:39 <Diablo-D3> I made DiabloMiner's GPU mining faster
353 2013-02-07 18:15:12 <kinlo> gavinandresen: is there a procedure like gitian to build mac clients?
354 2013-02-07 18:15:25 <gavinandresen> kinlo: no
355 2013-02-07 18:15:36 <kinlo> gavinandresen: so you build them manually?
356 2013-02-07 18:15:49 <gavinandresen> yes
357 2013-02-07 18:16:07 <moore> Luke-Jr, with your OOM issues are you on a multi processor ( not just multi core ) system
358 2013-02-07 18:16:10 <gavinandresen> kinlo: see doc/build_process.txt for details
359 2013-02-07 18:16:13 <kinlo> gavinandresen: so far, the gitian instructions are perfectly clear and rather easy btw...
360 2013-02-07 18:16:35 <gavinandresen> kinlo: great!
361 2013-02-07 18:16:41 <kinlo> gavinandresen: it's now building the base vm's, I'll probably have a binary build in an hour (it's an old laptop!)
362 2013-02-07 18:17:56 <kinlo> gavinandresen: so how can I help, build the binaries from time to time whenever you ask? I can do that...
363 2013-02-07 18:19:20 <gavinandresen> kinlo: yes; and clone the https://github.com/bitcoin/gitian.sigs repo, then add your gpg-signed build.assert / build.assert.sig files and submit a pull request so everybody can see "kinlo also built this and got the same result"
364 2013-02-07 18:20:20 <gavinandresen> kinlo: if you're starting from scratch, it will take you much longer than an hour to compile all dependencies and then compile bitcoin on an old laptop
365 2013-02-07 18:21:22 <kinlo> I have time
366 2013-02-07 18:21:27 <kinlo> well, my laptop has time :p
367 2013-02-07 18:22:44 <Luke-Jr> moore: I'm not even sure how to determine that. /proc/cpuinfo shows 8 cores total
368 2013-02-07 18:24:02 <moore> the "physical id" tells you I think
369 2013-02-07 18:25:41 <Luke-Jr> moore: all physical id 0, but I'm not aware of any 8-core CPUs :x
370 2013-02-07 18:26:16 <moore> it is a four core 8 thread cpu probably
371 2013-02-07 18:26:30 <Luke-Jr> aha, that sounds likely
372 2013-02-07 18:26:43 <Diablo-D3> amd's doesnt quality to be called just 8 thread
373 2013-02-07 18:27:01 <moore> I asked about the number of processes because numa causes strange memmory issues
374 2013-02-07 18:27:09 <Diablo-D3> it has 8 cores worth of integer alus, 4 cores worth of fp alus
375 2013-02-07 18:27:24 <moore> where you can run out of memory on one processer module but not the other
376 2013-02-07 18:27:36 <Diablo-D3> moore: yeah, but where are you going to find a numa box here?
377 2013-02-07 18:27:41 <moore> so you think like WTF why am I swaping with only half my memory used
378 2013-02-07 18:27:52 <Luke-Jr> moore: this system isn't even swapping..
379 2013-02-07 18:27:55 <moore> when really it bacause you have run out on a core
380 2013-02-07 18:28:07 <Luke-Jr> Mem: 24689104k total, 17410300k used, 7278804k free, 119216k buffers
381 2013-02-07 18:28:08 <Luke-Jr> Swap: 2097144k total, 1955024k used, 142120k free, 11434056k cached
382 2013-02-07 18:28:12 <Luke-Jr> 11 GB cached
383 2013-02-07 18:28:21 <Diablo-D3> Luke-Jr: ...
384 2013-02-07 18:28:22 <moore> I have tons of NUMA boxes
385 2013-02-07 18:28:30 <Diablo-D3> moore: wwwwwwwwhhhhhhhhyyyyyyyyyyyyyyyyy
386 2013-02-07 18:28:36 <moore> any multi processer box is one
387 2013-02-07 18:28:37 <Diablo-D3> thats such a waste of hardware
388 2013-02-07 18:28:41 <Diablo-D3> bullshit.
389 2013-02-07 18:28:48 <Luke-Jr> Diablo-D3: ... what?
390 2013-02-07 18:28:52 <Diablo-D3> multi-socket boxes are not numa.
391 2013-02-07 18:29:00 <moore> yes they are
392 2013-02-07 18:29:05 <Diablo-D3> nope, they are not.
393 2013-02-07 18:29:20 <moore> um I am sory to say you don't know what you are talking about
394 2013-02-07 18:29:22 <Diablo-D3> they may be numa-like when you have multiple banks
395 2013-02-07 18:29:27 <Diablo-D3> but they are not numa.
396 2013-02-07 18:29:43 <moore> you have the wrong deffinition of NUMA I think
397 2013-02-07 18:29:52 <Diablo-D3> numa means non-uniform memory architecture.
398 2013-02-07 18:29:54 <Diablo-D3> look it up.
399 2013-02-07 18:29:57 <moore> yess
400 2013-02-07 18:30:01 <midnightmagic> Diablo-D3: Are you narrowly defining NUMA again to mean something specific that exists only on the first machine it ever existed on?
401 2013-02-07 18:30:08 <Diablo-D3> midnightmagic: nope
402 2013-02-07 18:30:12 <Luke-Jr> lol
403 2013-02-07 18:30:37 <moore> and when you have non uniforme access to memory like in a multi socket system it is nonuniform
404 2013-02-07 18:30:39 <midnightmagic> Diablo-D3: Okay, this is the part where you tell us what *is* NUMA, or provide an example of a NUMA machine so we know what you're talking about. :)
405 2013-02-07 18:30:54 <Diablo-D3> moore: thats the problem, it IS uniform
406 2013-02-07 18:31:18 <moore> I suppose by some definitions
407 2013-02-07 18:31:24 <Diablo-D3> even when you have multiple banks in the same machine, its still uniform access
408 2013-02-07 18:31:30 <moore> but not the one the linux kernel uses
409 2013-02-07 18:31:55 <Diablo-D3> dont confuse core hierarchy code in the kernel with numa
410 2013-02-07 18:32:28 <Diablo-D3> situations with cache non-locality != numa
411 2013-02-07 18:32:38 <moore> when revering to numa in the linux kernel people are talking about multi socket systmems with sepreat memery controlers per socket
412 2013-02-07 18:32:53 <moore> ok
413 2013-02-07 18:32:54 <HM> Numa numa
414 2013-02-07 18:33:19 <HM> ma-ia-hii, ma-ia-huu, ma-ia-hoo
415 2013-02-07 18:33:25 <Diablo-D3> moore: except they're not.
416 2013-02-07 18:34:22 <HM> http://www.mjmwired.net/kernel/Documentation/vm/numa
417 2013-02-07 18:34:29 <HM> that's the kernels take on it
418 2013-02-07 18:34:33 <Diablo-D3> moore: numa requires that some system memory cannot be ever accessed by other cpus in the system
419 2013-02-07 18:34:44 <Diablo-D3> multi-socket machines never have that problem
420 2013-02-07 18:35:20 <moore> well that is how it is
421 2013-02-07 18:35:31 <midnightmagic> How is it non-uniform when it's inaccessible?
422 2013-02-07 18:35:34 <moore> you can copy cache lines but not address the memory directly
423 2013-02-07 18:35:46 <moore> I am done with this argument as it is pointless
424 2013-02-07 18:35:58 <Diablo-D3> midnightmagic: you require a different method of access, usually involving slow page copying over whatever external bus network you use
425 2013-02-07 18:36:27 <midnightmagic> Diablo-D3: Like HT?
426 2013-02-07 18:36:50 <Diablo-D3> midnightmagic: if you're on an AMD machine, yeah, external HT
427 2013-02-07 18:37:35 <midnightmagic> Diablo-D3: So if I have a multi-socket Opteron machine, and each CPU has to ask other CPU for their memory, when they want to access another CPU's memory, is that NUMA by your definition?
428 2013-02-07 18:37:41 <Diablo-D3> midnightmagic: nope
429 2013-02-07 18:37:50 <Diablo-D3> because you have no HT bridges to _other machines_
430 2013-02-07 18:38:14 <Diablo-D3> midnightmagic: also, memory controllers on multi-socket AMD machines k8 and up act in parallel
431 2013-02-07 18:38:25 <midnightmagic> Diablo-D3: Whereby do you come upon this definition?
432 2013-02-07 18:38:38 <Diablo-D3> midnightmagic: ? its always been defined that way
433 2013-02-07 18:38:49 <Diablo-D3> NUMA is something used to build some supercomputers
434 2013-02-07 18:39:02 <midnightmagic> http://en.wikipedia.org/wiki/Non-Uniform_Memory_Access you should fix this article then.
435 2013-02-07 18:39:10 <gavinandresen> frickin frackin gitian???..
436 2013-02-07 18:39:18 <HM> People like to put labels on systems that are always changing.
437 2013-02-07 18:39:21 <Diablo-D3> midnightmagic: nothing in that article is wrong
438 2013-02-07 18:39:21 <midnightmagic> And you might want to tell Linus that he's using the term incorrectly. :)
439 2013-02-07 18:40:42 <midnightmagic> Diablo-D3: If it takes longer for an Opteron to access its sibling's memory banks than it does to access its own, and there are separately-configured physical banks of RAM which are most-directly handled by individual sockets, would this hypothetical machine I am describing in this line satisfy your definition of NUMA?
440 2013-02-07 18:41:06 <Diablo-D3> midnightmagic: nope.
441 2013-02-07 18:41:21 <moore> Diablo-D3, are you trolling?
442 2013-02-07 18:41:53 <midnightmagic> Diablo-D3: Then how is this line from the Wikipedia article *not* incorrect "Under NUMA, a processor can access its own local memory faster than non-local memory (memory local to another processor or memory shared between processors)."
443 2013-02-07 18:42:10 <Diablo-D3> midnightmagic: because it doesnt define processor, or assume single socket machines
444 2013-02-07 18:42:19 <Diablo-D3> * or it assumes
445 2013-02-07 18:42:41 <HM> Memory access isn't uniform on single socket machines, even if you disregard all CPU caches
446 2013-02-07 18:42:48 <Diablo-D3> midnightmagic: what you described with your hypothetical machine is still local.
447 2013-02-07 18:42:48 <gavinandresen> BlueMatt: I got my laptop to do a gitian re-build. Linux build matches??? Windows doesn't
448 2013-02-07 18:43:01 <midnightmagic> Right, all it says is "local". Are you saying that the individual memory banks that each socket in a multi-socket opteron, controls with each its own memory controllers, are all local to all sockets?
449 2013-02-07 18:43:10 <Diablo-D3> midnightmagic: yes.
450 2013-02-07 18:43:33 <Diablo-D3> midnightmagic: for one, cores arent wired into memory controllers, they send standard htx messages to memory controllers
451 2013-02-07 18:43:47 <Diablo-D3> midnightmagic: all memory controllers are local to all other sockets in that machine
452 2013-02-07 18:44:46 <midnightmagic> Diablo-D3: Then the speed of access of memory is not enough to satisfy your definition, regardless of the fact that it is different depending on whether a socketed (potentially multicore) processor is accessing the memory controlled by itself or one of its sibling socketed processors.
453 2013-02-07 18:44:55 <Diablo-D3> midnightmagic: some 8 socket opteron machines MAY qualify as NUMA depending on how Linux ends up viewing it
454 2013-02-07 18:45:05 <midnightmagic> Diablo-D3: And that line in English in that wikipedia article must be corrected.
455 2013-02-07 18:45:28 <Diablo-D3> midnightmagic: you're claiming its slower to access sibling memory, its not
456 2013-02-07 18:46:08 <midnightmagic> Diablo-D3: Whereby do you come by the data you are basing your conclusion on?
457 2013-02-07 18:47:02 <Diablo-D3> midnightmagic: lets say you have a 4 socket opteron box, its not uncommon to have multiple htx connections, one to each other socket, and aggregate bandwidth of the connections exceeds total system memory banwidth
458 2013-02-07 18:47:09 <Diablo-D3> midnightmagic: years and years of reading.
459 2013-02-07 18:47:45 <midnightmagic> Diablo-D3: Say there were data which showed that memory local to a processor were quicker to access than a sibling. Would this not falsify your claim?
460 2013-02-07 18:47:52 <Diablo-D3> midnightmagic: nope
461 2013-02-07 18:47:58 <Diablo-D3> because when I say slower, I mean magnitudes
462 2013-02-07 18:48:26 <Diablo-D3> this is why dealing with NUMA is such a pain in the ass
463 2013-02-07 18:48:36 <Diablo-D3> dragging in pages from somewhere else is really fucking slow
464 2013-02-07 18:49:46 <Diablo-D3> any form of synchronization is slow, and a lot of people dont even bother with threads, they just spawn multiple processes and use async messaging
465 2013-02-07 18:49:56 <Diablo-D3> and at that point, you dont even need NUMA, just build a standard cluster
466 2013-02-07 18:49:59 <midnightmagic> I've never encountered a single definition, or use, of the term NUMA, anywhere, that had a specific threshold of latency over which a machine could be classified as NUMA, and under which it could not.
467 2013-02-07 18:50:19 <Diablo-D3> midnightmagic: because you're trying to shoehorn multisocket machines into numa
468 2013-02-07 18:50:25 <Diablo-D3> do you have one box? yes? its not numa.
469 2013-02-07 18:50:25 <midnightmagic> I would love to know what, specifically, your threshold is.
470 2013-02-07 18:51:40 <Diablo-D3> midnightmagic: like, install numa utils on linux, and run numastat
471 2013-02-07 18:51:52 <Diablo-D3> if it says you have more than one node, linux thinks its numa.
472 2013-02-07 18:53:20 <midnightmagic> Diablo-D3: From where do you achieve authority for making this claim? More specifically, who would agree with you that what Linux calls NUMA is not, in fact, NUMA?
473 2013-02-07 18:53:36 <midnightmagic> Diablo-D3: Or is this one of your pet peeves and everyone else is doing it wrong?
474 2013-02-07 18:53:50 <HM> they're all the same concept
475 2013-02-07 18:53:53 <Diablo-D3> midnightmagic: very rarely does anyone try to claim multisocket is numa
476 2013-02-07 18:53:57 <HM> what is called what doesn't matter
477 2013-02-07 18:54:00 <midnightmagic> Diablo-D3: Also, you haven't answered me how much latency qualifies for the moniker NUMA.
478 2013-02-07 18:54:17 <Diablo-D3> midnightmagic: I did answert that
479 2013-02-07 18:54:18 <HM> please stop arguing
480 2013-02-07 18:54:44 <Diablo-D3> midnightmagic: you're trying to shoehorn multisocket into numa, and I said until you have multiple boxes its not numa.
481 2013-02-07 18:55:06 <midnightmagic> You said orders of magnitude of latency initially.
482 2013-02-07 18:55:25 <Diablo-D3> yes, I said it _is_ magnitudes
483 2013-02-07 18:56:03 <Diablo-D3> its not a requirement, but until quantum physics fixes the problem of reality, I dont see how you're going to beat that
484 2013-02-07 18:56:44 <gmaxwell> Man, Bitcoin sure is great. Remember bitcoin?
485 2013-02-07 18:57:07 <Diablo-D3> midnightmagic: ever see numa opteron hardware?
486 2013-02-07 18:57:20 <HM> gmaxwell: tell me about this thing you call 'bitcoin'
487 2013-02-07 18:58:01 <Diablo-D3> midnightmagic: you have special mobos with htx slots (they physically use a pci-e x16 slot thats turned around 180), then have these dummy bridge cards that plug into them, and then you connect all your boxes to an htx switch
488 2013-02-07 18:58:02 <midnightmagic> Diablo-D3: .. I'm fairly certain AMD calls multi-socket machines NUMA. developer.amd.com/wordpress/media/2012/10/51803A_OpteronLinuxTuningGuide_SCREEN.pdf
489 2013-02-07 18:58:25 <Diablo-D3> midnightmagic: make that clickable
490 2013-02-07 18:58:29 <Diablo-D3> and it'd be very unfortunate if they did
491 2013-02-07 18:58:39 <midnightmagic> Diablo-D3: http://developer.amd.com/wordpress/media/2012/10/51803A_OpteronLinuxTuningGuide_SCREEN.pdf
492 2013-02-07 18:58:44 <midnightmagic> Unfortunate for whom?
493 2013-02-07 18:59:11 <Diablo-D3> for them
494 2013-02-07 18:59:14 <gmaxwell> midnightmagic: YHBTYHLHAND
495 2013-02-07 18:59:20 <Diablo-D3> also, nothing in there says they consider their stuff numa
496 2013-02-07 18:59:26 <Diablo-D3> whats in there is how to use numa on numa-enabled machines
497 2013-02-07 18:59:46 <midnightmagic> Diablo-D3: Actually, they are referring to 2P machines, which means 2-processor machines.
498 2013-02-07 19:00:13 <Diablo-D3> 16382 MB
499 2013-02-07 19:00:13 <Diablo-D3> available: 4 nodes (0-3)
500 2013-02-07 19:00:13 <Diablo-D3> bash> numactl --hardware
501 2013-02-07 19:00:13 <Diablo-D3> node 0 cpus: 0 1 2 3 4 5 6 7
502 2013-02-07 19:00:13 <Diablo-D3> node 0 free:
503 2013-02-07 19:00:13 <Diablo-D3> node 0 size:
504 2013-02-07 19:00:15 <Diablo-D3> 14730 MB
505 2013-02-07 19:00:17 <Diablo-D3> node 1 cpus: 8 9 10 11 12 13 14 15
506 2013-02-07 19:00:19 <Diablo-D3> node 1 size:
507 2013-02-07 19:00:21 <Diablo-D3> 16384 MB
508 2013-02-07 19:00:23 <Diablo-D3> node 1 free:
509 2013-02-07 19:00:25 <Diablo-D3> 15539 MB
510 2013-02-07 19:00:27 <Diablo-D3> node 2 cpus: 16 17 18 19 20 21 22 23
511 2013-02-07 19:00:29 <Diablo-D3> node 2 size:
512 2013-02-07 19:00:31 <Diablo-D3> 16384 MB
513 2013-02-07 19:00:35 <Diablo-D3> node 2 free:
514 2013-02-07 19:00:37 <Diablo-D3> 13467 MB
515 2013-02-07 19:00:39 <Diablo-D3> node 3 cpus: 24 25 26 27 28 29 30 31
516 2013-02-07 19:00:40 <midnightmagic> stop that. a reference will do.
517 2013-02-07 19:00:41 <Diablo-D3> node 3 size:
518 2013-02-07 19:00:43 <Diablo-D3> 16368 MB
519 2013-02-07 19:00:45 <Diablo-D3> node 3 free:
520 2013-02-07 19:00:47 <Diablo-D3> 15401 MB
521 2013-02-07 19:00:49 <Diablo-D3> 4 cores per socket, 2 sockets per machine, 4 machines.
522 2013-02-07 19:01:08 <Diablo-D3> man, pdfs can go fuck themselves
523 2013-02-07 19:01:14 <Diablo-D3> why the fuck did it do that
524 2013-02-07 19:03:26 <Diablo-D3> midnightmagic: anyhow, I think someone out there is telling people single machines qualify to be called numa
525 2013-02-07 19:03:40 <midnightmagic> Yeah, if you're serious, and you're doing a good job of pretending to be serious if you aren't, I don't see a single shred of supporting evidence that anyone else is making the same distinctions you are.
526 2013-02-07 19:03:45 <Diablo-D3> moore isnt the first person Ive met to make the claim
527 2013-02-07 19:04:07 <Diablo-D3> midnightmagic: dude, you just found a pdf from amd that further proves my position
528 2013-02-07 19:04:09 <midnightmagic> Also, that's not what the PDF says.
529 2013-02-07 19:04:36 <Diablo-D3> thats exactly what the pdf says
530 2013-02-07 19:07:15 <gmaxwell> I figure that this is more effective than telling him that he's wrong??? yet again.
531 2013-02-07 19:07:25 <phantomcircuit> wat
532 2013-02-07 19:08:25 <andytoshi> i second "wat"
533 2013-02-07 19:08:34 <HM> I'm glad I'm always wrong and know it, so nobody has to tell me :)
534 2013-02-07 19:09:28 <andytoshi> HM: eventually you'll start being right ~50% of the time, and that's when things get dangerous
535 2013-02-07 19:09:37 <andytoshi> ;)
536 2013-02-07 19:14:39 <Eliel> That's a lot of text arguing about the meaning of one word :)
537 2013-02-07 19:34:34 <moore> Eliel, actually it is four words, one acronym :)
538 2013-02-07 19:38:13 <BlueMatt> gavinandresen: wait...your linux build matches your original linux build, but windows doesnt?
539 2013-02-07 19:38:40 <gavinandresen> BlueMatt: yes.
540 2013-02-07 19:38:43 <gavinandresen> :(
541 2013-02-07 19:38:47 <BlueMatt> gavinandresen: open the bitcoin-qt help page (or wherever it is) and look for the compiler version
542 2013-02-07 19:39:01 <BlueMatt> I found it with a hex diff last night, but i thought it was in the help page
543 2013-02-07 19:39:19 <gavinandresen> compiler version of the Linux bitcoin-qt ?
544 2013-02-07 19:39:41 <BlueMatt> yea
545 2013-02-07 19:39:54 <gavinandresen> one sec, gotta start the VM
546 2013-02-07 19:40:03 <BlueMatt> you should be able to see it if you open the binary
547 2013-02-07 19:40:20 <gavinandresen> what do you mean, "open the binary" ?
548 2013-02-07 19:40:22 <Luke-Jr> gavinandresen: booting my gitian vm now..
549 2013-02-07 19:40:33 <BlueMatt> ?
550 2013-02-07 19:40:33 <BlueMatt> ie run bitcoin-qt and click help -> about
551 2013-02-07 19:40:41 <gavinandresen> ah, right
552 2013-02-07 19:40:56 <gavinandresen> I thought you meant open with a hex editor or something
553 2013-02-07 19:41:02 <BlueMatt> well that works too
554 2013-02-07 19:41:13 <BlueMatt> but thats easier if you have something to diff against
555 2013-02-07 19:41:50 <Luke-Jr> did we have any dependency changes for 0.8?
556 2013-02-07 19:41:58 <BlueMatt> dont think so
557 2013-02-07 19:42:18 <Luke-Jr> ACTION wishes gitian told up front <.<
558 2013-02-07 19:42:29 <BlueMatt> we change version numbers when that happens, no?
559 2013-02-07 19:42:38 <BlueMatt> so it does tell you up front ;)
560 2013-02-07 19:44:01 <gavinandresen> Help->About bitcoin just tells me v0.8.0rc1-beta
561 2013-02-07 19:44:08 <BlueMatt> awww, alright hex editor :)
562 2013-02-07 19:44:16 <BlueMatt> you should be able to just find gcc-4.
563 2013-02-07 19:46:28 <gavinandresen> mmm??? not working: strings -n 3 bitcoin-qt | grep gcc just gets me libgcc_s.s0.1
564 2013-02-07 19:46:47 <BlueMatt> wtf?, I saw it in your binary last night when I was searching it
565 2013-02-07 19:47:07 <HM> try "g++"
566 2013-02-07 19:47:27 <gavinandresen> g++ gets me: gggg++++
567 2013-02-07 19:47:52 <Luke-Jr> grep -i ?
568 2013-02-07 19:48:04 <gavinandresen> binaries are up on sourceforge if y'all want to try
569 2013-02-07 19:48:22 <BlueMatt> gavinandresen: capital GCC
570 2013-02-07 19:48:46 <gavinandresen> ah: GCC_3.0
571 2013-02-07 19:48:54 <BlueMatt> heh, its 2 characters after that
572 2013-02-07 19:49:07 <HM> gavinandresen: what are you trying to discover?
573 2013-02-07 19:49:20 <BlueMatt> gavinandresen: just open up your hex editor and search for GCC, youll find it 10 seconds later :)
574 2013-02-07 19:49:25 <gavinandresen> BlueMatt thinks my gitian compiled with a different version of gcc
575 2013-02-07 19:49:40 <gavinandresen> "my hex editor" ???. what hex editor comes with Lucid?
576 2013-02-07 19:49:46 <BlueMatt> ghex works well enough
577 2013-02-07 19:49:51 <BlueMatt> package may be ghex2?
578 2013-02-07 19:50:33 <gavinandresen> mmm??? emacs works....
579 2013-02-07 19:50:38 <BlueMatt> ok, that too
580 2013-02-07 19:50:50 <HM> Okteta <3
581 2013-02-07 19:51:11 <BlueMatt> lucid's default gcc (which gitian /should/ be using) is 4.4.3-1ubuntu1 http://packages.ubuntu.com/lucid/gcc
582 2013-02-07 19:51:31 <gavinandresen> BlueMatt: I see three different GCC: (Ubuntu blah blah blah)'s in the binary
583 2013-02-07 19:51:40 <BlueMatt> whats the version?
584 2013-02-07 19:52:13 <gavinandresen> 4.4.3-4ubuntu5.1 and 4.4.2-1ubuntu4 and 4.4.3-4ubuntu5
585 2013-02-07 19:52:18 <Luke-Jr> HM: :>
586 2013-02-07 19:53:26 <BlueMatt> gavinandresen: yea....umm....wtf?
587 2013-02-07 19:53:39 <BlueMatt> gavinandresen: what does dpkg -l | grep gcc tell you on the vm that gitian is using?
588 2013-02-07 19:54:40 <gavinandresen> BlueMatt: one sec, let me start THAT VM up...
589 2013-02-07 19:56:48 <BlueMatt> or...better yet, gcc -v
590 2013-02-07 19:57:31 <gavinandresen> BlueMatt: http://pastebin.com/RGJQ99zF
591 2013-02-07 19:57:47 <BlueMatt> oh, tat is right
592 2013-02-07 19:57:53 <BlueMatt> so...wtf?
593 2013-02-07 20:00:10 <Luke-Jr> gavinandresen: are you doing any of the code signing stuff for this rc?
594 2013-02-07 20:00:51 <gavinandresen> Luke-Jr: yes, it would be good to test that. But I won't code-sign until gitian is happy
595 2013-02-07 20:03:37 <kinlo> mmmz, gitial keeps failing on me. Base vm ok, building gives errors
596 2013-02-07 20:03:47 <kinlo> ./bin/gbuild:21:in `system!': failed to run on-target -u root bash < target-bin/grab-packages.sh > var/base-lucid-i386.manifest (RuntimeError)
597 2013-02-07 20:05:59 <gavinandresen> kinlo: try running: ssh -oConnectTimeout=60 -oNoHostAuthenticationForLocalhost=yes -i ./var/id_dsa -p 2223 ubuntu@localhost ls
598 2013-02-07 20:06:20 <BlueMatt> gavinandresen: I dont get it...my output has 2 small differences, one seemingly arbitrary (Im too lazy to do big dumps) and one in that segment. Now that i read it more closely, its swapping out the second and last entries...
599 2013-02-07 20:06:20 <gavinandresen> ??? and if that fails, run with -v and see if it can connect at all
600 2013-02-07 20:06:30 <BlueMatt> gavinandresen: did you use lxc or kvm?
601 2013-02-07 20:06:36 <gavinandresen> BlueMatt: kvm
602 2013-02-07 20:07:00 <BlueMatt> well...I dont get it, mine never seem to match anymore :( oh well
603 2013-02-07 20:07:08 <BlueMatt> you can just delete my sigs if you get others that match
604 2013-02-07 20:07:10 <kinlo> that works from the second time, first time seemed to timeout (I ctrl-c'ed)
605 2013-02-07 20:07:23 <kinlo> let me retry the command once more
606 2013-02-07 20:07:42 <gavinandresen> kinlo: I had trouble with the timeout on my underpowered laptop. You can modify libexec/on_target to give it a big ssh timeout
607 2013-02-07 20:17:48 <Luke-Jr> 6761ee95da86c1eb354ca7858b49061f0a39702a3d7ec6fe1b91140e329c7590
608 2013-02-07 20:18:09 <Luke-Jr> BlueMatt: you didn't build/push sigs for win32?
609 2013-02-07 20:21:22 <kinlo> gavinandresen: that seems to solve it, it is now building
610 2013-02-07 20:22:30 <gavinandresen> kinlo: tail -f var/install.log or var/build.log is very useful to see what is actually happening
611 2013-02-07 20:24:40 <kinlo> yeah, found that already...
612 2013-02-07 20:25:24 <kinlo> gavinandresen: which one do you need now? 0.8.0rc1? I'm just building 0.7.2 as a test atm
613 2013-02-07 20:26:14 <gavinandresen> kinlo: yes, 0.8.0rc1. But 0.7.2 is a great test to see if your builds match ours
614 2013-02-07 20:26:27 <kinlo> indeed
615 2013-02-07 20:57:24 <kinlo> ok, so my output file build-res.yml matches the bitcoin-build.assert from the repo. So I guess I can build 0.8rc1 now
616 2013-02-07 21:22:18 <kinlo> hmmmz, do I need to clean between gitian builds somehow? I build 0.8 but got the same hashes for 32 bit and 64 bit
617 2013-02-07 21:23:03 <Aranjedeath> okay so my weird locale issues when building bitcoind on freebsd may have actually been boost related. there was a problem with the port.
618 2013-02-07 21:23:55 <moarrr> lol i thought you said, do I need to clean between genitals for a second
619 2013-02-07 21:31:00 <BlueMatt_> unc decided, in their infinite wisdom, to block my vps, so...I may be logged into irc but I wont see that for a while.....
620 2013-02-07 23:21:38 <kinlo> gavinandresen: pull request created, but the windows sigs will be tomorrow or later
621 2013-02-07 23:22:44 <kinlo> gavinandresen: seems bluematt's sigs are kinda off, mine matches yours
622 2013-02-07 23:23:17 <gavinandresen> kinlo: best news I've had all day, thanks!
623 2013-02-07 23:23:49 <kinlo> you must have had a bad day then :)
624 2013-02-07 23:25:02 <gavinandresen> definitely not a great day, too much time spent learning about stuff I don't really want to know about.
625 2013-02-07 23:29:11 <Luke-Jr> kinlo: interesting, mine don't match Gavin's :o
626 2013-02-07 23:29:15 <Luke-Jr> ACTION pokes BlueMatt to push his win32 sig
627 2013-02-07 23:29:40 <kinlo> Luke-Jr: push your sigs then, only saw BlueMatt's and gavin's in github
628 2013-02-07 23:31:26 <Luke-Jr> kinlo: I only did win32 build
629 2013-02-07 23:31:40 <kinlo> ic
630 2013-02-07 23:32:16 <kinlo> well, I need to go to bed, and given the time I spent on the linux sigs, I cannot do those win32's today....
631 2013-02-07 23:32:18 <kinlo> afk :)
632 2013-02-07 23:32:20 <gavinandresen> win32 builds seems to be different for everybody for some reason. I wonder what changed....
633 2013-02-07 23:32:40 <Luke-Jr> :/
634 2013-02-07 23:33:04 <Luke-Jr> anyone have an old KVM-based pre-OVZ-support gitian to try with?
635 2013-02-07 23:33:14 <Luke-Jr> could be a bug in newer gitians maybe?
636 2013-02-07 23:33:17 <gavinandresen> what is OVZ?
637 2013-02-07 23:34:35 <Luke-Jr> OpenVZ - I meant LXC though
638 2013-02-07 23:34:43 <gavinandresen> ah, right...
639 2013-02-07 23:35:18 <gavinandresen> I'm uploading a code-signed one-commit-after-rc1 OSX binary, then I'm done building releases for tonight