1 2013-03-31 00:00:45 <kermit_> postgresql is cool, but its superheavyweight and lacks good backup and restore proceudres
2 2013-03-31 00:01:43 <Anduck> i thought postgresql is light
3 2013-03-31 00:01:57 <Anduck> kermit_: what do you recommend?
4 2013-03-31 00:04:03 <kermit_> well, gmaxwell banned me once for mentioning oracle
5 2013-03-31 00:04:09 <kermit_> so i wont mention it)))
6 2013-03-31 00:04:52 <Anduck> uhh
7 2013-03-31 00:04:56 <Anduck> oracle..
8 2013-03-31 00:04:59 <Anduck> how about no
9 2013-03-31 00:05:12 <Stimpy> anyone eu fait with the libbBitcoin API?
10 2013-03-31 00:05:28 <Anduck> could be nice but what theyve done with java..
11 2013-03-31 00:06:33 <kermit_> the thing is you cannot f-around with your data, this has to work 100 % of the time
12 2013-03-31 00:06:36 <kermit_> you db that is
13 2013-03-31 00:06:47 <kermit_> so filesystem corrupetion, whatever, it needs to be restorable
14 2013-03-31 00:06:52 <kermit_> or failover
15 2013-03-31 00:07:12 <kermit_> you cannot get any (and this is the simple truth) db to have this kind of protection mechanism as the big O
16 2013-03-31 00:07:47 <kermit_> with the Oracle11gR2 features,.., its a breaze to implement
17 2013-03-31 00:09:57 <Anduck> well uh
18 2013-03-31 00:10:10 <Anduck> i googled postgresql.. "Hot backups and point-in-time recovery"
19 2013-03-31 00:10:10 <kermit_> anyway, my implementation is to load the tx pieces as seperate blocks in to blobs
20 2013-03-31 00:10:31 <sipa> the full transactions?
21 2013-03-31 00:10:34 <kermit_> yeah, postgresql had "tablespaces" too, (whwhaaa)
22 2013-03-31 00:10:36 <kermit_> to funny
23 2013-03-31 00:10:44 <kermit_> yup,
24 2013-03-31 00:10:54 <sipa> ... why?
25 2013-03-31 00:11:12 <kermit_> the blob storage is stellar,.., (look up basicfile vs securefile types)
26 2013-03-31 00:11:28 <kermit_> I want to harvest the blockchain for all kinds of info, it is a general ledger
27 2013-03-31 00:11:39 <sipa> ok, carry o
28 2013-03-31 00:17:13 <kermit_> with oracle you can have Rack, there is a broker (jms) you cant imagine what kind of sweets and candy you can unleash on data
29 2013-03-31 00:17:36 <kermit_> rack (mulitple compu;s acting as one db)
30 2013-03-31 00:29:38 <kermit_> oracle has a very steep learning curve, unfortunatly
31 2013-03-31 00:29:52 <kermit_> to make a good solution will take intimate knowledge of the product
32 2013-03-31 00:30:48 <sipa> can you take this to #oraclerocks or so?
33 2013-03-31 00:35:58 <kermit_> why do people hate oracle?
34 2013-03-31 00:36:05 <kermit_> is there a rational for this?
35 2013-03-31 00:36:27 <kermit_> ferrari is just a good car, whats wrong with that?
36 2013-03-31 00:36:39 <sipa> nothing wrong; just off-topic here
37 2013-03-31 00:36:57 <kermit_> ferraris are offtopic yes
38 2013-03-31 00:37:16 <kermit_> have a good db to store your data, (Wallet, blockchain et all) dont see how this is offtopic
39 2013-03-31 00:37:24 <sipa> so are long monologues about the quality of database systems
40 2013-03-31 00:37:25 <kermit_> postgresql was on topic?
41 2013-03-31 00:37:28 <kermit_> hmmm
42 2013-03-31 00:39:07 <Anduck> how much bitcoind eats ram?
43 2013-03-31 00:39:23 <sipa> Anduck: a lot, especially when you have many connections
44 2013-03-31 00:39:29 <sipa> Anduck: though we're working on that
45 2013-03-31 00:39:30 <Anduck> f.eq. how much ram vps needs to run bitcoind?
46 2013-03-31 00:39:33 <Anduck> aight,cool!
47 2013-03-31 00:39:45 <sipa> right now, I'd suggest 1 GB
48 2013-03-31 00:40:01 <Anduck> for a mysql/psql + apache/nginx running..?
49 2013-03-31 00:40:26 <Anduck> and the site being actively used all the time
50 2013-03-31 00:40:30 <Anduck> maybe 2gb?
51 2013-03-31 00:40:42 <Anduck> i think i'll go with 3 GB
52 2013-03-31 00:40:50 <sipa> i have irssi + lighttpd + bitcoind + dnsseed + some cronjobs, on 2 GiB
53 2013-03-31 00:46:58 <HM> nginx is my preference over lighttpd
54 2013-03-31 00:47:51 <CodeShark> I oppose all software products that have unpronounceable names
55 2013-03-31 00:48:33 <Anduck> engine-x?
56 2013-03-31 00:51:30 <HM> CodeShark: right, because httpd sounds real sweet
57 2013-03-31 00:51:56 <HM> light h-tea-tea-pee-dee
58 2013-03-31 00:52:27 <kjdsfjdkfds> last block was 38 minutes ago?
59 2013-03-31 00:55:22 <muhoo> sipa: how badly does it swap when first dl'ing the chain?
60 2013-03-31 00:58:50 <bergle> wouldnt it be light - te pe de?
61 2013-03-31 01:00:00 <BlueMatt> ACTION cant stand going to bed at 4 on weekends when he has to get up at like 8 on weekdays...so he goes to bed at like 3
62 2013-03-31 01:00:00 <BlueMatt> ACTION goes to move all his data off google
63 2013-03-31 01:00:00 <BlueMatt> ahh, ok good
64 2013-03-31 01:00:00 <BlueMatt> ahh, ok well then go on coding!
65 2013-03-31 01:00:00 <BlueMatt> also, what are you doing up at 4?
66 2013-03-31 01:00:00 <BlueMatt> caught fire?
67 2013-03-31 01:00:00 <BlueMatt> dont you have work on monday or something?
68 2013-03-31 01:00:00 <BlueMatt> F%@!%&ing Java
69 2013-03-31 01:00:00 <BlueMatt> had to get fs corruption
70 2013-03-31 01:00:00 <BlueMatt> hah
71 2013-03-31 01:00:00 <BlueMatt> haha
72 2013-03-31 01:00:00 <BlueMatt> heh, alright
73 2013-03-31 01:00:00 <BlueMatt> heh, yes
74 2013-03-31 01:00:00 <BlueMatt> Im pretty sure swap over nbd qualifies as sharp, pointy objects
75 2013-03-31 01:00:00 <BlueMatt> ran perfectly with no problems!
76 2013-03-31 01:00:00 <BlueMatt> wait...this was on something you managed?
77 2013-03-31 01:00:00 <CodeShark> as if we need yet another dumbed-down C++ :p
78 2013-03-31 01:00:00 <CodeShark> does any of it involve sharp, pointy objects?
79 2013-03-31 01:00:00 <CodeShark> does it involve an open window on a highrise?
80 2013-03-31 01:00:00 <CodeShark> I've often felt the same way
81 2013-03-31 01:00:00 <CodeShark> I was close :)
82 2013-03-31 01:00:00 <muhoo> BlueMatt: amen, brother
83 2013-03-31 01:00:00 <muhoo> i tried running bitcoind on a machine with 1GB RAM. it was... submoptimal
84 2013-03-31 01:00:00 <muhoo> syncing
85 2013-03-31 01:00:00 <sipa> 04:10:39 <@sipa> i could have :)
86 2013-03-31 01:00:00 <sipa> 3 weeks data lost
87 2013-03-31 01:00:00 <sipa> also, it's only sunday morning
88 2013-03-31 01:00:00 <sipa> BlueMatt: it was at a student hosting org
89 2013-03-31 01:00:00 <sipa> BlueMatt: it was something i managed, but before my time :)
90 2013-03-31 01:00:00 <sipa> BlueMatt: i've once seen a setup with a RAID1 over localdisk + NDBdisk, managed by a script that detected which computer was first, and made that one master, and the other the slave that just exported one disk over NDB
91 2013-03-31 01:00:00 <sipa> BlueMatt: nope
92 2013-03-31 01:00:00 <sipa> bonus points if you can guess what happened
93 2013-03-31 01:00:00 <sipa> coding?
94 2013-03-31 01:00:00 <sipa> even before recent patches, i never went above ~900 MiB RES
95 2013-03-31 01:00:00 <sipa> i could have :)
96 2013-03-31 01:00:00 <sipa> inadequate monitoring?
97 2013-03-31 01:00:00 <sipa> it wasn't
98 2013-03-31 01:00:00 <sipa> muhoo: ?
99 2013-03-31 01:00:00 <sipa> muhoo: i have slow I/O, but i don't think it swaps during sync
100 2013-03-31 01:00:00 <sipa> no :)
101 2013-03-31 01:00:00 <sipa> nope, no FS corruption!
102 2013-03-31 01:00:00 <sipa> now it's closer to 300 :P
103 2013-03-31 01:00:00 <sipa> oh, now i see the confusion
104 2013-03-31 01:00:00 <sipa> one of the machines went down, and nobody noticed
105 2013-03-31 01:00:00 <sipa> only the slave fileserver was, and as it only provided an NBD device to the master, which gracefully went into degraded mode, nothing was wrong
106 2013-03-31 01:00:00 <sipa> power came back up, and the one that had been down for 3 weeks was started first
107 2013-03-31 01:00:00 <sipa> public holiday
108 2013-03-31 01:00:00 <sipa> that should have been "it could have"
109 2013-03-31 01:00:00 <sipa> then there was a power failure, and the other went down
110 2013-03-31 01:00:00 <sipa> warren: no idea
111 2013-03-31 01:00:00 <sipa> when i joined, we changed it to one decent fileserver with 2 local disks in RAID, and a dedicated backup machine
112 2013-03-31 01:00:00 <sipa> yup!
113 2013-03-31 01:00:00 <warren> ACTION wondering how nobody noticed one of the machines being down
114 2013-03-31 01:00:00 <warren> I can see how that can corrupt itself
115 2013-03-31 01:00:00 <warren> I have lots of other bad advice available.
116 2013-03-31 01:00:00 <warren> I mean, how can you not notice your hosting service is down?
117 2013-03-31 01:00:00 <warren> lost data then
118 2013-03-31 01:00:00 <warren> muhoo: make a ramdisk on a machine on the local network. export it as NBD to the machine with 1GB RAM. nbd-client, mkswap, swapon
119 2013-03-31 01:00:00 <warren> oh. storage.
120 2013-03-31 01:00:00 <warren> sipa: how was performance?
121 2013-03-31 01:00:00 <warren> sipa: Those two machines had the same IP address?
122 2013-03-31 01:00:00 <warren> that was you!? =)
123 2013-03-31 01:00:00 <warren> then the old one came up first?
124 2013-03-31 01:00:00 <warren> the script could easily read a timestamp somewhere before choosing which is master
125 2013-03-31 01:05:37 <lianj> last block an hour ago, meh
126 2013-03-31 01:06:15 <sipa> ;;bc,tslb
127 2013-03-31 01:06:16 <gribble> Error: "bc,tslb" is not a valid command.
128 2013-03-31 01:06:24 <sipa> ;;bc,tblb 1h
129 2013-03-31 01:06:26 <gribble> Error: invalid syntax (<string>, line 1)
130 2013-03-31 01:12:46 <TheButterZone> ,,bc,tslb
131 2013-03-31 01:12:47 <gribble> Error: "bc,tslb" is not a valid command.
132 2013-03-31 01:12:52 <TheButterZone> seriously
133 2013-03-31 01:13:05 <TheLordOfTime> ;;tslb
134 2013-03-31 01:13:08 <gribble> Time since last block: 58 minutes and 17 seconds
135 2013-03-31 01:18:40 <EvilPete> 63 minutes now.. damn it, wtb blocks :)
136 2013-03-31 01:19:26 <Jere_Jones> Ask and ye shall receive
137 2013-03-31 01:19:29 <Jere_Jones> ;;tslb
138 2013-03-31 01:19:33 <gribble> Time since last block: 1 minute and 5 seconds
139 2013-03-31 01:20:19 <kjdsfjdkfds> is sometimes longer blockchain creation a common event or an artifact of higher hash difficulties?
140 2013-03-31 01:20:53 <BlueMatt> block times are a statistical distribution
141 2013-03-31 01:20:58 <BlueMatt> so, yes, it happens
142 2013-03-31 01:21:36 <sipa> kjdsfjdkfds: see http://en.wikipedia.org/wiki/Exponential_distribution :)
143 2013-03-31 01:21:50 <EvilPete> Yep, its mostly luck.. but when 30-40% of hashpower is tied to one pool, bad things can happen if something stops working
144 2013-03-31 01:22:09 <kjdsfjdkfds> thanks
145 2013-03-31 01:22:38 <EvilPete> that being btcguild having 36% of global hashpower from mar 24th through today.
146 2013-03-31 01:23:15 <cornfeedhobo> does anyone know what happened to the VouchX coupons? they were working fine and now they dont seem to be working at any exchange
147 2013-03-31 01:45:37 <BlueMatt> sipa: ok, finally updated secp256k1 java thinggyu
148 2013-03-31 01:46:58 <BlueMatt> sipa: ahh, what did you do, now I get segfault....
149 2013-03-31 01:48:33 <sipa> BlueMatt: do bench and test work?
150 2013-03-31 01:48:58 <BlueMatt> yes :(
151 2013-03-31 01:49:22 <BlueMatt> but I segfault in libgmp
152 2013-03-31 01:50:03 <CodeShark> valgrind?
153 2013-03-31 01:50:23 <BlueMatt> run java in valgrind?
154 2013-03-31 01:50:38 <CodeShark> oh, nvm
155 2013-03-31 01:50:50 <CodeShark> I suppose you could :p
156 2013-03-31 01:51:03 <gmaxwell> BlueMatt: add loops to scan over all the input data and printf it. see if it moves the crash. Take the data that gets printfed at the crash and move it into the bench tool.
157 2013-03-31 01:51:24 <BlueMatt> gmaxwell: was already working on that...
158 2013-03-31 01:55:20 <jgarzik> sipa/gmaxwell: is #2409 looking good under valgrind?
159 2013-03-31 01:55:25 <sipa> jgarzik: yes
160 2013-03-31 01:55:30 <jgarzik> (network optimizations pull req)
161 2013-03-31 01:55:31 <BlueMatt> jgarzik: yes
162 2013-03-31 01:56:52 <gmaxwell> I'm not yet at as much uptime since the last bug as the maximum gap between bugs found yet. :P
163 2013-03-31 01:57:07 <gmaxwell> if someone was eager to pull it onto master I wouldn't object, but we can't release it yet.
164 2013-03-31 03:00:08 <sipa> BlueMatt: to prevent namespace clashes
165 2013-03-31 03:00:21 <BlueMatt> sipa: it probably should, yes
166 2013-03-31 03:00:30 <jgarzik> hah, too funny:
167 2013-03-31 03:00:31 <jgarzik> bash: [cd: command not found...
168 2013-03-31 03:00:31 <jgarzik> Failed to search for file: GDBus.Error:org.gtk.GDBus.UnmappedGError.Quark._pk_5ftransaction_5ferror.Code14: Invalid input passed to daemon: char '[' in text!
169 2013-03-31 03:00:31 <jgarzik> [jgarzik@bd fileserver-backup]$ [cd
170 2013-03-31 03:00:38 <jgarzik> ACTION kicks complexity
171 2013-03-31 03:00:42 <BlueMatt> lol wtf
172 2013-03-31 03:01:36 <jgarzik> some Fedora shell mapping craziness it seems. dbus has infected everything.
173 2013-03-31 03:01:57 <BlueMatt> sipa: ok
174 2013-03-31 03:02:01 <kermit_> is used visual studios "find decleration but it went sideways"
175 2013-03-31 03:02:12 <BlueMatt> f'ing dbus and systemd...
176 2013-03-31 03:02:33 <CodeShark> isn't Java supposidely designed so that segfaults are practically impossible? :)
177 2013-03-31 03:02:40 <BlueMatt> its in native code
178 2013-03-31 03:03:12 <BlueMatt> and Ive seen segfaults in java jvms plenty
179 2013-03-31 03:03:29 <sipa> BlueMatt: merged
180 2013-03-31 03:03:31 <sipa> i've known javac segfaulting :p
181 2013-03-31 03:03:35 <BlueMatt> thanks
182 2013-03-31 03:03:42 <BlueMatt> damn
183 2013-03-31 03:03:57 <jgarzik> People still use VMs?
184 2013-03-31 03:04:04 <jgarzik> ACTION looks at gcj and LLVM
185 2013-03-31 03:04:12 <sipa> (it was in 1.3, and 1.4 fixed it)
186 2013-03-31 03:06:13 <jgarzik> sipa: RE #2409. We should time out sockets that fail to drain the send buffer after a certain amount of time. Do not think the code does that now, but could be wrong?
187 2013-03-31 03:06:35 <BlueMatt> sipa: you need to mkdir obj && rm -r obj/* && git add obj so that you dont get a build error when you first check out
188 2013-03-31 03:06:54 <sipa> jgarzik: if no data gets sent or received in 60s, the socket is disconnected
189 2013-03-31 03:07:07 <sipa> jgarzik: perhaps when the send buffer is non-empty, a smaller timeout should be used
190 2013-03-31 03:07:40 <jgarzik> Pulled. Let's see wider testing.
191 2013-03-31 03:07:42 <bitcoin122> does armory work on bsd?
192 2013-03-31 03:08:17 <sipa> jgarzik: i've been thinking about an fProcessing in CNode, which is set to false when nSendSize exceeds SendBufferSize(), and gets set back to true when it's below half of that
193 2013-03-31 03:08:23 <muhoo> dbus is a disease
194 2013-03-31 03:08:43 <sipa> jgarzik: to prevent continuously calling ProcessMessage for tiny bits of data
195 2013-03-31 03:08:51 <sipa> but it won't be noticable, i guess
196 2013-03-31 03:09:39 <Belxjander> muhoo: what?
197 2013-03-31 03:11:56 <realazthat> are the block indices used by the API zero-based? such that getblockcount() returns top index +1?
198 2013-03-31 03:11:58 <jgarzik> sipa: RE timeout, 60s is sufficient. Making it smaller might just punish crappy connections. Just wanted to make sure we covered that case, in the new code.
199 2013-03-31 03:12:21 <sipa> i think that getblockcount returns the height of the tip block
200 2013-03-31 03:13:11 <realazthat> and height == index right?
201 2013-03-31 03:13:16 <sipa> indeed
202 2013-03-31 03:14:12 <jgarzik> OK, found the inactivity checking, re-reviewed, looks good
203 2013-03-31 03:14:21 <realazthat> ty
204 2013-03-31 03:14:48 <jgarzik> sipa: ACK fProcessing
205 2013-03-31 03:15:00 <jgarzik> sipa: Seems like it would prevent an abuse?
206 2013-03-31 03:15:05 <sipa> i doubt it
207 2013-03-31 03:15:27 <sipa> it's just slightly ugly now that nSendSize is accessed without cs_vSendMsg
208 2013-03-31 03:15:32 <jgarzik> true
209 2013-03-31 03:15:43 <sipa> doing that for a boolean feels nicer :)
210 2013-03-31 03:16:34 <sipa> jgarzik: ACK on #2422?
211 2013-03-31 03:17:30 <jgarzik> sipa: if this was the kernel I would use a yummy primitive like the atomic test_and_set_bit
212 2013-03-31 03:17:57 <BlueMattBot> Project Bitcoin build #259: ABORTED in 1 day 19 hr: http://jenkins.bluematt.me/job/Bitcoin/259/
213 2013-03-31 03:18:07 <BlueMatt> we should move bitcoind to a kernel module!
214 2013-03-31 03:18:10 <sipa> ewww
215 2013-03-31 03:18:25 <BlueMatt> sorry
216 2013-03-31 03:18:31 <BlueMattBot> Project Bitcoin build #260: ABORTED in 20 sec: http://jenkins.bluematt.me/job/Bitcoin/260/
217 2013-03-31 03:18:42 <BlueMatt> BlueMattBot: stop trying it aint gonna work
218 2013-03-31 03:18:42 <BlueMattBot> BlueMatt you may not issue bot commands in this chat!
219 2013-03-31 03:18:50 <BlueMatt> ugg
220 2013-03-31 03:19:12 <Belxjander> BlueMatt: keep the daemon in userland but create a kernel interface as well?
221 2013-03-31 03:19:16 <BlueMattBot> jgarzik you may not issue bot commands in this chat!
222 2013-03-31 03:19:16 <jgarzik> BlueMattBot: Say what? Chronicles of Narnia!
223 2013-03-31 03:19:22 <Belxjander> use the kernels built-in material where logical?
224 2013-03-31 03:19:32 <sipa> jgarzik: worst case is that async sending sets fProcessing back to true, but due to a race condition ProcessMessages sees a false, and stops slightly earlier than it could, picking up the next iteration
225 2013-03-31 03:19:36 <sipa> jgarzik: nothing to worry about
226 2013-03-31 03:19:52 <sipa> though a comment explaining why it's safe would be nice :)
227 2013-03-31 03:19:55 <BlueMatt> Belxjander: hell no, whole thing in kernel mode
228 2013-03-31 03:19:55 <jgarzik> /dev/bitcoin
229 2013-03-31 03:19:59 <Luke-Jr> jgarzik: my daughter is enjoying that for the first time
230 2013-03-31 03:20:53 <BlueMatt> ACTION wonders why test-and-set isnt exposed to userland
231 2013-03-31 03:20:56 <Belxjander> BlueMatt: well THAT is binding it to a single OS
232 2013-03-31 03:20:57 <BlueMatt> or is it?
233 2013-03-31 03:21:07 <BlueMatt> Belxjander: yea, BitcoinOS
234 2013-03-31 03:21:13 <BlueMatt> no userspace at all, just bitcoin
235 2013-03-31 03:21:30 <BlueMattBot> Project Bitcoin build #261: FAILURE in 20 sec: http://jenkins.bluematt.me/job/Bitcoin/261/
236 2013-03-31 03:21:30 <sipa> BlueMatt: i'm sure pthreads has something like that
237 2013-03-31 03:21:40 <BlueMatt> I thought I told you to give up
238 2013-03-31 03:21:47 <BlueMatt> probably, but does it run in one instruction?
239 2013-03-31 03:21:49 <BlueMatt> you know what, you just got yourself killed
240 2013-03-31 03:21:53 <Belxjander> well I am at the moment getting an updated python build for AmigaOS working so I can then start looking at various other things as well
241 2013-03-31 03:22:07 <jgarzik> sipa: Any other messages besides 'getdata' that might generate huge returns?
242 2013-03-31 03:22:14 <jgarzik> sipa: getheaders, addr, ?
243 2013-03-31 03:22:27 <sipa> getheaders can return up to 2000*81 bytes, afaik
244 2013-03-31 03:22:31 <jgarzik> mempool
245 2013-03-31 03:23:06 <BlueMatt> sipa: this line generates segfault for me:
246 2013-03-31 03:23:07 <BlueMatt> const ECMultConsts &c = GetECMultConsts();
247 2013-03-31 03:23:19 <sipa> BlueMatt: oh!
248 2013-03-31 03:23:38 <sipa> you need to call secp256k1_ge_start()
249 2013-03-31 03:23:50 <BlueMatt> where? lib load?
250 2013-03-31 03:23:51 <deltab> Luke-Jr: reading them in the order they were written, starting with LW&W?
251 2013-03-31 03:24:01 <sipa> i guess that means a secp256k1_start function in secp256k1.cpp
252 2013-03-31 03:24:03 <sipa> BlueMatt: let me fix
253 2013-03-31 03:24:16 <Luke-Jr> deltab: apparently :/
254 2013-03-31 03:24:27 <Luke-Jr> deltab: she started without consulting me for my expert advice
255 2013-03-31 03:26:39 <deltab> Luke-Jr: I think it's better to read the Magician's Nephew later rather than first
256 2013-03-31 03:26:47 <jgarzik> sipa/gmaxwell: I would keep an eye out for increased socket disconnects, as we now become more strict about message parsing. Refresher description in commit message... https://github.com/sipa/bitcoin/commit/607dbfdeaf7ec053d959c47c125d60c0b7e7216a
257 2013-03-31 03:26:51 <deltab> more fun that way
258 2013-03-31 03:26:53 <jgarzik> Small risk
259 2013-03-31 03:26:53 <Luke-Jr> deltab: I disagree :P
260 2013-03-31 03:26:59 <jgarzik> and they should all be bad actors
261 2013-03-31 03:27:04 <jgarzik> but still, something to watch
262 2013-03-31 03:27:28 <jgarzik> CodeShark: sipa was telling me about a main/core split, you are working on?
263 2013-03-31 03:27:55 <jgarzik> BlueMatt: I think there's a C level atomics API floating about
264 2013-03-31 03:28:06 <jgarzik> BlueMatt: but if you want to get linux-specific, that stuff does work in userspace
265 2013-03-31 03:28:10 <jgarzik> it's just asm
266 2013-03-31 03:28:18 <jgarzik> with a lock prefix, usually
267 2013-03-31 03:28:21 <BlueMatt> and gcc doesnt use it?
268 2013-03-31 03:28:25 <BlueMatt> thats boring
269 2013-03-31 03:28:31 <jgarzik> it's complicating :)
270 2013-03-31 03:28:36 <BlueMatt> true
271 2013-03-31 03:28:38 <BlueMatt> anyway...
272 2013-03-31 03:30:00 <jgarzik> in fact, I think all spinlock/semaphore primitives in the kernel should work in userland
273 2013-03-31 03:30:18 <jgarzik> cannot think of any priveleged stuff in that area
274 2013-03-31 03:30:28 <BlueMatt> ACK
275 2013-03-31 03:30:40 <jgarzik> Speaking of putting crazy shit in the kernel, I always wanted to do /dev/perl as a joke
276 2013-03-31 03:30:42 <BlueMatt> go implement it (and convince linus to merge it!)
277 2013-03-31 03:30:45 <jgarzik> and write a Perl device driver
278 2013-03-31 03:30:48 <sipa> jgarzik: #2154, i think
279 2013-03-31 03:30:58 <BlueMatt> haha
280 2013-03-31 03:31:01 <BlueMatt> /dev/brainfuck
281 2013-03-31 03:31:07 <sipa> tomorrow seems like the day
282 2013-03-31 03:31:26 <BlueMatt> for?
283 2013-03-31 03:31:27 <Belxjander> BlueMatt: you don't need to merge sources... just package it up and have it build against a working kernel tree
284 2013-03-31 03:31:45 <sipa> < BlueMatt> /dev/brainfuck
285 2013-03-31 03:31:47 <BlueMatt> fine kernel module
286 2013-03-31 03:31:49 <jgarzik> BlueMatt: April 1st
287 2013-03-31 03:31:53 <BlueMatt> ohhh...
288 2013-03-31 03:32:10 <BlueMatt> hmm...
289 2013-03-31 03:32:22 <BlueMatt> ACTION is giving tech talks on april 1st...now I need to redo the first few slides...
290 2013-03-31 03:33:03 <BlueMatt> also, if jgarzik wants to drive to chapel hill to listen to BlueMatt talk about bitcoin anonymity/pseudonymity, he is welcome to do so
291 2013-03-31 03:33:14 <jgarzik> P.S. A tale of April 1 gone wrong: a semi well known FOSS programmer made a commit to a FOSS project repository, adding an obvious backdoor to a very well known program that runs priveleged on most Linux desktops today.
292 2013-03-31 03:33:27 <BlueMatt> oh god
293 2013-03-31 03:33:36 <jgarzik> It was an April 1 joke, and he commited to side branch of something like "this is a joke"
294 2013-03-31 03:33:58 <jgarzik> However, even private branch commits go out to this project's commit mailing list. all shitstorm broke loose.
295 2013-03-31 03:34:11 <BlueMatt> hahaha
296 2013-03-31 03:34:27 <Belxjander> jgarzik: what project had that happen?
297 2013-03-31 03:35:07 <sipa> grrrr
298 2013-03-31 03:35:34 <nikitis> So are the butterfly asic's a scam or real? I feel like people are lieing to me
299 2013-03-31 03:35:41 <sipa> i spent the last half hour figuring out why all my benchmarks ran as slow as before a performance improvement i made earlier
300 2013-03-31 03:35:42 <sipa> BlueMatt: pushed a commit
301 2013-03-31 03:35:56 <sipa> going back commit by commit to see what introduced it
302 2013-03-31 03:35:58 <sipa> BlueMatt: interface in include/secp256k1.h changed a bit, but more stable now
303 2013-03-31 03:36:02 <BlueMatt> jgarzik: also, though you arent a student so you probably dont care, there is free pizza
304 2013-03-31 03:36:02 <sipa> now i notice:
305 2013-03-31 03:36:05 <sipa> $ top
306 2013-03-31 03:36:20 <sipa> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
307 2013-03-31 03:36:23 <sipa> 23477 pw 20 0 1117m 293m 22m R 99 3.7 91:53.60 chromium-browse
308 2013-03-31 03:37:10 <BlueMatt> thanks
309 2013-03-31 03:37:18 <Belxjander> sipa: chrome eating the CPU and the benchmarks are still performing the same as before?
310 2013-03-31 03:37:28 <sipa> Belxjander: nope, slower
311 2013-03-31 03:37:38 <sipa> i suppose you can call secp256k1_start() when loading the .so file
312 2013-03-31 03:37:39 <BlueMatt> so I should call secp256k1_start on library load and secp256k1_stop on unload?
313 2013-03-31 03:37:42 <BlueMatt> ok
314 2013-03-31 03:37:42 <sipa> yeah
315 2013-03-31 03:37:48 <sipa> i7 CPU that only runs at top performance when 1 single core is loaded
316 2013-03-31 03:38:03 <sipa> that call has to happen before any other calls
317 2013-03-31 03:38:05 <Belxjander> that sounds about right actually
318 2013-03-31 03:38:21 <sipa> but after that, it should be thread-safe
319 2013-03-31 03:38:30 <BlueMatt> ok, good
320 2013-03-31 03:38:34 <Belxjander> I clocked a DualCore AMD down from 2.5GHz -> 2.2GHz and it actually performed better since the cores were not as "starved" from bus access
321 2013-03-31 03:38:57 <sipa> but the resulting degradation is performance was exactly the same as the improvement i made a few commits before
322 2013-03-31 03:39:09 <sipa> so i wondered why that didn't work anymore
323 2013-03-31 03:39:30 <kermit_> jgarzik
324 2013-03-31 03:39:36 <kermit_> I have a question
325 2013-03-31 03:39:48 <kermit_> there are 2 "main" functions in picocoin
326 2013-03-31 03:39:53 <kermit_> in main.c and brd.c
327 2013-03-31 03:40:09 <jgarzik> kermit_: yes
328 2013-03-31 03:40:25 <sipa> BlueMatt: the current precomputation table is quite huge, it may take like a second to start up
329 2013-03-31 03:40:39 <kermit_> can you explain? never used two main functions before
330 2013-03-31 03:40:56 <BlueMatt> sipa: "meh"
331 2013-03-31 03:41:00 <jgarzik> kermit_: There are two programs in that directory
332 2013-03-31 03:41:08 <kermit_> oh
333 2013-03-31 03:41:36 <jgarzik> kermit_: Makefile.am in that directory shows what programs and libraries are built
334 2013-03-31 03:41:44 <kermit_> ok
335 2013-03-31 03:43:18 <CodeShark> jgarzik: yes, I had started working on that
336 2013-03-31 03:43:34 <kermit_> ok,..,brd and picocoin
337 2013-03-31 03:43:43 <jgarzik> CodeShark: looks like the early commits could be merged right now
338 2013-03-31 03:43:44 <kermit_> seem to be 2 targets
339 2013-03-31 03:43:51 <jgarzik> CodeShark: the "reduce dependence on..." stuff
340 2013-03-31 03:43:52 <Perdos> sipa: both chrome and firefox on some websites periodically do that to me and i have to kill them
341 2013-03-31 03:43:57 <Perdos> murda!
342 2013-03-31 03:44:34 <sipa> BlueMatt: weird, i must have miscalculated something; it's only 0.04s here
343 2013-03-31 03:45:00 <sipa> ok, got it, it's expected
344 2013-03-31 03:45:03 <sipa> nvm!
345 2013-03-31 03:45:25 <kermit_> jgarzik, i am not really familiar with makefile.am...,
346 2013-03-31 03:45:33 <kermit_> i think this is from autotools,?
347 2013-03-31 03:45:35 <jgarzik> kermit_: google "automake table of contents"
348 2013-03-31 03:45:45 <CodeShark> jgarzik: I was going to restructure the commits - it's been a couple months since I last really took a look at that
349 2013-03-31 03:45:55 <BlueMatt> sipa: realloc: invalid pointer
350 2013-03-31 03:46:00 <BlueMatt> in gmp
351 2013-03-31 03:46:02 <sipa> BlueMatt: bleh
352 2013-03-31 03:46:13 <jgarzik> CodeShark: those preparatory commits are ordered first, and that seems like the correct order
353 2013-03-31 03:46:23 <jgarzik> CodeShark: they are also independent of a core/main split
354 2013-03-31 03:46:43 <kermit_> jgarzik->http://www.sourceware.org/autobook/autobook/autobook_toc.html ?
355 2013-03-31 03:46:51 <CodeShark> ok, cool - I'd very much like to merge them
356 2013-03-31 03:46:51 <sipa> BlueMatt: stack trace?
357 2013-03-31 03:47:44 <CodeShark> gotta go - will be back later...
358 2013-03-31 03:48:07 <BlueMatt> sipa: in secp256k1_start...nfc where
359 2013-03-31 03:48:13 <BlueMatt> gotta backtrace manually
360 2013-03-31 03:48:53 <jgarzik> kermit_: anywhere that GNU Automake is sold :)
361 2013-03-31 03:49:28 <kermit_> what does brd do, it looks like a demon
362 2013-03-31 03:49:32 <kermit_> deamon
363 2013-03-31 03:49:43 <BlueMatt> sipa: secp256k1_fe_start
364 2013-03-31 03:49:43 <sipa> brd == block relay daemon
365 2013-03-31 03:49:52 <kermit_> aha
366 2013-03-31 03:50:14 <kermit_> what is a block relay daemon?
367 2013-03-31 03:50:52 <sipa> BlueMatt: got it
368 2013-03-31 03:51:09 <BlueMatt> secp256k1_num_set_bin
369 2013-03-31 03:51:12 <BlueMatt> ok
370 2013-03-31 03:52:25 <kermit_> brd is a service, to service queries from blockinformation
371 2013-03-31 03:52:27 <kermit_> ?
372 2013-03-31 03:53:07 <sipa> kermit_: no, it just relays blocks
373 2013-03-31 03:53:24 <kermit_> sipa what does that mean?
374 2013-03-31 03:53:48 <sipa> it receives blocks from peers, verifies them, and relays them to other peers
375 2013-03-31 03:53:56 <kermit_> aha
376 2013-03-31 03:54:27 <kermit_> does it add blocks to the blockchain?
377 2013-03-31 03:54:57 <sipa> which blockchain?
378 2013-03-31 03:55:31 <sipa> BlueMatt: should be fixed
379 2013-03-31 03:56:11 <realazthat> how long does a -reindex usually take
380 2013-03-31 03:56:19 <sipa> depends on lot on your CPU and disk
381 2013-03-31 03:56:31 <realazthat> both suck
382 2013-03-31 03:56:39 <BlueMatt> sipa: works now
383 2013-03-31 03:56:56 <sipa> realazthat: setting -dbcache higher (to 1000 or so) will improve reindex speed
384 2013-03-31 03:56:58 <BlueMatt> sipa: pushed working version
385 2013-03-31 03:57:16 <jgarzik> kermit_: brd maintains a local copy of the blockchain, yes
386 2013-03-31 03:57:18 <realazthat> sipa: can I do that in the bitcoin.conf?
387 2013-03-31 03:57:21 <sipa> realazthat: yes
388 2013-03-31 03:57:28 <sipa> realazthat: it's a number in MiB, though
389 2013-03-31 03:57:39 <kermit_> jgarzik, in main.c there is a call to "chain_set" but i see only one chain_set function and it has static scope in brd
390 2013-03-31 03:57:50 <kermit_> how is this possible it is not visable from main.c
391 2013-03-31 03:57:57 <kermit_> or am i missing something
392 2013-03-31 03:58:19 <realazthat> sipa: does it effect ram or disk usage?
393 2013-03-31 03:58:41 <jgarzik> brd.c:static void chain_set(void)
394 2013-03-31 03:58:41 <jgarzik> brd.c:\tchain_set();
395 2013-03-31 03:58:41 <jgarzik> [jgarzik@bd src]$ grep -w chain_set *.c
396 2013-03-31 03:58:41 <jgarzik> main.c:static void chain_set(void)
397 2013-03-31 03:58:59 <jgarzik> kermit_: get a grep :)
398 2013-03-31 03:59:04 <kermit_> i am working in windows
399 2013-03-31 03:59:11 <kermit_> grep would be nice yes
400 2013-03-31 03:59:11 <sipa> realazthat: RAM
401 2013-03-31 03:59:16 <realazthat> kermit_: msys, cygwin
402 2013-03-31 03:59:31 <realazthat> sipa: hehe, the ram sucks too :D
403 2013-03-31 03:59:59 <sipa> BlueMatt: doesn't __attach need to be static, or get a more unique name?
404 2013-03-31 04:18:28 <realazthat> ./bitcoind.exe getrawtransaction 212e530f9fef1dd9c5f6e9d4f5525aac783f8316521a1242cace1943bddf4729 works fine,
405 2013-03-31 04:18:44 <realazthat> but the same txid with gettransaction returns error: {"code":-5,"message":"Invalid or non-wallet transaction id"}
406 2013-03-31 04:19:02 <gmaxwell> "non-wallet transaction id"
407 2013-03-31 04:19:03 <BlueMatt> do you have -txindex
408 2013-03-31 04:19:15 <gmaxwell> BlueMatt: huh, that doesn't matter for that case.
409 2013-03-31 04:19:17 <phantomcircuit> gmaxwell wins
410 2013-03-31 04:19:29 <BlueMatt> oh, I thought txindex would let you get a non-wallet txid
411 2013-03-31 04:19:32 <gmaxwell> realazthat: what are you trying to get? a decode of the transaction?
412 2013-03-31 04:19:38 <gmaxwell> BlueMatt: not with gettransaction
413 2013-03-31 04:19:49 <BlueMatt> oh :(
414 2013-03-31 04:20:03 <gmaxwell> BlueMatt: No frown needed. getrawtransaction will let you get it.
415 2013-03-31 04:20:17 <BlueMatt> oh, ok...meh
416 2013-03-31 04:20:29 <gmaxwell> BlueMatt: watcha mehing for?
417 2013-03-31 04:20:39 <realazthat> gmaxwell: I am trying to get the tx info
418 2013-03-31 04:20:47 <BlueMatt> because its a non-intuitive api
419 2013-03-31 04:20:47 <gmaxwell> realazthat: what info?
420 2013-03-31 04:21:05 <BlueMatt> generally, we have a million rpcs and some of them could be consolidated
421 2013-03-31 04:21:12 <gmaxwell> BlueMatt: yea, but gettransaction suddenly returning transactions which aren't yours when it was previously wallet stuff would be bad.
422 2013-03-31 04:21:31 <realazthat> gmaxwell: the blockchain info
423 2013-03-31 04:21:40 <realazthat> like, hash, inputs, outputs etc.
424 2013-03-31 04:21:54 <BlueMatt> gmaxwell: yes, doesnt mean it doesnt look weird now
425 2013-03-31 04:22:07 <gmaxwell> realazthat: add a 1 to your getrawtransaction call (verbose)
426 2013-03-31 04:22:09 <realazthat> or some way to get at transaction info
427 2013-03-31 04:22:15 <realazthat> ah ok
428 2013-03-31 04:22:39 <realazthat> gmaxwell: perfect, thanks
429 2013-03-31 04:23:13 <BlueMatt> ;;bc,blocks
430 2013-03-31 04:23:13 <gribble> 228874
431 2013-03-31 04:23:34 <gmaxwell> BlueMatt: I agree it's surprising, I think all the wallet rpcs should be prefixed with wallet or something like that... I guess.
432 2013-03-31 04:24:03 <BlueMatt> hmm...looks like bitcoinj may be able to import full chain in around 3 hours to tmpfs with sigchecking everything thanks to sipa
433 2013-03-31 04:25:03 <gmaxwell> slooow. :P
434 2013-03-31 04:25:08 <BlueMatt> its also java
435 2013-03-31 04:25:11 <gmaxwell> True.
436 2013-03-31 04:25:24 <BlueMatt> what does bitcoind do if you check all sigs
437 2013-03-31 04:25:25 <gmaxwell> Fassssst (all things considered)
438 2013-03-31 04:27:13 <gmaxwell> BlueMatt: 47:21 to 210000 with openssl, 37:10 with the old HAL patch which was only 20% faster than openssl (vs sipa's code which is more like 4x+ faster)
439 2013-03-31 04:27:37 <gmaxwell> (thats with checkpoints off, I don't have numbers newer than that right now)
440 2013-03-31 04:28:07 <BlueMatt> ok, well that is kinda in line with what I'd expect from java then
441 2013-03-31 04:28:18 <BlueMatt> ~3x as slow, though to tmpfs
442 2013-03-31 04:28:27 <gmaxwell> yea, mine was on a SSD.
443 2013-03-31 04:32:42 <MC-Droid> i learn there was once a FVN implementation for android
444 2013-03-31 04:32:53 <MC-Droid> i want to try it lol
445 2013-03-31 04:33:59 <BlueMatt> ~12 hours in, 270 connections, 365m RES, 789m VIRT
446 2013-03-31 04:34:40 <BlueMatt> :D
447 2013-03-31 04:39:12 <Scrat> does the automatic wallet flush lock the entire bitcoind?
448 2013-03-31 04:42:28 <gmaxwell> Scrat: because it basically shuts down and closes the wallet database in order to flush it out.... and other parts of the code expect to access it
449 2013-03-31 04:43:08 <Scrat> takes 250 ms - 9 sec (varies wildly) with a wallet of 2100 keys
450 2013-03-31 04:45:23 <sipa> realazthat: you need getrawtransaction, and enabled -txindex
451 2013-03-31 04:48:08 <BlueMatt> Imported 228535 blocks.
452 2013-03-31 04:48:08 <BlueMatt> real\t100m31.443s
453 2013-03-31 04:48:18 <BlueMatt> so...a "bit" less than 3 hours
454 2013-03-31 04:48:24 <BlueMatt> I think my calculation was a bit wrong
455 2013-03-31 04:48:37 <BlueMatt> thanks sipa!
456 2013-03-31 04:49:15 <phantomcircuit> ;;bc,blocks
457 2013-03-31 04:49:16 <gribble> 228876
458 2013-03-31 04:49:26 <sipa> BlueMatt: which num/field implementation?
459 2013-03-31 04:49:37 <BlueMatt> Im importing from .bitcoin that hasnt been synced in a bit
460 2013-03-31 04:49:39 <BlueMatt> gmp non-asm
461 2013-03-31 04:50:00 <BlueMatt> sipa: seriously? its now 9 in the morning...
462 2013-03-31 04:50:25 <EvilPete> Oh this is funny.. my bitcoind doesn't recognize itself and is trying to connect back to itself over and over..
463 2013-03-31 04:50:32 <sipa> BlueMatt: yeah, i figure i'd try to fix by biorythm a bit :)
464 2013-03-31 04:50:36 <gmaxwell> BlueMatt: yea thats more reasonable.
465 2013-03-31 04:51:36 <BlueMatt> sipa: lol
466 2013-03-31 04:51:48 <Scrat> Flushed 64 addresses to peers.dat 14305ms
467 2013-03-31 04:51:54 <BlueMatt> wtf?
468 2013-03-31 04:51:54 <Scrat> why oh why
469 2013-03-31 04:51:59 <BlueMatt> valgrind?
470 2013-03-31 04:52:00 <Scrat> load average < 2
471 2013-03-31 04:52:19 <Scrat> iowait < 10%
472 2013-03-31 04:52:27 <sipa> Scrat: is your filesystem on a network drive mounted over RFC 1149?
473 2013-03-31 04:52:35 <Scrat> sipa: lol
474 2013-03-31 04:52:38 <gmaxwell> Scrat: what crazy thing have you done with your wallet? how many transactions are in ot?
475 2013-03-31 04:52:38 <gribble> 228878
476 2013-03-31 04:52:38 <phantomcircuit> ;;bc,blocks
477 2013-03-31 04:53:20 <gmaxwell> Scrat: somethinglike .. bitcoind listtransactions "*" 100000 | grep txid | wc -l
478 2013-03-31 04:53:53 <BlueMatt> ok, seriously, Im giving a talk on april 1st and I need to start with something good...any suggestions?
479 2013-03-31 04:53:59 <Scrat> gmaxwell: not many, 210
480 2013-03-31 04:55:20 <gmaxwell> BlueMatt: whats the subject matter of the talk?
481 2013-03-31 04:55:30 <BlueMatt> bitcoin/tor anonymity
482 2013-03-31 04:55:41 <gmaxwell> BlueMatt: Give the talk in german.
483 2013-03-31 04:56:16 <BlueMatt> hah
484 2013-03-31 04:56:18 <sipa> BlueMatt: put the logo of the CIA on your first slide, with a text that you're completely and totally hidden from all authorities when using Tor
485 2013-03-31 04:57:09 <Scrat> gmaxwell: well I don't know, does bitcoind do anything weird I/O wise? no other daemon I'm running has this kind of variance
486 2013-03-31 04:57:28 <gmaxwell> Scrat: the DB flushing involve sync writes. Do you have a weird filesystem?
487 2013-03-31 04:57:31 <Scrat> drive has no errors
488 2013-03-31 04:57:46 <Scrat> ext4 on mdadm raid 1
489 2013-03-31 04:58:11 <sipa> writing to peers.dat doesn't need any DB interaction
490 2013-03-31 04:58:22 <sipa> i figure something was locking addrman
491 2013-03-31 04:58:33 <gmaxwell> BlueMatt: you could browse to an underground market which features copies of the classes test solutions for sale. The problem with that is that the screen is hard to read so the joke might be missed.
492 2013-03-31 04:58:35 <sipa> so the write blocked retrieving the data from addrman
493 2013-03-31 04:58:39 <Scrat> my keypool is 1000
494 2013-03-31 04:58:47 <gmaxwell> oh I thought he was saying the wallet took 14 seconds.
495 2013-03-31 04:58:50 <sipa> Scrat: all irrelevant
496 2013-03-31 04:59:01 <gmaxwell> I missed that it was peers.dat! crazy.
497 2013-03-31 04:59:10 <gmaxwell> That does no special filesystem anything and doesn't touch the wallet.
498 2013-03-31 04:59:24 <gmaxwell> ACTION looks on scrat's claims of a non-failing disk with skepticism.
499 2013-03-31 04:59:46 <sipa> i think something locking addrman.cs is more likely than a disk problem
500 2013-03-31 05:00:05 <Scrat> maybe an addrwoman.cs?
501 2013-03-31 05:00:20 <gmaxwell> I don't think we do anything blocking, like opening a new connection, while holding the lock though AFAIK
502 2013-03-31 05:00:22 <Scrat> ACTION lame joke
503 2013-03-31 05:00:29 <sipa> Scrat: we don't want nondeterminism there
504 2013-03-31 05:00:53 <gmaxwell> ACTION feels guilty for chuckling at sipa's sexist humor
505 2013-03-31 05:01:20 <sipa> ACTION breakfast!
506 2013-03-31 05:01:46 <BlueMatt> lol
507 2013-03-31 05:02:13 <BlueMatt> gmaxwell: hmmm...I wonder what rfcs we'll get this year
508 2013-03-31 05:03:26 <MC1984> heh
509 2013-03-31 05:03:47 <sydna> so, bitcoin transactions rely on the security of their private keys.
510 2013-03-31 05:03:58 <sydna> then why are so many libraries generating them in stupid ways?
511 2013-03-31 05:04:03 <muhoo> BlueMatt: announce that all the devs and bitcoin has been bought by visa
512 2013-03-31 05:04:10 <sydna> I've seen at least three PHP libraries that use mt_rand()
513 2013-03-31 05:04:24 <Scrat> sydna: because CSPRNGs are hard
514 2013-03-31 05:04:32 <Scrat> sydna: ok in this case stupidity
515 2013-03-31 05:04:40 <gwillen> sydna: you have seen php libraries generating bitcoin keys?
516 2013-03-31 05:04:51 <sydna> gwillen: yuh.
517 2013-03-31 05:05:14 <sydna> Scrat: easier to use openssl_random_pseudo_bytes() though, that's cryptographically secure
518 2013-03-31 05:05:17 <muhoo> BlueMatt: and visa will buy the complete market cap of bitcoins, from now on, bitcoin will be known as VisaCoin
519 2013-03-31 05:05:42 <gwillen> sydna: I can't picture why someone would be doing that
520 2013-03-31 05:05:51 <gwillen> sydna: can you shed light?
521 2013-03-31 05:06:04 <BlueMatt> muhoo: mmm, dunno, maybe discuss all the backdoor in bitcoin or something
522 2013-03-31 05:06:09 <BlueMatt> s
523 2013-03-31 05:07:10 <BlueMatt> announce a new standard for bitcoin over RFC 5514
524 2013-03-31 05:07:11 <sydna> gwillen: just payment processing scripts in this case. I've made mention of it to their developers.
525 2013-03-31 05:07:28 <gwillen> sydna: but presumably they are using bitcoind to talk to the network; why not let bitcoind generate addresses?
526 2013-03-31 05:07:33 <gmaxwell> BlueMatt: they bypass the regular RFC editorial queue, so they're actually not known until published.
527 2013-03-31 05:07:48 <gwillen> oh god
528 2013-03-31 05:07:50 <gwillen> it's that time of year
529 2013-03-31 05:07:53 <sydna> gwillen: one does, the other ones are talking to APIs like blockchain.info
530 2013-03-31 05:07:55 <gwillen> ACTION had almost managed to forget.
531 2013-03-31 05:08:02 <gwillen> sydna: ahh, *nod*
532 2013-03-31 05:08:07 <Scrat> gwillen: there are some php scripts that generate addresses themselves and use a 3rd party api to track incoming payments
533 2013-03-31 05:08:18 <gwillen> this tempts me
534 2013-03-31 05:08:20 <Scrat> sydna damn you
535 2013-03-31 05:08:22 <gwillen> but not enough to actually do it
536 2013-03-31 05:08:28 <BlueMatt> gmaxwell: well if there is a fun IP-over-Shit standard, Ill probably talk about how Bitcoin is the first to adopt IP-over-Shit
537 2013-03-31 05:08:30 <gwillen> to look through the blockchain for keys generated with MT
538 2013-03-31 05:08:34 <gwillen> and try to crack them ;-)
539 2013-03-31 05:08:38 <gwillen> ACTION wonder if anybody will
540 2013-03-31 05:08:50 <gwillen> it's probably not even hard.
541 2013-03-31 05:08:50 <sydna> you'd probably have some luck, mt_rand is very predictable
542 2013-03-31 05:08:55 <gwillen> not only that
543 2013-03-31 05:09:01 <Scrat> gwillen: only doable if you know one (or a few) of them
544 2013-03-31 05:09:02 <gwillen> you probably know the exact time it was called
545 2013-03-31 05:09:10 <Scrat> need 600 bytes of MT afaik
546 2013-03-31 05:09:12 <gmaxwell> Scrat: there are things using mt_rand() ... man, I went and checked some JS implementations and I wasn't super happy with what I found I didn't see any of that!
547 2013-03-31 05:09:19 <Scrat> to calculate state
548 2013-03-31 05:09:19 <sydna> most of them just do mt_rand(1,16), turn it into hex, and concatenate it multiple times
549 2013-03-31 05:09:30 <gmaxwell> BlueMatt: well there is the IP over social networks from a couple years ago.
550 2013-03-31 05:09:46 <gmaxwell> BlueMatt: if you'd like you can use petertodd's stuff to run bitcoin over twitter.
551 2013-03-31 05:09:47 <BlueMatt> yea...<BlueMatt> announce a new standard for bitcoin over RFC 5514
552 2013-03-31 05:09:57 <BlueMatt> thats the one
553 2013-03-31 05:10:03 <sipa> ACTION actually used that
554 2013-03-31 05:10:07 <BlueMatt> same
555 2013-03-31 05:10:17 <BlueMatt> petertodd's thing?
556 2013-03-31 05:10:22 <sipa> no, RFC 5514
557 2013-03-31 05:10:41 <gwillen> Scrat: well, you need either 600 bytes or you need the seed
558 2013-03-31 05:10:42 <BlueMatt> sorry, yes, Ive used 5514, I meant to ask what gmaxwell was referring to
559 2013-03-31 05:10:58 <gwillen> Scrat: and if you can figure out what time the app first called it
560 2013-03-31 05:10:59 <gmaxwell> BlueMatt: some airlines are allowing access to ebay on flights for free.. contemplated setting up IP over Ebay.
561 2013-03-31 05:11:04 <gwillen> Scrat: you have part of the seed
562 2013-03-31 05:11:23 <gwillen> gmaxwell: oh dear lord.
563 2013-03-31 05:11:27 <sydna> if they expose mt_rand() variables in any part of their site, they're in trouble essentially
564 2013-03-31 05:11:28 <sipa> gmaxwell: wouldn't IP-over-DNS be easier?
565 2013-03-31 05:11:33 <gwillen> gmaxwell: just use ... yeah, what sipa said.
566 2013-03-31 05:11:37 <BlueMatt> gmaxwell: many of them allow google access, so you can do IP-over-Googleapps pretty trivially
567 2013-03-31 05:11:40 <BlueMatt> though its not new
568 2013-03-31 05:11:46 <muhoo> i've seen a demo of someone downloading IIRC a movie torrent over 5514 or something
569 2013-03-31 05:11:54 <gmaxwell> sydna: not even that mt rand is a fine rng for things but its not cryptographically strong...
570 2013-03-31 05:11:57 <BlueMatt> sipa: IP-over-DNS on airplanes is sloooowww (I use it all the time, you can get IRC, thats about it)
571 2013-03-31 05:11:58 <gwillen> you can also just set your UA to be something mobile-looking
572 2013-03-31 05:12:00 <gwillen> and pay half price
573 2013-03-31 05:12:01 <sydna> I relied on Iodine (IP over DNS) for my internet connection for a year.
574 2013-03-31 05:12:04 <gwillen> for for full speed
575 2013-03-31 05:12:18 <sipa> BlueMatt: i've never been on an airplane that had internet access in the first place
576 2013-03-31 05:12:46 <BlueMatt> sipa: North Carolina <-> California's 5-hour has internet usually
577 2013-03-31 05:12:48 <sipa> then again, for these 1h flights within europe, it's not that urgent either :)
578 2013-03-31 05:12:50 <muhoo> i just discovered that t-mobile will "block tethering" by inspecting your user-agent, and capture-portal'ing you if you are using windoze
579 2013-03-31 05:13:00 <sydna> gmaxwell: some of the implementations using mt_rand are even worse than the RNG itself.
580 2013-03-31 05:13:10 <BlueMatt> ACTION is waiting for the international flights with internet
581 2013-03-31 05:13:20 <BlueMatt> muhoo: this is why you use a vpn
582 2013-03-31 05:13:21 <sipa> xkcd #221 comes to mind
583 2013-03-31 05:13:22 <Scrat> gmaxwell: what's funnier is that JS implementations are up to the interpreter. there are quite a few cases of JS engine prng implementations that provided less-than-random outputs
584 2013-03-31 05:13:35 <sydna> gmaxwell: for one system I looked over, the encryption keys were "randomly generated" using sha1(mg_rand())
585 2013-03-31 05:13:36 <muhoo> BlueMatt: heh, i was just typing that: add to my reasons for VPNing
586 2013-03-31 05:13:43 <gmaxwell> sydna: also, if people sign using a poor rng for the signing nonce the key is effectively disclosed.
587 2013-03-31 05:13:51 <BlueMatt> ACTION wonders if anyone would notice if I had a screenshot of bitcoin that had a xkcd 221-style function in it
588 2013-03-31 05:14:27 <sydna> gmaxwell: hm. does that mean that some of those javascript signing tools might be exposing keys too?
589 2013-03-31 05:14:52 <sydna> I know for a fact most of them don't use the CSRNG functions.
590 2013-03-31 05:15:00 <gmaxwell> Scrat: the ones out there I've seen use window.crypto.getRandomValues ... but then have dumb stuff if its not available.
591 2013-03-31 05:15:13 <EvilPete> I have a bitcoin-related question. "Terracoin" uses 2 minute block interval and retargets every 30 blocks. Imagine the v1/v2 enforcement a while back. There was no way for v1 miners to establish a fork because 2016 is a long way to go to regarget with just a few percent of hashpower. Am I right in assuming the high speed retargeting makes accidental forking easy?
592 2013-03-31 05:15:13 <gmaxwell> scrat: like randomly polling the pointers position and then surrendering. :(
593 2013-03-31 05:15:27 <gwillen> muhoo: the interesting thing is, tmo only does that in some locations
594 2013-03-31 05:15:35 <gwillen> muhoo: when I was in pittsburgh they never did it, but when I visited seattle they did
595 2013-03-31 05:16:00 <gwillen> I think they may have just been experimenting with it back then
596 2013-03-31 05:16:00 <Scrat> gwillen: looked at the source, seed is based on time (in milliseconds) and pid
597 2013-03-31 05:16:02 <Scrat> quite bad
598 2013-03-31 05:16:05 <gmaxwell> EvilPete: there were some pretty fun attacks against alts with fasted windowed difficulty, e.g. sc 1.0.
599 2013-03-31 05:16:06 <gwillen> yeah
600 2013-03-31 05:16:07 <EvilPete> (I used terracoin as an example, I mostly want to clear up the concept in my head about why 2016 is a good idea)
601 2013-03-31 05:16:21 <muhoo> bitaddres.org appears to use window.SecureRandom
602 2013-03-31 05:16:42 <sydna> muhoo: what's that?
603 2013-03-31 05:16:59 <EvilPete> muhoo: it says "secure", what more can you ask for?
604 2013-03-31 05:17:02 <gwillen> Scrat: PID is probably 16 bits, time you can probably guess down to a handful of bits
605 2013-03-31 05:17:08 <realazthat> how do I get the block reward for a particular block height?
606 2013-03-31 05:17:24 <gmaxwell> muhoo: qustion is what it does when its not there. some js implementations just have random() which is an insecure LCG in most implementations and in some browsers always returns 0.
607 2013-03-31 05:17:24 <sydna> realazthat: blockexplorer.com has an API call for that
608 2013-03-31 05:17:37 <Scrat> muhoo: used a homebrew RC4 RNG last I checked
609 2013-03-31 05:17:38 <sydna> muhoo: window.SecureRandom is their own implimentation, not a CSRNG
610 2013-03-31 05:17:43 <realazthat> sydna: I am trying to implement it myself
611 2013-03-31 05:18:09 <sipa> realazthat: 50.00000000 / 2**(int(height / 210000))
612 2013-03-31 05:18:29 <realazthat> sipa: ah ty
613 2013-03-31 05:18:45 <realazthat> should I worry about different chains having a different policy for that?
614 2013-03-31 05:19:07 <sipa> if you want to support multiple chains, yes :)
615 2013-03-31 05:19:10 <sipa> otherwise no
616 2013-03-31 05:19:32 <sydna> gmaxwell: Scrat: that window.SecureRandom function bitaddress.org uses looks a little fragile
617 2013-03-31 05:19:35 <realazthat> ie. is there a way to retreive that via the rpc api?
618 2013-03-31 05:19:44 <sipa> realazthat: not really
619 2013-03-31 05:19:58 <gmaxwell> sydna: thats the one that polls the pointer position in a busy loop?
620 2013-03-31 05:20:03 <sipa> realazthat: you can look at a block's coinbase outputs, but that includes fees
621 2013-03-31 05:20:19 <muhoo> i don't quite understand it, it looks like it uses ARCFOUR
622 2013-03-31 05:20:21 <realazthat> right, I've been grapling with the math trying to separate it
623 2013-03-31 05:20:27 <realazthat> but I don't think there is a way
624 2013-03-31 05:20:29 <sydna> gmaxwell: polls the position, uses window.Crypto, or falls back silently to Math.random()
625 2013-03-31 05:20:37 <realazthat> ie. separating the fees/reward
626 2013-03-31 05:20:38 <sipa> realazthat: you can if you look at the inputs too
627 2013-03-31 05:20:51 <realazthat> oh can I
628 2013-03-31 05:20:52 <sipa> realazthat: but don't do that :)
629 2013-03-31 05:20:57 <sipa> that will be slow
630 2013-03-31 05:20:58 <gmaxwell> yea thats what I meant by not being happy. means that e.g. in IE5 you'll be screwed.
631 2013-03-31 05:21:02 <realazthat> I am looking at the inputs already hehe
632 2013-03-31 05:21:05 <gmaxwell> but then again maybe it doesn't work in IE5.
633 2013-03-31 05:21:09 <Scrat> it feeds the mouse into an RC4 cipher
634 2013-03-31 05:21:14 <realazthat> ok
635 2013-03-31 05:21:24 <BlueMatt> ok, lets see if anyone notices the return 4; in GetRand screenshot
636 2013-03-31 05:21:29 <sipa> realazthat: well then you can calculate each non-coinbase transaction's fee by computing the different between outputs and inputs
637 2013-03-31 05:21:33 <gmaxwell> Scrat: yea thats not really all that useful. Assume the mouse is still there are only e.g. 1024x768 possible locations.
638 2013-03-31 05:21:46 <sipa> realazthat: and subtrack the sum of all those fees from the sum of the coinbase outputs
639 2013-03-31 05:21:48 <gmaxwell> BlueMatt: lol!
640 2013-03-31 05:22:15 <realazthat> sipa: yeah makes sense, but for now I'll back off that, and make that optional later
641 2013-03-31 05:22:26 <gmaxwell> BlueMatt: you could buy something online via bitcoin and have it instantly delivered to the room. :P
642 2013-03-31 05:22:31 <sipa> realazthat: by the way, that is not the actual subsidy
643 2013-03-31 05:22:38 <sipa> realazthat: but only the claimed part of it
644 2013-03-31 05:22:42 <sipa> realazthat: which in theory can be lower
645 2013-03-31 05:22:51 <realazthat> right
646 2013-03-31 05:22:53 <sydna> gmaxwell: there's probably some use in it all, but I'm not entirely certain that their whitening is well designed. at any rate, there's certainly not as much entropy as a standard address.
647 2013-03-31 05:23:04 <BlueMatt> gmaxwell: only if jgarzik is willing to drive over and deliver
648 2013-03-31 05:23:42 <muhoo> Scrat: sydna : yep, it's this https://www.refheap.com/paste/13141
649 2013-03-31 05:23:45 <Scrat> sydna: if people use it for microtransactions then it's fine (same goes for anything javascript based)
650 2013-03-31 05:23:58 <gmaxwell> sydna: what I'm saying is that if window.Crypto not available and the mouse is still.. Math.random probably only has 31 to zero bits of entropy.
651 2013-03-31 05:24:06 <muhoo> not cryptographically-secure :-/
652 2013-03-31 05:24:07 <MC1984> A latency of several hours has an impact on the transport protocols.
653 2013-03-31 05:24:08 <MC1984> UDP SHOULD be used, and TCP SHOULD NOT be used.
654 2013-03-31 05:24:12 <MC1984> is....is this serious?
655 2013-03-31 05:24:33 <sipa> MC1984: is that from the avian carriers RFC?
656 2013-03-31 05:24:46 <MC1984> internet over facebook thing
657 2013-03-31 05:24:50 <gmaxwell> MC1984: tcp can't be used when the latency is over 2 minutes.
658 2013-03-31 05:24:56 <sipa> MC1984: look at the publication date
659 2013-03-31 05:24:57 <sydna> I don't think there's any use for insecure keys. as someone pointed out on reddit, the brain wallet sha256('i love bitcoin') held 500BTC at some point.
660 2013-03-31 05:25:17 <MC1984> welp
661 2013-03-31 05:25:27 <gmaxwell> sydna: do these sites do signing too?
662 2013-03-31 05:25:46 <sydna> gmaxwell: one of them does
663 2013-03-31 05:25:47 <MC1984> i bet it even works
664 2013-03-31 05:25:52 <sipa> MC1984: it did
665 2013-03-31 05:26:04 <gmaxwell> if they use bad random numbers in the signing they'll leak the private keys, even if the private keys were securely generated.
666 2013-03-31 05:26:10 <sipa> it's one of the few april fools' RFC's with an actually working implementation
667 2013-03-31 05:26:21 <muhoo> ok folks. is there a tool that generates SECURE keys and will print out a paper wallet?
668 2013-03-31 05:26:24 <sydna> gmaxwell: http://brainwallet.org/#sign
669 2013-03-31 05:26:31 <MC1984> lawd
670 2013-03-31 05:26:53 <muhoo> because, bitaddress.org RNG does not inspire confidence
671 2013-03-31 05:27:10 <gmaxwell> muhoo: uh. bitcoind dumpprivkey .. and lpr?
672 2013-03-31 05:27:15 <gmaxwell> :P
673 2013-03-31 05:27:22 <muhoo> with QR codes
674 2013-03-31 05:27:44 <gmaxwell> and gnuqrcodemaker
675 2013-03-31 05:27:45 <kadoban> there's simple tools to make QR codes
676 2013-03-31 05:27:45 <sipa> ./bitcoind dumpprivkey | qrencode | lpr
677 2013-03-31 05:27:46 <gmaxwell> ? :P
678 2013-03-31 05:28:12 <kadoban> anyway, if you want something fancy, doesn't armory have that kind of stuff?
679 2013-03-31 05:28:20 <muhoo> i don't mean for ME, smartys :-P
680 2013-03-31 05:28:29 <kadoban> never used it, but that's what i plan to look into when i have enough coins to care
681 2013-03-31 05:28:37 <muhoo> my keys are all in emacs org mode
682 2013-03-31 05:28:49 <gmaxwell> muhoo: those sites were already off my list to recommend because they're sites. :P
683 2013-03-31 05:29:16 <muhoo> i didn't say site. i said tool. like, a program.
684 2013-03-31 05:29:27 <gmaxwell> Plus joric's entry into bitcoin address generation was bruteforcing keys, using his site seems ... foolish even though I'm sure he's a swell guy
685 2013-03-31 05:30:07 <gmaxwell> muhoo: armory, I assume??? thought I've not used that functionality of it... etotheipi's tried to make the paper backup stuff pretty good as I understand it
686 2013-03-31 05:30:20 <muhoo> cool, will look into that.
687 2013-03-31 05:30:28 <sydna> gmaxwell: you've got me looking for mt_rand in things now. I'm genuinely horrified.
688 2013-03-31 05:30:55 <gmaxwell> I like the gambling sites that have used it.
689 2013-03-31 05:31:21 <sydna> ACTION rolls eyes
690 2013-03-31 05:31:39 <gmaxwell> sydna: also people using sites that give out "random numbers" lol.
691 2013-03-31 05:35:09 <sydna> gmaxwell: nothing would surprise me at this point.
692 2013-03-31 05:35:27 <sipa> sydna: not even a man eating his own head?
693 2013-03-31 05:36:16 <muhoo> eating? or inserting it in the other end?
694 2013-03-31 05:36:55 <sydna> sipa: before I saw a php payment script that was "validating" input by stripping words like "SELECT" out. a man eating his own head wouldn't even make me blink.
695 2013-03-31 05:37:22 <muhoo> little bobby tables
696 2013-03-31 05:38:19 <sipa> do people still pass input data directly to SQL? i thought prepared statements solved just about any problem possible with that...
697 2013-03-31 05:38:23 <MC1984> why does bitcoin.org claim qt takes a day to sync
698 2013-03-31 05:38:56 <kadoban> sipa: yeah, but that's not described in horrible PHP tutorials that geniuses learn from
699 2013-03-31 05:39:00 <sydna> sipa: I've seen maybe two or three people use PDO in PHP, the rest still pass user input straight to mysql_query()
700 2013-03-31 05:39:26 <phantomcircuit> MC1984, that's about how long it takes on a slowish laptop
701 2013-03-31 05:39:42 <gmaxwell> the world is pretty much doomed.. I mean.. I've found the wrong value of pi in a number of different things.
702 2013-03-31 05:39:59 <sydna> 3.1416. that's good enough.
703 2013-03-31 05:40:04 <gmaxwell> as soon as someone builds a computer system which must be correct or we all die.. we'll all die
704 2013-03-31 05:40:14 <sydna> ACTION has pi memorised up to 150 decimal places
705 2013-03-31 05:40:22 <MC1984> i like pie
706 2013-03-31 05:40:40 <gmaxwell> sydna: hopefully you were careful where you got it from, wouldn't it be sad to have memorized 150 digits of bad pi?
707 2013-03-31 05:40:40 <MC1984> phantomcircuit i should benchmark it again on this machine
708 2013-03-31 05:40:56 <MC1984> ~6 yr oldlaptop
709 2013-03-31 05:41:09 <gmaxwell> MC1984: under promise over deliver.
710 2013-03-31 05:41:13 <gwillen> sydna: the last php code I had to work on used sprintf and mysql_real_escape_string
711 2013-03-31 05:41:19 <gwillen> sydna: which I think is at least correct, if not elegant
712 2013-03-31 05:41:20 <gmaxwell> it can take a day, if your disk is failing or you just get unlucky with peers.
713 2013-03-31 05:41:52 <MC1984> telling people to wait a day is scary though
714 2013-03-31 05:41:52 <sydna> gmaxwell: as far as I've seen it's the right digits
715 2013-03-31 05:42:03 <MC1984> its where i start to buy the scaring off newbs argument
716 2013-03-31 05:42:47 <Scrat> sipa: I had to consult for a big PHP project a few months ago and they had written every single SQL command manually (quite a lot of them). needless to say it was full of gems like \\'" + $_GET["pagenumber"] + \\'"
717 2013-03-31 05:42:59 <gmaxwell> it takes under an hour on my laptop, so long as peer selection doesn't get in the way. I synced over tor last week and it took about two and a half hours
718 2013-03-31 05:43:25 <Scrat> I told the guy that it's horrible and he told me to cut the bullshit, it's fine because he "encased it in quotes"
719 2013-03-31 05:43:28 <sydna> gwillen: the issue isn't really that it is inelegant, it's that people insist on not escaping input. yes, they do all the $_GET variables, but not input that has come from other places
720 2013-03-31 05:43:38 <Scrat> then I proceded to drop his tables in front of him
721 2013-03-31 05:44:19 <sydna> ugh.
722 2013-03-31 05:44:20 <gwillen> "Help my mouse is moving by it self"
723 2013-03-31 05:44:27 <gmaxwell> hah
724 2013-03-31 05:44:27 <MC1984> peer selection seems to catch out a few people
725 2013-03-31 05:44:42 <MC1984> it just gets stuck on a slow peer and no one things to restart it
726 2013-03-31 05:44:57 <sydna> the issue with PHP isn't PHP, it's that the community is full of idiots writing production code.
727 2013-03-31 05:45:00 <MC1984> and then bitcoin is indelibly shit in a newbs mind
728 2013-03-31 05:45:42 <gmaxwell> MC1984: so where were you when TD was telling sipa to not work on that earlier? :P
729 2013-03-31 05:46:12 <warren> Haha. litecoin screwed up in a non-fatal way that will worry the users.
730 2013-03-31 05:46:21 <gmaxwell> hum?
731 2013-03-31 05:46:36 <warren> gmaxwell: http://pastebin.com/btg1evia http://p2pool-ltc.info:9327/static/
732 2013-03-31 05:46:37 <MC1984> my eyelid is twitching again....
733 2013-03-31 05:46:59 <gmaxwell> warren: block version numbers?
734 2013-03-31 05:47:04 <gmaxwell> knocked everyone into safemode?
735 2013-03-31 05:47:09 <warren> gmaxwell: the official binaries from mid-2012 are from coblee's tree. Then someone bumped a version for no reason in the project fork.
736 2013-03-31 05:47:10 <warren> oh?
737 2013-03-31 05:47:19 <warren> hmm
738 2013-03-31 05:47:29 <warren> gmaxwell: not everyone, only people who built from the other git repo
739 2013-03-31 05:47:41 <warren> i'm trying to track down exactly what changed
740 2013-03-31 05:48:24 <warren> gmaxwell: that error knocks people into safemode?
741 2013-03-31 05:48:37 <MC1984> what do people think about changing the GUI langue from "downlaoding blocks" or whatever it is to "processing blocks"
742 2013-03-31 05:49:02 <MC1984> because too many people are assuming that the download is slow, when its the processing that is holding them up
743 2013-03-31 05:49:09 <MC1984> then they go asking for a torrent etc
744 2013-03-31 05:49:19 <MC1984> when it probably wont help
745 2013-03-31 05:49:42 <kadoban> MC1984: does it even say downloading? the text i see says like.."Synchronizing with network" and other stuff says "Processed blah blah out of blah blah"
746 2013-03-31 05:50:00 <sipa> MC1984: right now, it is likely the downloading part that is slow :)
747 2013-03-31 05:50:08 <MC1984> i cant remember
748 2013-03-31 05:50:38 <MC1984> hmm is it really, early blocks are tiny and come at hundreds per second
749 2013-03-31 05:50:45 <MC1984> later blocks put your CPU in a tarpit
750 2013-03-31 05:50:58 <gmaxwell> it used to be the case that the download wasn't the slow part, it often is now.
751 2013-03-31 05:51:14 <MC1984> ok
752 2013-03-31 05:51:28 <gmaxwell> as mentioned before, my laptop can sync in under an hour. .. but it could also take several if peers are unkind.
753 2013-03-31 05:52:40 <sydna> gmaxwell: I was trying to think of where I'd seen bitcoin signatures; then I realised that I've never actually seen one. they're rarer than GPG.
754 2013-03-31 05:52:44 <MC1984> yeah just from watching the logtail the p2p logic seems pretty basic right now
755 2013-03-31 05:55:27 <gmaxwell> sydna: huh? they're in every transaction.
756 2013-03-31 05:55:28 <gmaxwell> :P
757 2013-03-31 05:56:07 <sydna> gmaxwell: I meant publicly, as a message identifier
758 2013-03-31 05:56:32 <MC1984> speaking of which http://imgur.com/TVXp04A pimp my bitcoind
759 2013-03-31 05:56:45 <gmaxwell> sydna: they're used for OTC auth by a number of people.
760 2013-03-31 05:56:59 <Scrat> sydna: someone should make a gpg alternative that uses the same curve as bitcoin (or something like ed25519)
761 2013-03-31 05:57:00 <sydna> of course!
762 2013-03-31 05:57:04 <Scrat> RSA/DSA is so 1990
763 2013-03-31 05:57:35 <sydna> Scrat: getting people to use GPG itself is hard enough
764 2013-03-31 05:57:48 <Scrat> you need sexy UIs
765 2013-03-31 05:57:57 <gmaxwell> MC1984: http://www.sostars.com/hackers-font/
766 2013-03-31 06:00:13 <sydna> MC1984: they should call you mario, because you just got one-upped ??? http://i.imgur.com/Uk7Fo9T.jpg
767 2013-03-31 06:02:14 <MC1984> thats nerdy
768 2013-03-31 06:02:33 <MC1984> level 17 wizard mage nerdy
769 2013-03-31 06:02:54 <sydna> nothing like using an i7 processor and a 56kbp/s terminal emulator.
770 2013-03-31 06:04:51 <gmaxwell> sydna: totally ungeeked by the use of mac os. Sorry, youve been outdone: https://people.xiph.org/~greg/screenshot.png
771 2013-03-31 06:05:27 <gmaxwell> (the 'apple2' screensaver thats part of xscreensaver can be used as a terminal emulator 0_o)
772 2013-03-31 06:05:54 <Scrat> hmm, apparently there is ECC support in GPG 2.1beta
773 2013-03-31 06:06:04 <sydna> yikes, that's completely unusable.
774 2013-03-31 06:06:11 <MC1984> "ignoring 1dice"
775 2013-03-31 06:06:17 <sydna> ^ I like it
776 2013-03-31 06:06:40 <Scrat> LOL
777 2013-03-31 06:06:49 <sydna> hey wait. you'd be ignoring my addresses too then. I use 1dice addresses for my TX :\\
778 2013-03-31 06:07:07 <MC1984> lol cnsorship 4 u
779 2013-03-31 06:08:13 <gmaxwell> gives me lots of orphans to test with.
780 2013-03-31 06:08:19 <sydna> gmaxwell: how many nodes would have to be ignoring 1dice TX for it to have a noticable effect?
781 2013-03-31 06:08:35 <sydna> ** noticeable
782 2013-03-31 06:08:39 <MC1984> thats what menagle said you monster!
783 2013-03-31 06:09:02 <gmaxwell> MC1984: lol
784 2013-03-31 06:09:34 <Scrat> sydna: I doubt it would have any effect. they wait for confirmations now
785 2013-03-31 06:09:38 <gmaxwell> sydna: nearly all of them.
786 2013-03-31 06:09:41 <Scrat> did they get finney'd a lot?
787 2013-03-31 06:10:49 <gmaxwell> Scrat: yep, though that didn't have anything to do with relay policy.
788 2013-03-31 06:11:06 <Scrat> yeah, it had to do with eligius
789 2013-03-31 06:12:31 <sydna> if they wait for confirmations now, a big pool like BTC Guild ignoring them (I don't know if they do or not) would have a huge impact on them
790 2013-03-31 06:13:01 <gmaxwell> hm. I see some weird behavior, right after a new block I got flooded with reannounced transactions.
791 2013-03-31 06:13:46 <sydna> blockchain.info has the same behaviour, as soon as a block is processed they show a flood of new TX
792 2013-03-31 06:13:56 <gmaxwell> thats ... broken.
793 2013-03-31 06:14:51 <TheButterZone> I was told "basically http://blockchain.info/??doesnt adhere to the standard fees of the reference client, so some of their txns never get relayed"
794 2013-03-31 06:15:11 <TheButterZone> when i was trying to work on my /pushtx not getting 1 confirm for days
795 2013-03-31 06:15:39 <gmaxwell> TheButterZone: okay, but that shouldn't have anything to do with just flooding transactions after a block
796 2013-03-31 06:15:48 <sydna> I was under the impression that they paid a standard 0.0005BTC fee
797 2013-03-31 06:16:17 <gmaxwell> All my peers claim to be one of
798 2013-03-31 06:16:19 <gmaxwell> "subver" : "/Satoshi:0.7.0.2/",
799 2013-03-31 06:16:19 <gmaxwell> "subver" : "/Satoshi:0.7.0.3/",
800 2013-03-31 06:16:19 <gmaxwell> "subver" : "/Satoshi:0.7.1/",
801 2013-03-31 06:16:19 <gmaxwell> "subver" : "/Satoshi:0.8.0/",
802 2013-03-31 06:16:21 <gmaxwell> "subver" : "/Satoshi:0.8.1/",
803 2013-03-31 06:16:28 <gmaxwell> and none of those should do that. :-/
804 2013-03-31 06:16:54 <Scrat> one of them is faking it!
805 2013-03-31 06:17:01 <gmaxwell> yea likely.
806 2013-03-31 06:17:45 <sydna> connect back to each one individually and compare the TX? that's easy enough
807 2013-03-31 06:18:03 <sydna> I've done that a couple of times to compare the latency between nodes.. for fun.
808 2013-03-31 06:18:23 <gmaxwell> I'll just switch to a patch that logs IPs on incoming transactions and see when it happens again.
809 2013-03-31 06:18:55 <sydna> you'll get different peers that way though, won't you?
810 2013-03-31 06:19:58 <sydna> a couple of instances of bitcoin-sniffer is more fun.
811 2013-03-31 06:33:17 <warren> gmaxwell: we figured out what happened. It's kind of funny actually.
812 2013-03-31 06:34:36 <xQuasar> BITCOINS CAN SUCK MY FUCKING COCK NIGGERS
813 2013-03-31 06:34:57 <gmaxwell> warren: someone got pool software that was fixed for the bitcoin block version change.
814 2013-03-31 06:35:40 <gmaxwell> warren: you can tell me that I'm right at any point in time.
815 2013-03-31 06:35:53 <warren> yup
816 2013-03-31 06:36:26 <gmaxwell> "copy bitcoin stuff less blindly nexttime genuises" :P
817 2013-03-31 06:36:31 <warren> gmaxwell: and we know exactly who it is, the pool that's been ~50% of the hash power for the past few months ... I'm guessing they went ahead and rebased litecoind as a competitive advantage.
818 2013-03-31 06:37:14 <gmaxwell> warren: oh I think you're probably not correct there.
819 2013-03-31 06:37:21 <warren> oh? ok.
820 2013-03-31 06:38:34 <gmaxwell> I expect they updated to a later eloipool or whatever daemon they are using to generate blocks. and whatever it is was panic updated to support height in coinbase in the last month and now produces v2 blocks.
821 2013-03-31 06:39:05 <warren> hah
822 2013-03-31 06:39:27 <sydna> I'm guessing that nobody really holds LTC in high regard?
823 2013-03-31 06:39:49 <gmaxwell> hm? my comments are not reflecting any regard for ltc in any particular direction.
824 2013-03-31 06:39:58 <warren> sydna: the speculators do, amazingly. I'm shaking my head amazed that this hasn't blown up due to lack of attention.
825 2013-03-31 06:41:02 <warren> It's really a testament to Satoshi that it hasn't killed itself yet.
826 2013-03-31 06:41:13 <Perdos> if it blew up the attention would increase
827 2013-03-31 06:41:30 <warren> gmaxwell: hmm, first v2 block in LTC began Sun, 12 Aug 2012 19:50:06 GMT
828 2013-03-31 06:41:55 <gmaxwell> that probably is someone screwing around.
829 2013-03-31 06:42:19 <gmaxwell> do the current block in question have height in coinbase?
830 2013-03-31 06:45:14 <warren> I'm not sure how to see that.
831 2013-03-31 07:21:09 <doublec> warren: what pool software is this?
832 2013-03-31 07:21:51 <doublec> oh litecoin pool, I see
833 2013-03-31 07:22:28 <warren> doublec: gmaxwell's explanation makes the most sense
834 2013-03-31 08:01:36 <MC1984> anyone still producing empty blocks?
835 2013-03-31 08:08:06 <Bluetegu> Hi there. Yesterday I compiled bitcoind and run it as daemon. This morning I found out the server crashed with out-of-memory errors. Any idea where to start fixing this? Thanks
836 2013-03-31 08:08:40 <sipa> Bluetegu: we've been working on memory issues very recently
837 2013-03-31 08:09:10 <sipa> Bluetegu: if you don't mind running bleeding-edge code, you can test current git master
838 2013-03-31 08:09:37 <Bluetegu> I actually got it from git master yesterday
839 2013-03-31 08:10:15 <sipa> the patch i'm referring to was merged about 6 hours ago
840 2013-03-31 08:10:58 <Bluetegu> Yes, I pulled and got something
841 2013-03-31 08:11:31 <Bluetegu> Should I simply recompile and run, or are there any flags that would help with debugging?
842 2013-03-31 08:12:04 <sipa> not really; just running the new code as-is should improve memory usage a lot; especially when you have many connections
843 2013-03-31 08:12:36 <Bluetegu> Ok, great. Many thanks for your prompt response
844 2013-03-31 08:37:53 <MC1984> <MC1984> no its reindexing it
845 2013-03-31 08:37:53 <MC1984> <sam> but it seems stuck at 3360 blocks ago (ie. the blockchain fork)
846 2013-03-31 08:37:53 <MC1984> <sam> mmmh I upgraded an old box from client 0.6.2 to 0.8.1 and it's now downloading all the old blockchain histry
847 2013-03-31 08:39:19 <sipa> how long has it been stuck? wait an hour or so
848 2013-03-31 08:41:50 <Bluetegu> I compiled the new code, and run it. For some reason it didn't return the prompt after I run bitcoind -daemon. Is this ok?
849 2013-03-31 08:42:07 <Bluetegu> It is running and returning getinfo