1 2012-11-09 01:45:28 <owowo> <3
2 2012-11-09 01:49:15 <gmaxwell> Why can't more people be like that?
3 2012-11-09 02:06:29 <Luke-Jr> lol
4 2012-11-09 02:29:19 <BlueMatt> nanotube: nice
5 2012-11-09 02:34:49 <phantomcircuit> gmaxwell, extremely strange?
6 2012-11-09 02:45:13 <astor> has anybody tried to fix lock ordering issues in the bitcoin software?
7 2012-11-09 02:46:04 <BlueMatt> ...
8 2012-11-09 02:46:16 <BlueMatt> uhh...yea, like 1.5 years ago?
9 2012-11-09 02:48:52 <Luke-Jr> lol
10 2012-11-09 02:48:53 <astor> no I mean, I went through the software roughly a year ago and annotated all fields with locking info. I remember finding a few issues, but I've lost my repository. One of my main concerns at the time was that the code allowed for recursive locking, which is a pretty dangerous practice.
11 2012-11-09 02:49:59 <BlueMatt> there were a few issues nearly 2 years ago but they are long since fixed
12 2012-11-09 02:50:28 <BlueMatt> using recursive locking...yea, the code flow isnt very obvious, but once you study it some its not absolutely godawful, its just kinda bad...
13 2012-11-09 03:00:29 <astor> actually I found my repo
14 2012-11-09 03:01:57 <astor> it's a 1500 lines patch that makes it mostly -Wthread-safety safe with clang.
15 2012-11-09 03:26:42 <Luke-Jr> astor: so where's the pullreq?
16 2012-11-09 03:45:37 <Diablo-D3> gmaxwell: well that was fun
17 2012-11-09 03:46:00 <Diablo-D3> I spent like two hours seeing if I could get zero latency screen casting
18 2012-11-09 03:46:42 <Diablo-D3> got pretty close
19 2012-11-09 03:46:52 <gmaxwell> Diablo-D3: use wayland.
20 2012-11-09 03:47:14 <Diablo-D3> ffmpeg -fflags nobuffer -flags:v low_delay -f x11grab -r 60 -s 1920x1200 -i :0.0 -vcodec libx264 -profile:v high444 -preset ultrafast -tune zerolatency,fastdecode,stillimage -qp 18 -f mpegts -g 1 tcp://192.168.2.2:8888?listen
21 2012-11-09 03:47:27 <gmaxwell> x11grab is always lossy.
22 2012-11-09 03:47:37 <Diablo-D3> then on another box, ffplay -fflags nobuffer -i http://192.168.2.2:8888
23 2012-11-09 03:47:44 <Diablo-D3> its about a quarter of a second behind for me
24 2012-11-09 03:48:16 <Diablo-D3> gmaxwell: well, I doubt theres a faster method
25 2012-11-09 03:48:25 <Diablo-D3> and dont say wayland, that doesnt work on fglrx
26 2012-11-09 03:48:58 <gmaxwell> I feel bad for you son, I've got 99 problems but a desktop with non-intel video aint one.
27 2012-11-09 03:49:03 <Luke-Jr> :D
28 2012-11-09 03:49:06 <Diablo-D3> >intel
29 2012-11-09 03:49:08 <Diablo-D3> >bitcoin
30 2012-11-09 03:49:09 <Diablo-D3> quat
31 2012-11-09 03:49:29 <Diablo-D3> gmaxwell: wait, why is x11grab lossy?
32 2012-11-09 03:50:13 <gmaxwell> Because it misses frames, and has tearing because the grab isn't atomic.
33 2012-11-09 03:50:22 <Diablo-D3> oh, that
34 2012-11-09 03:50:25 <gmaxwell> (it's also slow because god knows why)
35 2012-11-09 03:50:39 <Diablo-D3> its slow because grabbing out of the X framebuffer is fucking shit
36 2012-11-09 03:50:46 <Diablo-D3> its also why x11vnc and whatever else is slow
37 2012-11-09 03:51:11 <gjs278> x11vnc is great
38 2012-11-09 03:51:27 <Diablo-D3> gjs278: yeah, but its no thin client replacement
39 2012-11-09 03:51:45 <gjs278> I truly believe my high school tried to use it as one
40 2012-11-09 03:52:07 <Diablo-D3> gjs278: no, that was NT remote thin client
41 2012-11-09 03:52:09 <gmaxwell> gjs278: its a neat hack... but it's slow.
42 2012-11-09 03:52:22 <Diablo-D3> its rdp, which is very similar to vnc, and still slow as fuck
43 2012-11-09 03:52:29 <gmaxwell> I used x11 vnc to 'hack' a friend of mine who bragged his system was unhackable.
44 2012-11-09 03:52:41 <gjs278> bro run this program. k now portforward.
45 2012-11-09 03:52:42 <Luke-Jr> lol
46 2012-11-09 03:52:47 <gmaxwell> no way man.
47 2012-11-09 03:52:48 <Diablo-D3> but yeah, like
48 2012-11-09 03:52:52 <Diablo-D3> seen redhat spice?
49 2012-11-09 03:52:58 <Diablo-D3> it uses mjpeg for motion encoding
50 2012-11-09 03:53:13 <Diablo-D3> my method is lower latency than mjpeg <3
51 2012-11-09 03:53:43 <gmaxwell> I had him SSH into a system of mine to look at something as a pretext. Of course, he had X11 forwarding enabled. I started up x11vnc as him on my system, connected.. then waited until he got up and grabbed a root terminal he left up and added my ssh key.
52 2012-11-09 03:56:36 <gmaxwell> (there are lots of fun things you can do with ssh forwarding, x11vnc is just the simplest)
53 2012-11-09 03:57:38 <Arnavion> Do tell
54 2012-11-09 03:58:40 <gmaxwell> Arnavion: I meant ssh x11 forwarding... if you can connect to someone's xserver you can sniff their keyboard and such.
55 2012-11-09 03:59:25 <Diablo-D3> wait
56 2012-11-09 03:59:30 <Diablo-D3> how did you do that without his cookies?
57 2012-11-09 04:00:29 <gmaxwell> ssh x11 forwarding. It setups up the auth on the system you log into (where I am root)
58 2012-11-09 04:00:42 <gmaxwell> otherwise you couldn't run X clients yourself.
59 2012-11-09 04:01:12 <Diablo-D3> Im aware of how ssh x11 forwarding works, but it only works for the user running X
60 2012-11-09 04:01:24 <Diablo-D3> it cant hijack other people's ssh sessions
61 2012-11-09 04:01:28 <Luke-Jr> ACTION facepalms
62 2012-11-09 04:01:43 <Luke-Jr> obviously gmaxwell tricked him into SSHing to gmaxwell's box, with X11 forwarding enabled
63 2012-11-09 04:01:51 <Luke-Jr> and from there hijacked the connection
64 2012-11-09 04:01:54 <gmaxwell> Right.
65 2012-11-09 04:02:12 <Luke-Jr> gmaxwell: does the sniffing crap work even with -X ?
66 2012-11-09 04:02:15 <Luke-Jr> ie, not -Y
67 2012-11-09 04:02:23 <Diablo-D3> you should have to have been logged in as him
68 2012-11-09 04:02:27 <Luke-Jr> ???
69 2012-11-09 04:02:32 <Luke-Jr> Diablo-D3: su victim
70 2012-11-09 04:03:04 <gmaxwell> "I started up x11vnc as him on my system"
71 2012-11-09 04:03:30 <Diablo-D3> ...
72 2012-11-09 04:03:33 <Diablo-D3> well isnt that ugly.
73 2012-11-09 04:03:51 <gmaxwell> Luke-Jr: I don't think ~any thing bothers using the secure grab stuff to get passwords. Does it?
74 2012-11-09 04:04:16 <gmaxwell> Diablo-D3: this is why you should _never_ x11 forward to a host you don't fully trust.
75 2012-11-09 04:04:47 <Luke-Jr> gmaxwell: I'd hope ssh/gpg agent do!
76 2012-11-09 04:05:02 <Diablo-D3> gmaxwell: yeah, I already have ForwardX11 no in my .ssh/config
77 2012-11-09 04:05:18 <Luke-Jr> gmaxwell: I'd also have expected -X to deny crap like "grab the framebuffer" in the first place :/
78 2012-11-09 04:07:12 <gmaxwell> Luke-Jr: I imagine a lot of things wouldn't work.
79 2012-11-09 04:08:07 <gmaxwell> Luke-Jr: not that they need it, but it's like with SE Linux??? a lot of stuff makes crazy irrelevant syscalls just because they can... try to selinux lock stuff down and you either have to permit a lot of crazy or fix a lot of bugs.
80 2012-11-09 04:08:54 <phantomcircuit> gmaxwell, ForwardX11 defaults to no
81 2012-11-09 04:09:04 <gmaxwell> phantomcircuit: it does now. It did not use to.
82 2012-11-09 04:09:11 <phantomcircuit> that's hilarious
83 2012-11-09 04:09:19 <Luke-Jr> gmaxwell: sigh, I wonder when someone will make an OS with a reasonable security model :P
84 2012-11-09 04:09:53 <Luke-Jr> though current CPU designs, and C specs, etc might make that non-trivial
85 2012-11-09 04:10:13 <phantomcircuit> Luke-Jr, there's one where things are isolated using hardware virtualization
86 2012-11-09 04:10:17 <phantomcircuit> forget it's name
87 2012-11-09 04:10:33 <Luke-Jr> (by reasonable, I'm thinking of something like MOO where code executes with its programmer's permissions by default and needs to elevate to do anything the programmer couldn't do himself)
88 2012-11-09 04:11:52 <Luke-Jr> the problem with CPUs/C compatibility being that library functions need to execute with different permissions than the code calling it; and who knows how a library intends to use a pointer :/
89 2012-11-09 04:11:55 <gmaxwell> Luke-Jr: there are a couple neat papers on cpu architectures that have additional tags for all memory words, so you can do things like valgrind everything with no performance hit.
90 2012-11-09 04:15:02 <Luke-Jr> gmaxwell: hmm, more like the CPUs have the performance hit for everything by default? :0
91 2012-11-09 04:15:03 <Luke-Jr> ;)*
92 2012-11-09 04:15:22 <Diablo-D3> gmaxwell: DO WANT
93 2012-11-09 04:17:23 <gmaxwell> Luke-Jr: certantly less than having to emulate+jit everything in software! I think for a lot of applications it would be worth having a half speed chip. Esp since it could do even fancier things like multi-level taint tracking.
94 2012-11-09 04:18:01 <Luke-Jr> gmaxwell: hmm, but that still didn't sound like it'd solve the library problem
95 2012-11-09 04:19:12 <gmaxwell> Luke-Jr: well, it would allow enforcement. E.g. app data and library data would have different tags. And you trap if you violate the tags... then some handler wakes up and does something smart.
96 2012-11-09 04:20:14 <Luke-Jr> gmaxwell: hmm, I was imagining implementing it by running each code-user in a separate process and doing some kind of internal RPC for cross-user calls
97 2012-11-09 04:22:03 <gmaxwell> Luke-Jr: ever seen selinux sanbox thats part of fedora/rhel?
98 2012-11-09 04:22:18 <phantomcircuit> lol
99 2012-11-09 04:22:29 <phantomcircuit> gmaxwell, runs in permissive mode iirc
100 2012-11-09 04:22:37 <phantomcircuit> blacklist instead of whitelist
101 2012-11-09 04:22:55 <Luke-Jr> gmaxwell: no
102 2012-11-09 04:23:05 <gmaxwell> It's a wrapper 'sandbox ./scarry-program' and basically scarry-program runs as a one-shot sandbox user that can only do IO outside of a chroot like jail.
103 2012-11-09 04:23:32 <Luke-Jr> gmaxwell: sure, but that's the entire program ???
104 2012-11-09 04:23:35 <gmaxwell> phantomcircuit: yes, but it does run the program as another user at least, and its substantilly (though indeed, not completely) clamped down.
105 2012-11-09 04:23:56 <phantomcircuit> i tried running gentoo hardened w/ selinux in strict mode
106 2012-11-09 04:24:02 <phantomcircuit> couldn't even login
107 2012-11-09 04:24:05 <phantomcircuit> :|
108 2012-11-09 04:27:06 <Diablo-D3> brb
109 2012-11-09 04:33:35 <jgarzik> ACTION wonders if sandbox could be combined with cgroups
110 2012-11-09 04:33:57 <phantomcircuit> jgarzik, selinux has resource controls also
111 2012-11-09 04:34:06 <phantomcircuit> i suspect you could modify a role's permissions or something
112 2012-11-09 04:34:30 <jgarzik> a true sandbox limits cpu, memory, disk usage, ...
113 2012-11-09 04:34:43 <jgarzik> I know Google uses cgroups heavily, to do that for their processes
114 2012-11-09 04:38:37 <phantomcircuit> jgarzik, i use cgroups a lot for momentovps
115 2012-11-09 04:51:35 <Joric> mourning! does anyone know how to hide a part of gihub comment under a spoiler? ie markdown for expandable/collapsible content
116 2012-11-09 04:53:04 <astor> I've added my patch at https://github.com/alexanderkjeldaas/bitcoin There are pages and pages of locking warnings still with clang, so I'm not sure if a pull request is the right thing to do. Please take a look.
117 2012-11-09 05:10:27 <jgarzik> -c --cgroups
118 2012-11-09 05:10:27 <jgarzik> Use cgroups to control this copy of seunshare. Specify parame???
119 2012-11-09 05:10:32 <jgarzik> nice
120 2012-11-09 07:07:18 <sipa> ;;bc,blocks
121 2012-11-09 07:07:19 <gribble> 207147
122 2012-11-09 07:26:25 <abrkn> hmm i cant get my testnet in a box up
123 2012-11-09 07:26:40 <abrkn> bt qt says 1 connection but "wallet out of sync"
124 2012-11-09 07:26:43 <abrkn> has been running for ages
125 2012-11-09 07:28:14 <gmaxwell> abrkn: what version of bitcoin?
126 2012-11-09 07:28:50 <abrkn> my client is 1.7.0
127 2012-11-09 07:29:03 <abrkn> err
128 2012-11-09 07:29:05 <abrkn> 0.7.0 ofc
129 2012-11-09 07:30:05 <abrkn> https://gist.github.com/4044447
130 2012-11-09 07:31:14 <abrkn> i pretty much just extracted the testnet-in-a-box to some dir and spun a daemon and a client
131 2012-11-09 07:32:16 <sipa> gmaxwell: found the reason for my crashing mining node
132 2012-11-09 07:32:40 <sipa> there's an assert in CreateNewBlock that is only active undef (fDebug) ...
133 2012-11-09 07:32:47 <sipa> *under
134 2012-11-09 07:37:27 <abrkn> is there a different one i can use? http://sourceforge.net/projects/bitcoin/files/Bitcoin/testnet-in-a-box/
135 2012-11-09 07:38:28 <gmaxwell> abrkn: testnet in a box is a special two node testnet setup. Normal testnet is just part of the regular bitcoin client. (which is at 0.7.1)
136 2012-11-09 07:38:51 <abrkn> gmaxwell: right, i'd rather have the testnet-in-a-box
137 2012-11-09 07:38:51 <sipa> hmmm, is testnet in a box updated for testnet3?
138 2012-11-09 07:38:58 <gmaxwell> abrkn: for testnet in a box just having one connection is expected.
139 2012-11-09 07:39:14 <abrkn> gmaxwell: right, but it says out of sync and nothing gets generated
140 2012-11-09 07:39:16 <gmaxwell> sipa: dunno, gavin has a tniab set of files but I dunno whats on that page.
141 2012-11-09 07:39:39 <gmaxwell> sipa: ah, see the url, there is a testnet3 archive there.
142 2012-11-09 07:39:48 <gmaxwell> abrkn: you used the testnet3 archive, right?
143 2012-11-09 07:41:22 <abrkn> now i'm unsure, downloaded a few days ago. trying a clean run
144 2012-11-09 07:41:32 <sipa> still strange how transactions can enter the mempool without their parents being in there
145 2012-11-09 07:41:59 <sipa> i know it is possible during reorgs
146 2012-11-09 07:42:14 <sipa> but there is nothing special in my debug.log
147 2012-11-09 07:46:38 <abrkn> ok, testnet3 up and running thanks
148 2012-11-09 07:47:32 <sipa> :o
149 2012-11-09 07:48:00 <sipa> CTxMempool::remove doesn't remove dependent transactions
150 2012-11-09 07:48:49 <sipa> hmm, it shouldn't, actually
151 2012-11-09 08:49:53 <thermoman> hi there. we have issued a transaction via 0.7.1 client but it didn't make it into the bitcoin network, confirmations=0. any idea? http://pastebin.com/aFU8MKN8
152 2012-11-09 08:50:52 <edcba> fee=0 ?
153 2012-11-09 08:51:38 <edcba> time 12345 ??
154 2012-11-09 08:51:44 <edcba> you replaced it ?
155 2012-11-09 09:08:17 <sturles> thermoman: The transaction you pasted isn't valid. Sure you didn't change anything?
156 2012-11-09 09:09:28 <thermoman> sturles: it's redacted
157 2012-11-09 09:09:42 <thermoman> edcba: yes
158 2012-11-09 09:10:09 <thermoman> the real values were replaced
159 2012-11-09 09:10:22 <sturles> In that case it is no use pasting it. Check your debug.log.
160 2012-11-09 09:10:39 <sipa> thermoman: how long has the node been running since creating the tx?\\
161 2012-11-09 09:10:48 <thermoman> sipa: days
162 2012-11-09 09:11:11 <thermoman> sipa: 8 days or so
163 2012-11-09 09:21:31 <abrkn> why do i keep getting banned from bitcoin-otc?
164 2012-11-09 09:21:36 <abrkn> ;;ident abrkn
165 2012-11-09 09:21:36 <gribble> Nick 'abrkn', with hostmask 'abrkn!~pialur@195.159.164.228', is not identified.
166 2012-11-09 09:23:08 <abrkn> ;;ident abrkn
167 2012-11-09 09:23:08 <gribble> Nick 'abrkn', with hostmask 'abrkn!~pialur@195.159.164.228', is identified as user abrkn, with GPG key id C73739F294E8EB0E, key fingerprint 9184ABC7B25EDAEB4D1C6737C73739F294E8EB0E, and bitcoin address 1abrknajSFpnz7MHjLkVnuvCbwd96wSYt
168 2012-11-09 09:25:11 <thermoman> are information in debug log regarding CTransaction sensitive?
169 2012-11-09 09:25:54 <sipa> depends what you call sensitive
170 2012-11-09 09:26:00 <sipa> it won't reveal private keys
171 2012-11-09 09:27:21 <thermoman> what about "WalletUpdateSpent found spent coin" and the long string at the end?
172 2012-11-09 09:31:33 <thermoman> ok, its a transaction
173 2012-11-09 09:31:34 <sipa> that's the transaction id
174 2012-11-09 09:31:59 <sipa> debug.log may reveal privacy-sensitive information, but will never allow stealing of coins
175 2012-11-09 09:32:14 <thermoman> that's the problem
176 2012-11-09 09:32:51 <thermoman> i'm looking at a transaction on blockexplorer
177 2012-11-09 09:33:07 <thermoman> it says input: amount X
178 2012-11-09 09:33:12 <thermoman> 2x output
179 2012-11-09 09:33:21 <thermoman> one of these outputs has "Not yet redeemed"
180 2012-11-09 09:37:25 <thermoman> i'll talk to your developer
181 2012-11-09 09:37:40 <thermoman> sipa: can i query you?
182 2012-11-09 09:40:56 <sipa> yes
183 2012-11-09 09:54:36 <sipa> thermoman: do you know the transaction's inputs?
184 2012-11-09 10:56:46 <drizztbsd> hi, I have a little question
185 2012-11-09 10:57:01 <drizztbsd> should I rename archlinux bitcoin-daemon package to bitcoind?
186 2012-11-09 10:57:07 <drizztbsd> for coherency?
187 2012-11-09 11:19:02 <Luke-Jr> drizztbsd: probably
188 2012-11-09 14:54:06 <MC1984> crqzy thought
189 2012-11-09 14:54:50 <MC1984> what if standarisation on the OPUS audio codec is a good way to get a national surveillance mechanism for voice going
190 2012-11-09 14:54:53 <MC1984> easy DPI
191 2012-11-09 14:55:59 <MC1984> before i am assuced of tinfoil hats, i live in a country where the government has stated it wants access to communications streams from xbox and other online gaming for 'anti-terrorism'
192 2012-11-09 14:56:39 <MC1984> and has mentioned quote "black boxes" multiple times to that effect
193 2012-11-09 15:00:14 <abrkn> they will control our thoughts through xbox
194 2012-11-09 15:00:20 <abrkn> or maybe they already are
195 2012-11-09 15:05:05 <rdponticelli> They control our thoughts through massive media
196 2012-11-09 15:05:41 <MC1984> no guys really this realates to an actual stated desire by a major western polity
197 2012-11-09 15:06:05 <abrkn> i only read hacker news, which is written 100% by arrogant, 14-year-old python programmers
198 2012-11-09 15:06:27 <abrkn> shit in shit out
199 2012-11-09 15:10:44 <rdponticelli> Whatever anybody can read is more or less, directly or indirectly, heavily exposed to mass media at some stage
200 2012-11-09 15:11:34 <rdponticelli> It's like those "investments" we had, everything in the end was exposed to the big black hole
201 2012-11-09 15:11:41 <rdponticelli> /ot
202 2012-11-09 15:16:16 <rg> lol
203 2012-11-09 15:50:16 <gmaxwell> MC1984: totally OT for #bitcoin-dev
204 2012-11-09 15:52:28 <copumpkin> on-topic?
205 2012-11-09 15:52:31 <copumpkin> ACTION runs
206 2012-11-09 15:52:40 <sipa> Open Transactions>
207 2012-11-09 15:52:51 <gmaxwell> ACTION stabs all
208 2012-11-09 15:52:53 <sipa> ").fixTypo();
209 2012-11-09 16:14:07 <slush1> jgarzik: You were asking about how to do atomic coin swapping. Did you already find the solution?
210 2012-11-09 16:14:29 <slush1> Does it need something in script language,which is not enabled yet?
211 2012-11-09 16:15:24 <helo> swapping between altcoins?
212 2012-11-09 16:15:41 <slush1> helo: no,swapping of some particular coins of bitcoin blockchain, in single transaction
213 2012-11-09 16:15:58 <helo> for colored coin?
214 2012-11-09 16:16:01 <slush1> yes
215 2012-11-09 16:17:44 <sipa> you just need a mapping rule
216 2012-11-09 16:18:11 <sipa> like colors are inherited by equally-numbered outputs as inputs
217 2012-11-09 16:20:06 <slush1> sipa: I know. But how to create one transaction which must be signed by both parties?
218 2012-11-09 16:20:07 <helo> while that may increase fees and help the hashrate, i fear colored coin could devalue bitcoin by making mandatory fees too high for simple currency transactions to be feasible
219 2012-11-09 16:21:02 <Eliel> helo: I don't understand how they could do anything but increase bitcoin's value
220 2012-11-09 16:21:24 <Eliel> why would they increase mandatory fees too high?
221 2012-11-09 16:21:35 <slush1> Eliel: it don't need to increate bitcoin value significantly. You can use just satoshis
222 2012-11-09 16:21:38 <helo> it could fill blocks up
223 2012-11-09 16:21:59 <helo> which will put upward pressure on fees
224 2012-11-09 16:22:05 <slush1> I agree that it can affect tx fees. but this is free market. If people will see colored coins useful, they'll do it
225 2012-11-09 16:23:36 <Eliel> helo: so, basically, it could do the same thing as regular increasing usage.
226 2012-11-09 16:25:36 <helo> i think it may differ a little from simple increasing usage
227 2012-11-09 16:26:37 <helo> or rather, the potential for increased usage from colored bitcoin is a lot higher than increased usage as a currency
228 2012-11-09 16:26:55 <jgarzik> slush: yes, found and posted the solution
229 2012-11-09 16:27:17 <helo> given that one colored satoshi could be valued arbitrarily high
230 2012-11-09 16:27:20 <jgarzik> slush: https://bitcointalk.org/index.php?topic=112007.msg1212356#msg1212356
231 2012-11-09 16:28:09 <helo> so the issue rate of useful units with colored bitcoin is 5,000,000,000 every 10 minutes
232 2012-11-09 16:29:47 <helo> i haven't thought about the demand for currency bitcoin that would be created by colored bitcoin, though... people would demand bitcoin to pay the fees on their colored coin transfers
233 2012-11-09 16:30:11 <Eliel> precisely
234 2012-11-09 16:31:03 <jgarzik> helo: Not just fees... pay bitcoins for colored-coin property
235 2012-11-09 16:31:22 <helo> yeah, the atomic swap would make that pretty compelling...
236 2012-11-09 16:32:49 <helo> as in: holy crap pervasive wired asset zero trust trade
237 2012-11-09 16:33:35 <helo> making all wired property just as non-reversible as bitcoin
238 2012-11-09 16:34:28 <helo> so now my opinion has changed significantly... colored coin w/ atomic swapping could be the end-all killer app for bitcoin
239 2012-11-09 16:35:42 <jgarzik> helo: OTOH, we don't want to bloat the main chain with smart-property tokens, if this becomes super-popular ;p
240 2012-11-09 16:35:58 <jgarzik> helo: essentially turns the block chain into a property registry
241 2012-11-09 16:38:32 <helo> it would really make a double spending not-quite-so-scary
242 2012-11-09 16:39:27 <aurigae1> What is the correct algorithm to calculate worker & pool hashrate? THX
243 2012-11-09 16:39:48 <helo> aurigae1: if you don't get an answer here, try #bitcoin-mining
244 2012-11-09 16:39:55 <aurigae1> thx
245 2012-11-09 16:42:53 <helo> jgarzik: i'm not sure if that would be bad... the minimum value that someone would want to exchange using bitcoin would be higher (given they don't want to pay more than some % of the value in fees), excluding it from use for everyday transactions
246 2012-11-09 16:43:09 <helo> but we already know bitcoin won't scale to allow its use in everyday transactions
247 2012-11-09 16:45:01 <maaku> helo: use #opentransactions; it does the job of managing assets better than colored coins, and doesn't bloat the blockchain
248 2012-11-09 16:45:29 <maaku> keep the block chain for bitcoin (currency) only
249 2012-11-09 16:45:48 <helo> i don't think we get to choose how bitcoin is used :/
250 2012-11-09 16:47:01 <helo> avoiding the addition of coin tracking and atomic swapping to the reference client would go a long way, though
251 2012-11-09 16:47:21 <maaku> sure we do, individually of course
252 2012-11-09 16:47:34 <maaku> don't make the next satoshi-dice bit-spam service
253 2012-11-09 16:48:46 <etotheipi_> request for devs: could someone update the wiki page for BIP 21 to make all the bitcoin: URI examples clickable? It's exactly what I need to test updates to my URL handling, but for some reason that page doesn't have any clickable examples (https://en.bitcoin.it/wiki/BIP_0021)
254 2012-11-09 16:49:29 <slush1> jgarzik: thanks!
255 2012-11-09 16:49:43 <helo> as far as a currency is what people use for relatively low-valued everyday transactions, and that bitcoin can't scale to handle the global volume of such transactions, it kind of makes sense to not worry about applications that create demand for bitcoin-the-currency
256 2012-11-09 16:50:33 <helo> what i was trying to say there was that bitcoin isn't really well suited as an 'everyday currency' due to scalability
257 2012-11-09 16:51:38 <maaku> bitcoin is an excellent clearing-house for off-the-chain services
258 2012-11-09 16:52:05 <maaku> however it's only designed to be as such for one currency/commodity
259 2012-11-09 16:52:45 <maaku> colored coins has the potential to push us against scalability limits, even if only used for clearing-house operations on new assets
260 2012-11-09 16:53:05 <helo> bitcoin's biggest problem is arguably that it solves the zero-trust problem in only one direction, so trust is still very much required
261 2012-11-09 16:53:18 <maaku> open-transactions solves the asset issuance and manipulation problem off-chain, so why use the block chain?
262 2012-11-09 16:53:19 <helo> colored bitcoin would allow it to solve the zero-trust problem bidirectionally
263 2012-11-09 16:53:42 <helo> (with atomic swapping)
264 2012-11-09 16:53:56 <helo> raising transaction fees doesn't make it useless as a currency
265 2012-11-09 16:54:11 <helo> it just raises the minimum value that it is useful for transferring
266 2012-11-09 16:55:09 <gmaxwell> maaku: because opentransactions does not??? rather it solves it far more weakly.
267 2012-11-09 16:55:50 <helo> we want bitcoin to be used widely, and have high fees to keep the hashrate up
268 2012-11-09 16:56:09 <gmaxwell> or to be more specific, it solves _nothing_ because its a toolkit not a solution. But the things you could build out of it in theory have a much weaker non-inflation property than bitcoin does.
269 2012-11-09 16:56:10 <sipa> maaku: OT is centralized but anonymous; bitcoin is decentralized but pseudonymous
270 2012-11-09 16:56:11 <helo> atomic colored bitcoin could do just that
271 2012-11-09 16:56:16 <aurigae1> i still need to know how to calculat the hashrate of my pool and the workers to display it on my website, anybody with a solution???
272 2012-11-09 16:58:01 <Diapolo> sipa: Your suggestion on the allocator stuff would induce the need to change all functions using ``std::vector<unsigned char>&`` into ``std::vector<unsigned char, zero_after_free_allocator<unsigned char> >`` AFAIK!?
273 2012-11-09 16:59:09 <sipa> hmm, perhaps
274 2012-11-09 16:59:14 <Diapolo> e.g. DecodeBase58Check(psz, vchTemp); and the ones that get called from that one on
275 2012-11-09 16:59:26 <sipa> right
276 2012-11-09 16:59:34 <sipa> i'll have a look myself later
277 2012-11-09 16:59:57 <Diapolo> I can try to do this, but you need to carefully look at the pull then :).
278 2012-11-09 17:00:03 <Luke-Jr> maaku: you can run something like OT within the framework of Bitcoin. just have people transfer their Bitcoins to your centralized address and make withdrawls as they like
279 2012-11-09 17:00:25 <maaku> sipa: asset issuance inherently involves centralized trust in the issuer; OT requires trust in the issuer but not the server
280 2012-11-09 17:04:11 <sipa> no trust in the server? can't the server just create as many tokens as it likes?
281 2012-11-09 17:16:21 <Diapolo> sipa: I updated the pull without the discussed changes, you are free to participate or guide me through.
282 2012-11-09 17:17:12 <etotheipi_> sipa, Luke-Jr, gmaxwell, etc: could someone update the wiki page for BIP 21 to make all the bitcoin: URI examples clickable? It has exactly what I need to test updates to my URL handling, but only if they are clickable (https://en.bitcoin.it/wiki/BIP_0021)
283 2012-11-09 17:17:51 <etotheipi_> (I imagine other client devs would appreciate it, too)
284 2012-11-09 17:19:01 <sipa> Diapolo: it was just a suggestion; if it requires too many changes it's not worth it
285 2012-11-09 17:19:30 <sipa> etotheipi_: not sure whether i have edit rights
286 2012-11-09 17:20:11 <etotheipi_> sipa: well I certainly don't
287 2012-11-09 17:20:17 <etotheipi_> (I tried)
288 2012-11-09 17:32:21 <Luke-Jr> etotheipi_: even with edit rights, I don't think the wiki supports bitcoin URIs yet
289 2012-11-09 17:32:45 <sipa> time to poke theymox/MagicalTux then
290 2012-11-09 17:33:55 <Luke-Jr> if it were me, I'd make it autodetect addresses and linkify them :P
291 2012-11-09 17:38:49 <BlueMatt> no one else finds it a bad idea to start linking bitcoin addresses from the wiki page when anyone who signs up for an account can throw in their own address?
292 2012-11-09 17:49:04 <sipa> ha
293 2012-11-09 17:50:39 <Diapolo> I always wanted to change the example address in Qt to mine :-P, no just kidding.
294 2012-11-09 17:50:44 <Luke-Jr> etotheipi_: perhaps most relevant, the URIs in the BIPs should NOT be legal ;)
295 2012-11-09 17:57:15 <etotheipi_> Luke-Jr: why not? (besides the few that aren't supposed to be legal)
296 2012-11-09 17:57:47 <etotheipi_> I imagine that a significant number of users accessing that page have a reason to want to test URIs
297 2012-11-09 17:57:52 <Luke-Jr> etotheipi_: because people might accidentally send to them
298 2012-11-09 17:57:57 <maaku> sipa: the issuer can create as many tokens as they like; OT server != issuer
299 2012-11-09 17:58:19 <maaku> sipa: but that's entirely the point--assets are a form of credit backed by an issuer
300 2012-11-09 17:58:26 <etotheipi_> why is that a concern? someone can click on a bitcoin: link from any page, they still have to confirm it
301 2012-11-09 17:59:28 <Luke-Jr> etotheipi_: they might accidentally confirm it? :p
302 2012-11-09 17:59:42 <sipa> and on a wiki, anyone can change the URI
303 2012-11-09 17:59:48 <etotheipi_> I'm just annoyed that every time I need to test URI handling, I have to go web searching, or through my email to find examples
304 2012-11-09 17:59:48 <sipa> (except on locked pages)
305 2012-11-09 18:00:31 <BlueMatt> etotheipi_: uhh...you should be able to copy/paste examples into your browser's address bar and just hit enter (well...maybe with a down key if you are using chrome)
306 2012-11-09 18:00:58 <etotheipi_> BlueMatt: that's a good point
307 2012-11-09 18:01:02 <gmaxwell> The right thing to do is just use an address with a bad checksum
308 2012-11-09 18:01:31 <gmaxwell> And yea, people will change it... but ::shrugs:: it'll be a fast way of identifying people who need to be banned.
309 2012-11-09 18:01:38 <etotheipi_> I don't understand who we're protecting by saying it's not wise to have valid URIs there
310 2012-11-09 18:02:13 <BlueMatt> its more of a "eh, do we have a valid reason why we need it, because its a fairly minor avenue for abuse"
311 2012-11-09 18:02:16 <etotheipi_> people who are completely clueless shouldn't be on that page anyway, and even if they are... there's multiple opportunities for them to cancel the transaction
312 2012-11-09 18:02:57 <etotheipi_> it's not a major thing... but it seems silly to me to have a webpage talking about standardizing webpage links, and having zero working examples
313 2012-11-09 18:03:13 <BlueMatt> rfcs?
314 2012-11-09 18:03:19 <Luke-Jr> etotheipi_: eligius.st has a few links IIRC
315 2012-11-09 18:03:46 <etotheipi_> I have a few examples of links... but what I like is that the wiki page has a diverse example set
316 2012-11-09 18:03:55 <etotheipi_> including some failures
317 2012-11-09 18:08:22 <Lachesis> so why doesn't bitcoin's encryptwallet feature use public-private key encryption? then there's be no need to worry about maintaining the keypool
318 2012-11-09 18:08:34 <etotheipi_> okay, it's just a recommendation
319 2012-11-09 18:09:59 <etotheipi_> Lachesis: you're mixing up the various cryptography components used in Bitcoin and wallets
320 2012-11-09 18:10:13 <sipa> he is right though, it could have been used
321 2012-11-09 18:10:28 <gmaxwell> sipa: we discussed it.
322 2012-11-09 18:10:34 <sipa> i know
323 2012-11-09 18:10:38 <gmaxwell> as I responded in #bitcoin, Lachesis: because then a hostile party with temporay access to your wallet could fill its keypool with keys he controls.
324 2012-11-09 18:10:40 <Lachesis> sorry for starting a cross-channel discussion
325 2012-11-09 18:10:46 <gmaxwell> Lachesis: and then your change will happily go to him.
326 2012-11-09 18:10:48 <gmaxwell> I admit, a minor concern... but the simpler implementation avoids it.
327 2012-11-09 18:12:22 <etotheipi_> okay, here's an idea I'm pondering, I'd like feedback -- a "secure" plugin mechanism for Armory (useful on other apps, too)
328 2012-11-09 18:12:59 <etotheipi_> there's a plugin directory where users can dump .py files containing new "tabs" for the main window
329 2012-11-09 18:13:23 <sipa> how much access do these plugins have to the core/
330 2012-11-09 18:13:26 <etotheipi_> Armory would auto-detect the ones in that directory, and add a tab for each one, reading that .py file to know what code to run to display the tag
331 2012-11-09 18:13:48 <etotheipi_> theoretically the plugins could do anything Armory could do... but that's up for debate
332 2012-11-09 18:13:51 <etotheipi_> the key is this:
333 2012-11-09 18:14:04 <sipa> etotheipi_: btw, how much does armory still depend on bitcoind?
334 2012-11-09 18:14:19 <etotheipi_> there must be an associated plugin.py.signature file with the .py file
335 2012-11-09 18:14:23 <sipa> haven't used it in a while
336 2012-11-09 18:14:29 <etotheipi_> if a plugin is detected on load, it will check for the signature file
337 2012-11-09 18:14:55 <etotheipi_> for non-official plugins, they wouldn't be loaded until the user explicitly signs it (Armory will prompt them)
338 2012-11-09 18:15:03 <etotheipi_> it will be signed with the root key of one of their wallets
339 2012-11-09 18:15:44 <etotheipi_> or, plugins that devs (me) supports can be signed and one of the signing keys to check
340 2012-11-09 18:16:37 <sipa> that also means someone can repackage the program, with a pre-signed malicious plugin
341 2012-11-09 18:16:47 <etotheipi_> sipa: but that's always a risk, anyway
342 2012-11-09 18:16:48 <sipa> but that's something a repackager can always do of course
343 2012-11-09 18:17:22 <etotheipi_> if the attacker has root access, they can swap out the armory installation with anything they want
344 2012-11-09 18:17:36 <etotheipi_> this wouldn't increase that aspect of the attack surface
345 2012-11-09 18:17:56 <etotheipi_> but it does prevent someone who's not so resourceful, from just dumping malicious plugins into your plugin dir
346 2012-11-09 18:18:19 <sipa> well i'd mostly fear social engineering to make people sign malicious things
347 2012-11-09 18:18:29 <etotheipi_> on the other hand, if they have that level of access, they can probably get your encryption key to sign it for you
348 2012-11-09 18:18:31 <sipa> but that's always a risk with third party plugins
349 2012-11-09 18:19:32 <etotheipi_> I'm not entirely clear how this would work if the user doesn't have any full wallets on their system
350 2012-11-09 18:19:42 <etotheipi_> they'd have to sign the plugin from their offline machine
351 2012-11-09 18:19:49 <sipa> etotheipi_: still, how much does armory still depend on bitcoidn?
352 2012-11-09 18:19:55 <etotheipi_> sipa: still completely
353 2012-11-09 18:20:08 <sipa> you're aware of the upcoming changes for 0.8?
354 2012-11-09 18:20:15 <etotheipi_> I have made no effort to re-implement the networking engine for Armory
355 2012-11-09 18:20:21 <etotheipi_> sipa: what changes are those?
356 2012-11-09 18:20:35 <etotheipi_> sipa: also note that Armory connects to Bitcoin-Qt/bitcoind as a peer
357 2012-11-09 18:20:40 <etotheipi_> so any RPC changes won't affect it
358 2012-11-09 18:20:46 <Luke-Jr> hmm, speaking of 0.8, is append-only HD wallets still planned for it? :o
359 2012-11-09 18:20:59 <sipa> etotheipi_: https://bitcointalk.org/index.php?topic=119525.0
360 2012-11-09 18:21:02 <Luke-Jr> etotheipi_: blkindex is gone etc
361 2012-11-09 18:21:05 <etotheipi_> oooh, blkxxx.dat files
362 2012-11-09 18:21:38 <sipa> the blk000??.dat files have the same format as before, but they're smaller, different names, and in a subdir
363 2012-11-09 18:21:52 <etotheipi_> sipa: that's excellent to know
364 2012-11-09 18:22:06 <Luke-Jr> sipa: really? same format? :/
365 2012-11-09 18:22:08 <etotheipi_> Armory handles any number and size of them... but having the correct path would be useful :)
366 2012-11-09 18:22:21 <sipa> Luke-Jr: because of compatibility issues, yes
367 2012-11-09 18:22:35 <Luke-Jr> sipa: didn't jgarzik have some simple compatible change that would enable sendfile?
368 2012-11-09 18:22:37 <etotheipi_> sipa: is it still a 4-digit numbre in there?
369 2012-11-09 18:22:43 <sipa> yes
370 2012-11-09 18:22:47 <sipa> Luke-Jr: in picocoin, yes
371 2012-11-09 18:22:58 <Luke-Jr> sipa: why not use it for bitcoind as well?
372 2012-11-09 18:23:09 <etotheipi_> so I relaly just have to convert "path/to/bitcoin/blkxxxx.dat" to "path/to/bitcoin/blocks/blkxxxx.dat"
373 2012-11-09 18:23:27 <sipa> etotheipi_: 5 digits instead of 4, and starts at 0
374 2012-11-09 18:23:34 <sipa> Luke-Jr: meg
375 2012-11-09 18:23:36 <sipa> Luke-Jr: meh
376 2012-11-09 18:23:46 <etotheipi_> sipa: will there be a transition?
377 2012-11-09 18:23:48 <Luke-Jr> sipa: better than changing it later IMO
378 2012-11-09 18:23:51 <sipa> sendfile works just as fine now, you just have to send the header separately
379 2012-11-09 18:23:55 <Luke-Jr> hmm
380 2012-11-09 18:24:20 <sipa> it's not worth breaking compatibility for
381 2012-11-09 18:24:23 <Luke-Jr> ok
382 2012-11-09 18:24:41 <Luke-Jr> I don't see how it'd break compatibility, but if it works anyway might as well save the disk space
383 2012-11-09 18:24:41 <sipa> there's something nice about being able to use block files and bootstrap.dat interchangeable between all versions
384 2012-11-09 18:24:42 <Luke-Jr> :P
385 2012-11-09 18:24:49 <etotheipi_> will 0.8 clients convert their blk files to the one, or keep using the old system?
386 2012-11-09 18:25:06 <sipa> etotheipi_: probably just move them, and reindex
387 2012-11-09 18:25:11 <sipa> but that's not implemented yet
388 2012-11-09 18:25:27 <sipa> Luke-Jr: if i put some effort into it, i'm sure i can make block files 30-40% smaller
389 2012-11-09 18:25:37 <sipa> Luke-Jr: that'd be worth breaking compatibility for, imho
390 2012-11-09 18:25:55 <etotheipi_> so some users who currently have blk0001.dat and blk0002.dat will now have blocks/blk00000.dat (2GB), blocks/blk00001.dat (2GB), and then add 128 MB files to that?
391 2012-11-09 18:26:00 <sipa> yes
392 2012-11-09 18:26:10 <sipa> well, maybe
393 2012-11-09 18:26:14 <sipa> still to be discussed, i guess
394 2012-11-09 18:26:15 <etotheipi_> haha
395 2012-11-09 18:26:24 <etotheipi_> sounds like there's no imminent threat of it
396 2012-11-09 18:26:32 <Luke-Jr> sipa: it might make sense to copy the data just to defragment
397 2012-11-09 18:26:45 <sipa> Luke-Jr: if you're going to copy, you can just use loadblock
398 2012-11-09 18:26:46 <etotheipi_> I will update my search function to look for blk0001+, and if it's not there, it will look for blocks/blk00000+
399 2012-11-09 18:26:52 <Luke-Jr> sipa: sure
400 2012-11-09 18:27:08 <etotheipi_> think that's safe?
401 2012-11-09 18:27:10 <Luke-Jr> etotheipi_: I'd do the inverse. Some people may keep the old data.
402 2012-11-09 18:27:10 <sipa> which has many advantages, but also requires temporarily 2x as much storage space
403 2012-11-09 18:27:17 <etotheipi_> oh, look for the new data first
404 2012-11-09 18:27:20 <etotheipi_> makes sense
405 2012-11-09 18:27:29 <sipa> agree, first check new data
406 2012-11-09 18:27:42 <sipa> or you could judge by mtime
407 2012-11-09 18:27:50 <Luke-Jr> or load both
408 2012-11-09 18:27:52 <Luke-Jr> :p
409 2012-11-09 18:28:19 <Luke-Jr> ACTION grumbles about testnet's target being too high
410 2012-11-09 18:29:20 <sipa> about HD wallets / append-only wallet format... whatever I (or someone else) manages to implement, I guess
411 2012-11-09 18:29:31 <sipa> having a daytime job doesn't help :)
412 2012-11-09 18:30:05 <Luke-Jr> ACTION gets sipa fired from dayjob.
413 2012-11-09 18:30:06 <Luke-Jr> j/k ;)
414 2012-11-09 18:30:13 <etotheipi_> oh, I'm not the only one with a real job, here? :)
415 2012-11-09 18:30:27 <etotheipi_> or do you all chat on IRC from work?
416 2012-11-09 18:30:38 <sipa> it's 8:30 pm here...
417 2012-11-09 18:30:40 <Luke-Jr> etotheipi_: I control when/how I work. ;)
418 2012-11-09 18:32:16 <sipa> hmm, idea... if blocks/ exists, but blktree/ doesn't, I could run a -reindex automatically
419 2012-11-09 18:32:48 <sipa> so you could just drop an archived set of blk*.dat files in your datadir, and have them imported with zero additional storage
420 2012-11-09 18:34:10 <D34TH> sipa: sounds good
421 2012-11-09 18:34:11 <D34TH> :D
422 2012-11-09 18:34:40 <D34TH> but will it blend
423 2012-11-09 18:35:12 <sipa> it's already the case that if you have blocks/ + matching blktree/, but no coins/ is not up-to-date, it gets updated at startup automatically
424 2012-11-09 18:35:21 <sipa> -no
425 2012-11-09 18:35:55 <sipa> i like having these "if it's broken, delete it, and it gets regenerated" practices
426 2012-11-09 18:36:23 <D34TH> question, whats the difference between the .sst and the .log in coins/
427 2012-11-09 18:36:35 <sipa> .sst as sstables, .log are logs :)
428 2012-11-09 18:36:43 <sipa> .sst's are sorted, .log's are not
429 2012-11-09 18:37:08 <D34TH> ahh, because i was going to say the log looks just like the ssts
430 2012-11-09 18:37:12 <D34TH> completly unreadable
431 2012-11-09 18:37:19 <sipa> .log's get written synchronously when updates are written, and they are converted to .sst's in the background
432 2012-11-09 18:37:37 <sipa> and then .sst's get merged/rewritten when necessary
433 2012-11-09 18:37:54 <D34TH> so its more a tx log
434 2012-11-09 18:37:58 <D34TH> than a reading log
435 2012-11-09 18:38:25 <sipa> it's a transaction log, yes
436 2012-11-09 18:38:41 <D34TH> that answers my question, thanks
437 2012-11-09 18:39:37 <jgarzik> Luke-Jr, sipa: pynode stores blocks a P2P "block" messages, suitable for direct sendfile(2). However, it is just as valid for bitcoind to send the P2P message header via send(2) with MSG_MORE flag, then calling sendfile(2) on the block data itself.
438 2012-11-09 18:39:53 <jgarzik> picocoin does not store full blocks, so not really applicable to picocoin.
439 2012-11-09 18:39:58 <sipa> oh, pynode
440 2012-11-09 18:40:02 <jgarzik> picocoin just stores merkle tx's.
441 2012-11-09 18:40:08 <sipa> ACTION confuses all these jgarzik nodes
442 2012-11-09 18:40:11 <jgarzik> ;p
443 2012-11-09 18:40:39 <D34TH> useragent("jgarzik/picocoin")
444 2012-11-09 18:40:40 <D34TH> :D
445 2012-11-09 18:41:01 <sipa> useragent("matrix/smith")
446 2012-11-09 18:41:19 <D34TH> Mr. Anderson, we meet again
447 2012-11-09 18:41:27 <sipa> surprised to see me?
448 2012-11-09 18:41:41 <D34TH> :3
449 2012-11-09 18:45:09 <Luke-Jr> ACTION wonders if pynode shows up on the client list yet
450 2012-11-09 18:45:44 <Luke-Jr> ACTION ponders wtf /bitsofproof:0.5/ is
451 2012-11-09 18:46:10 <jgarzik> Luke-Jr: new full node client from grau(sp?)
452 2012-11-09 18:46:10 <Luke-Jr> no sign of pynode yet :\\
453 2012-11-09 18:46:23 <D34TH> Luke-Jr, https://bitcointalk.org/index.php?topic=122013.0
454 2012-11-09 18:46:28 <jgarzik> Luke-Jr: pynode does not listen, only connects out
455 2012-11-09 18:46:50 <Luke-Jr> aha
456 2012-11-09 18:47:03 <sipa> i'm really glad to see some many active listening ipv6 nodes
457 2012-11-09 18:47:11 <sipa> >350 in my seeds.txt
458 2012-11-09 18:48:25 <gmaxwell> seems almost suspect!
459 2012-11-09 18:48:41 <D34TH> ACTION grabs his box of tinfoil hats
460 2012-11-09 18:48:44 <D34TH> 1 btc per hat
461 2012-11-09 18:48:48 <D34TH> orderly line please
462 2012-11-09 18:49:51 <jgarzik> in many cases, it is easier to have ipv6 listening nodes, than ipv4.
463 2012-11-09 18:50:47 <sipa> however, my onion node on my laptop doesn't seem to get detected by my seeder
464 2012-11-09 18:51:05 <jgarzik> thanks to amiller, pynode was upgraded from asyncore shite, so it just required coder time to make pynode a full node.
465 2012-11-09 18:51:11 <jgarzik> still lacks all the DoS protections
466 2012-11-09 18:51:41 <jgarzik> some of the "DoS protections" are in actuality necessities for pynode, e.g. signature cache, otherwise performance is slow.
467 2012-11-09 18:52:03 <sipa> that's not a DoS protections; that's a performance optimization
468 2012-11-09 18:54:36 <jgarzik> sipa: Sergio Damian Lerner (sp?) considered it DoS protection as well, at least
469 2012-11-09 18:54:44 <jgarzik> ACTION doesn't care about labelling, call it what you will
470 2012-11-09 18:55:21 <sipa> good point
471 2012-11-09 18:58:15 <sipa> damn... i've wondered for a long time what (sp?) means... it just means (spelling?) ?
472 2012-11-09 18:58:17 <Luke-Jr> slush1: ping
473 2012-11-09 18:58:25 <Luke-Jr> sipa: yes
474 2012-11-09 18:58:32 <Lachesis> sipa, yeppers
475 2012-11-09 18:58:42 <Lachesis> sipa, i'm an ipv6 node :)
476 2012-11-09 18:59:06 <slush1> Luke-Jr: yes?
477 2012-11-09 18:59:33 <Luke-Jr> slush1: set_difficulty allows any Number, or only integers?
478 2012-11-09 19:00:09 <forrestv> ACTION wishes that stratum used normal hex targets instead of "difficulty"
479 2012-11-09 19:00:35 <Luke-Jr> ^
480 2012-11-09 19:00:37 <slush1> Luke-Jr: good question. For some reasons, after discussion with ckolivas and Eleuthria I limited it to integers only, but now I cannot remember the reason
481 2012-11-09 19:00:41 <Luke-Jr> forrestv: bring it up during the BIP process :D
482 2012-11-09 19:00:54 <Luke-Jr> slush1: so there's no way to request non-truncated shares?
483 2012-11-09 19:01:09 <slush1> what do you mean by non-truncated shares?
484 2012-11-09 19:01:24 <Luke-Jr> slush1: full true-bits after the initial zero-bits
485 2012-11-09 19:02:04 <Luke-Jr> eg, pdiff 1 = 32 false-bits + 224 true-bits
486 2012-11-09 19:02:04 <slush1> Luke-Jr: well, my mining proxy and poclbm accepts difficulty with decimal places, but ckolivas rejected it :)
487 2012-11-09 19:02:14 <slush1> as far as I remember because of implementation details in C
488 2012-11-09 19:02:32 <slush1> (dividing big numbers with float precision)
489 2012-11-09 19:03:00 <sipa> using difficulty in any format always requires a division, and that seems like asking from problems
490 2012-11-09 19:03:02 <slush1> and I didn't see a reason for full precision, so I accepted it
491 2012-11-09 19:03:31 <slush1> actually why we need so precisious difficulty?
492 2012-11-09 19:03:55 <Luke-Jr> well, if it works in the reference proxy, I'll just consider that standard for now
493 2012-11-09 19:04:02 <Luke-Jr> slush1: otherwise miners will lose shares
494 2012-11-09 19:04:08 <slush1> how?
495 2012-11-09 19:04:22 <Luke-Jr> they won't submit 00000000FFFF1FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
496 2012-11-09 19:05:21 <TD> ACTION wonders what is up with people proclaiming their bitcoin reimplementations to be the dawn of a new age
497 2012-11-09 19:05:24 <slush1> afaik diff1 == 0x00000000ffff0000....
498 2012-11-09 19:05:32 <slush1> so yes, such share won't be accepted for difficulty 1
499 2012-11-09 19:05:37 <sipa> TD: haha
500 2012-11-09 19:06:20 <sipa> TD: are you referring to cbitcoin or bitsofproof?
501 2012-11-09 19:06:24 <TD> both
502 2012-11-09 19:06:27 <TD> it seems to be a trend
503 2012-11-09 19:07:09 <Luke-Jr> slush1: bdiff1 = 0x00000000ffff0000???, but pdiff1 = 0x00000000ffffffff???
504 2012-11-09 19:07:16 <slush1> what the hell is pdiff?
505 2012-11-09 19:07:23 <Luke-Jr> pool difficulty
506 2012-11-09 19:07:32 <slush1> what pool difficulty?
507 2012-11-09 19:07:41 <Luke-Jr> the one pools have used since you started yours 2 years ago
508 2012-11-09 19:07:43 <Luke-Jr> :p
509 2012-11-09 19:07:52 <sipa> just have a well-defined difficulty
510 2012-11-09 19:07:53 <slush1> yes, I made a mistake and everybody copied it from me :-D
511 2012-11-09 19:08:04 <Luke-Jr> slush1: no, it's well-reasoned!
512 2012-11-09 19:08:09 <Luke-Jr> it's easier to check pdiffs
513 2012-11-09 19:08:15 <slush1> 0x00000000ffff0000 is exactly diff1
514 2012-11-09 19:08:33 <Luke-Jr> 0x00000000ffff0000 is pdiff1 passed through bitcoin's custom floating point
515 2012-11-09 19:08:54 <Luke-Jr> there's no reason to use it for pools
516 2012-11-09 19:09:40 <slush1> well, what means "difficulty 1" exactly?
517 2012-11-09 19:09:52 <sipa> it means whatever you define it to mean
518 2012-11-09 19:10:01 <sipa> but do define it :)
519 2012-11-09 19:10:03 <slush1> or, better, where's defined that diff1 is 0x00000000fffffffffffff....
520 2012-11-09 19:10:09 <Luke-Jr> slush1: it could mean pdiff1 or bdiff1 :p
521 2012-11-09 19:10:26 <slush1> I never heard pdiff/bdiff, where did you find these terms?
522 2012-11-09 19:10:33 <sipa> slush1: in his head
523 2012-11-09 19:10:39 <sipa> do people really use "difficulty 1" when referring to 0x00000000FFFFFFFF... ?
524 2012-11-09 19:10:42 <Luke-Jr> yes
525 2012-11-09 19:10:57 <slush1> sipa: yes, because I made a mistake. no other reason
526 2012-11-09 19:11:07 <sipa> well, doesn't matter
527 2012-11-09 19:11:12 <Luke-Jr> sipa: it's insanely close
528 2012-11-09 19:11:12 <sipa> just specify what you mean
529 2012-11-09 19:11:16 <sipa> i know
530 2012-11-09 19:11:35 <Luke-Jr> slush1: for performance too
531 2012-11-09 19:11:42 <slush1> when I built first pool code, I just wanted to make checks easy, so I checked first eight zeros
532 2012-11-09 19:11:47 <slush1> obviously a hack
533 2012-11-09 19:11:52 <Luke-Jr> checking count of false-bits is faster than doing bignum math
534 2012-11-09 19:12:03 <sipa> Luke-Jr: no need to explain math to me, but i'd certainly call it difficulty 0.999985 :)
535 2012-11-09 19:12:19 <slush1> Luke-Jr: it works good, unless you want dynamic difficulty ;)
536 2012-11-09 19:12:28 <Luke-Jr> slush1: even with dynamic difficulty!
537 2012-11-09 19:12:32 <slush1> and diff1 pools are going to die
538 2012-11-09 19:12:45 <Luke-Jr> slush1: I'm storing share difficulties as "number of false-bits minus 32"
539 2012-11-09 19:13:16 <slush1> I see this as an over optimization
540 2012-11-09 19:13:23 <slush1> doing "full check" is super cheap
541 2012-11-09 19:14:17 <Luke-Jr> not in terms of coding time
542 2012-11-09 19:14:24 <Luke-Jr> *unless you're using Python
543 2012-11-09 19:14:48 <slush1> or any other sane language
544 2012-11-09 19:14:57 <slush1> like Scala :)
545 2012-11-09 19:16:00 <Luke-Jr> respect teh C!
546 2012-11-09 19:16:01 <slush1> well, ckolivas solved this by not doing full check on all bits, but he took only 64 bits after initial 32bits of zeroes
547 2012-11-09 19:16:06 <slush1> gives "good enough" math, even in C
548 2012-11-09 19:16:35 <slush1> I see 64bit precision good enough even for pools
549 2012-11-09 19:18:37 <Luke-Jr> slush1: you may be right on this, but it's still annoying that stratum is tied to integer bdiffs. I guess that can be discussed in more depth after the initial draft is done though
550 2012-11-09 19:19:10 <sipa> i really consider that a detail
551 2012-11-09 19:19:20 <sipa> if it's well-defined, there is no problem
552 2012-11-09 19:19:57 <slush1> actually the only reason again integer difficulty is support for alt chains (using <diff1 shares)
553 2012-11-09 19:20:00 <slush1> but...
554 2012-11-09 19:20:47 <Luke-Jr> slush1: ?
555 2012-11-09 19:21:09 <slush1> some alt chains are using difficulty below 1
556 2012-11-09 19:21:33 <slush1> obviously it's impossible to define so small target with integer difficulty.
557 2012-11-09 19:21:48 <slush1> But it's not a problem for bitcoin, so not a problem for me
558 2012-11-09 19:21:55 <Luke-Jr> slush1: btw, you know stratum isn't JSON-RPC 2.0 compatible?
559 2012-11-09 19:22:15 <sipa> i didn't know stratum was even JSON-RPC
560 2012-11-09 19:22:25 <Luke-Jr> lol
561 2012-11-09 19:22:33 <slush1> yes :) I forgot in which part, but I know there's some difference
562 2012-11-09 19:22:42 <Luke-Jr> slush1: JSON-RPC 2.0 doesn't allow bidirectional
563 2012-11-09 19:22:45 <sipa> afaik, it's just JSON over TCP?
564 2012-11-09 19:22:59 <D34TH> whoop whoop, got the full pynode working on win32
565 2012-11-09 19:23:01 <slush1> Luke-Jr: oh, that's something different
566 2012-11-09 19:23:06 <Luke-Jr> sipa: JSON-RPC defines the format of the JSON stuff, it doesn't require HTTP
567 2012-11-09 19:23:11 <slush1> Luke-Jr: can you send me a link to this statement?
568 2012-11-09 19:24:07 <etotheipi_> so does the rest of the world hate the US as much as the US hates itself?
569 2012-11-09 19:24:47 <slush1> Luke-Jr: oh, I remember. It is different in error handling. JSON-RPC defines error as object, but I defined it as 3-tuple
570 2012-11-09 19:25:18 <slush1> Luke-Jr: but I cannot find anything about that bi-directional statement
571 2012-11-09 19:25:25 <Luke-Jr> slush1: the spec talks about client-server stuff, I'm not sure on the original source of the bidirectional thing
572 2012-11-09 19:25:56 <Luke-Jr> though I'm noticing that notify should not have any result, not even null, and "jsonrpc":"2.0" is required
573 2012-11-09 19:26:20 <slush1> yes, that's another thing. passing "jsonrpc": "2.0" for every method seems incredibly stupid
574 2012-11-09 19:26:38 <D34TH> should be assumed by default
575 2012-11-09 19:26:47 <D34TH> pass 1.x if not
576 2012-11-09 19:26:48 <D34TH> :D
577 2012-11-09 19:28:10 <slush1> well, maybe I should say it is json-rpc 1.1 compatible, and problem is gone :)
578 2012-11-09 19:28:15 <Luke-Jr> ;)
579 2012-11-09 19:28:32 <D34TH> i totally want https://github.com/bitcoin/bitcoin/pull/1549 pulled
580 2012-11-09 19:28:52 <Luke-Jr> D34TH: that was planned for 0.7 originally, but fell through the cracks XD
581 2012-11-09 19:28:59 <D34TH> i see
582 2012-11-09 19:29:03 <D34TH> now its for 0.8.0
583 2012-11-09 19:29:26 <D34TH> would be very handy
584 2012-11-09 19:29:38 <D34TH> now only if pynode also had that
585 2012-11-09 19:31:33 <Luke-Jr> slush1: one problem I have having understanding is how to multiplex miners on a single stratum link
586 2012-11-09 19:32:17 <slush1> Luke-Jr: any specific problem?
587 2012-11-09 19:32:41 <Luke-Jr> slush1: no, I don't see how it's possible at all
588 2012-11-09 19:33:03 <Luke-Jr> I would have expected a unique extranonce1 per miner, but I don't see that in the spec
589 2012-11-09 19:33:08 <slush1> maybe I don't understand what you mean by "multiples on a single link"
590 2012-11-09 19:33:23 <slush1> no, there's just one subscription
591 2012-11-09 19:33:54 <slush1> but the client can add some unique byte to extranonce1 and lower extranonce2_size, to every sub-miner
592 2012-11-09 19:33:55 <Luke-Jr> slush1: different users on the same stratum connection
593 2012-11-09 19:34:07 <Luke-Jr> o
594 2012-11-09 19:34:27 <jgarzik> D34TH: neat! Did pynode require any modifications, to work on Windows?
595 2012-11-09 19:34:32 <D34TH> nope
596 2012-11-09 19:34:40 <jgarzik> D34TH: picocoin would, sadly, require replacement of fork(2) to work on Windows.
597 2012-11-09 19:34:49 <slush1> Luke-Jr: for example proxy is now adding one byte extra, allowing 255 sub-miners
598 2012-11-09 19:34:55 <D34TH> i used my libleveldb from my bitcoin pullreq
599 2012-11-09 19:35:06 <D34TH> snappy was a little weird to i used that
600 2012-11-09 19:35:27 <slush1> but I'm thinking to raise make extranonce2_size on the pool, eight seems a bit low
601 2012-11-09 19:36:03 <jgarzik> slush: the main thing JSON-RPC 2.0 adds is permitting batches of requests, and batches of responses, in a single message. You pass an array at the top-level object, containing a list of JSON-RPC requests. The server responds with an array of JSON-RPC responses. Very useful for reducing round-trip times, if you have multiple requests.
602 2012-11-09 19:36:10 <jgarzik> slush1: ^
603 2012-11-09 19:36:22 <jgarzik> so batch support would be nice
604 2012-11-09 19:36:28 <slush1> jgarzik: oh, so I should redefine it as json-rpc 1.1 compatible
605 2012-11-09 19:36:53 <slush1> I see batches as useful for HTTP
606 2012-11-09 19:37:05 <slush1> but it just adds unnecessary complexity for tcp streams...
607 2012-11-09 19:37:11 <Luke-Jr> yeah, not sure if batches makes sense for TCP streams
608 2012-11-09 19:37:28 <jgarzik> TCP or HTTP, batches still make sense
609 2012-11-09 19:37:39 <jgarzik> of course, "makes sense" depends on your usage
610 2012-11-09 19:37:54 <slush1> especially for mining, you want to submit share as fast as you can, not cache shares and send them in batch...
611 2012-11-09 19:38:06 <jgarzik> sure
612 2012-11-09 19:38:27 <jgarzik> but the stratum protocol is proposed for other uses than just fast mining, I thought?
613 2012-11-09 19:38:34 <D34TH> i like how bitcoin only uses ~175 mb ram now
614 2012-11-09 19:39:03 <Luke-Jr> jgarzik: how do batches make sense for TCP?
615 2012-11-09 19:39:07 <slush1> jgarzik: yes, but still, I cannot imagine useful use-case anyway
616 2012-11-09 19:39:51 <slush1> it's still about some balance. of course, it's possible to compress json overhead by batches
617 2012-11-09 19:41:00 <jgarzik> Luke-Jr: reduces overhead, uses fewer packets, versus pipelined requests of same
618 2012-11-09 19:41:28 <jgarzik> if your protocol has transactional features, or you need to request a great many small items, bundling may help on server side too
619 2012-11-09 19:43:28 <slush1> jgarzik: is a batch considered as a transaction? I think that implementing transactions accross multiple RPC calls is far from trivial...
620 2012-11-09 19:43:52 <jgarzik> slush1: it is what you make of it, like any JSON. it can, or cannot be, as you define.
621 2012-11-09 19:44:20 <jgarzik> "batch" is simply aggregating multiple json-rpc requests into a single array
622 2012-11-09 19:44:26 <jgarzik> (ditto responses)
623 2012-11-09 19:44:36 <D34TH> needs moar brackets
624 2012-11-09 19:45:07 <slush1> I understand. Still, I hope I won't need transactions for long time ;)
625 2012-11-09 19:45:25 <slush1> need to implement, I mean
626 2012-11-09 20:34:33 <Luke-Jr> slush: seems to me set_difficulty and multiple users don't work together in theory?
627 2012-11-09 20:34:59 <sipa> ;;bc,block
628 2012-11-09 20:35:00 <gribble> Error: "bc,block" is not a valid command.
629 2012-11-09 20:35:00 <sipa> ;;bc,blocks
630 2012-11-09 20:35:02 <gribble> 207241
631 2012-11-09 20:36:22 <sipa> ok, -reindex with default cache size and sigchecking disabled: 13 minutes
632 2012-11-09 20:51:10 <sipa> Luke-Jr: any interest in updating your full hashes in debug.log patch?
633 2012-11-09 20:51:32 <Luke-Jr> sipa: blah, it needs rebasing? :|
634 2012-11-09 20:52:01 <Luke-Jr> sipa: I'll try to get to it later tonight maybe
635 2012-11-09 20:52:20 <sipa> probably easier to rewrite from scratch... i wouldn't even bother with the helper functions... hash.ToString.c_str() works fine :)
636 2012-11-09 21:03:49 <Diapolo> Is there any release schedule for 0.8?
637 2012-11-09 21:04:44 <sipa> no feature freeze yet
638 2012-11-09 21:05:14 <sipa> as it may still take some time for the recent changes to mature
639 2012-11-09 21:06:49 <Diapolo> that's fine, what are the biggest things not yet merged or created to make it into 0.8 from your point of view?
640 2012-11-09 21:07:50 <sipa> auto-upgrading old datadir, bloom filtering
641 2012-11-09 21:09:01 <Diapolo> oh btw. I would love to see your remove detach pull get merged ^^
642 2012-11-09 21:09:17 <sipa> it's not merged yet?
643 2012-11-09 21:09:29 <Diapolo> there are many ACKed things not merged
644 2012-11-09 21:10:10 <sipa> Diapolo: if you want to hack on db.{cpp,h}, a large part of it can be removed now that's only used for wallets
645 2012-11-09 21:11:22 <sipa> there you go
646 2012-11-09 21:11:26 <Diapolo> I just stumpled upon that part and that function is still in use ^^ (::Open()).
647 2012-11-09 21:11:49 <Diapolo> Thought it's a good thing to simplify that many path variables ^^.
648 2012-11-09 21:12:38 <Diapolo> sipa: thanks
649 2012-11-09 21:12:57 <Diapolo> You are able to run and compile the Qt version right? Can you test a small thing?
650 2012-11-09 21:15:48 <Diapolo> sipa: https://github.com/bitcoin/bitcoin/blob/master/src/init.cpp#L514 insert a "return false;" there and start Qt, what happens for you?
651 2012-11-09 21:22:40 <sipa> in current head?
652 2012-11-09 21:22:57 <sipa> probably not...
653 2012-11-09 21:25:02 <sipa> Diapolo: where exactly? 514 is currently inside the handling of -salvagewallet
654 2012-11-09 21:26:05 <Diapolo> right
655 2012-11-09 21:26:35 <Diapolo> exactly after the first part there which calls ::Open()
656 2012-11-09 21:27:02 <sipa> I see no ::Open in init.cpp
657 2012-11-09 21:27:36 <sipa> you mean bitdb.Open(GetDataDir()) ?
658 2012-11-09 21:28:36 <Diapolo> return false;
659 2012-11-09 21:29:11 <sipa> ok, building
660 2012-11-09 21:29:51 <Diapolo> this triggers a Data Execution Prevention Exception for me ... I want to know if that is my build or Windows only
661 2012-11-09 21:31:52 <sipa> just quits; no problems
662 2012-11-09 21:32:16 <Diapolo> strange ...
663 2012-11-09 21:33:38 <Diapolo> I found that "bug?" by setting an invalid proxy, which triggers it even with official 0.7.1 because of return InitError();
664 2012-11-09 21:44:19 <Diapolo> I disabled the DEP protection and get a stack related crash of Bitcoin-Qt ...
665 2012-11-09 21:44:42 <Diapolo> seems the DEP one is safer anyway, but there is something not right
666 2012-11-09 21:45:25 <edcba> why is there a need for a proxy ?
667 2012-11-09 21:45:27 <sipa> i ran bitcoin-qt after that extra return false under valgrind, and couldn't even see any unexpected warnings
668 2012-11-09 21:46:02 <Diapolo> sipa: seems Windows related, unsure about Mac but not affects Linux
669 2012-11-09 21:46:25 <Diapolo> edcba: that was just to trigger an startup abort
670 2012-11-09 21:46:37 <Diapolo> Are you using Windows :)?
671 2012-11-09 21:54:21 <Guest73974> Project Bitcoin build #133: FAILURE in 39 min: http://jenkins.bluematt.me/job/Bitcoin/133/
672 2012-11-09 21:56:43 <Diapolo> at least my debugger seems to work I'll look into it
673 2012-11-09 22:00:51 <x18882> hi
674 2012-11-09 22:01:42 <x18882> is it possible to run bitcoind with no blockchain? kind of in a "relay-only" mode?
675 2012-11-09 22:03:13 <x18882> the idea is to run it to help the bitcoin network, even if I don't have enough storage or bandwidth to store the blockchain for now
676 2012-11-09 22:05:00 <phantomcircuit> x18882, that doesn't really help actually
677 2012-11-09 22:05:07 <phantomcircuit> you just become yet another consumer
678 2012-11-09 22:06:39 <Diapolo> sipa: the crash comes from backup_holder, which is a class of Boost variant...
679 2012-11-09 22:13:02 <sipa> there, build fixed
680 2012-11-09 22:21:23 <Diapolo> sipa: you keep pushing, nice :D
681 2012-11-09 22:27:08 <D34TH> i like how simple blockchain backups can be now
682 2012-11-09 22:27:13 <D34TH> just upload the dats
683 2012-11-09 22:27:17 <D34TH> and it will rebuild it
684 2012-11-09 22:27:27 <D34TH> instead of 2gb tar.gz
685 2012-11-09 22:31:41 <Diapolo> sipa: What about my explicit bind pull? I'm tired to still see it open, counter-motivating as you ACKed some time ago.
686 2012-11-09 22:35:20 <Guest73974> Project Bitcoin build #134: STILL FAILING in 40 min: http://jenkins.bluematt.me/job/Bitcoin/134/
687 2012-11-09 22:36:00 <D34TH> so it nickname
688 2012-11-09 22:36:01 <D34TH> D:
689 2012-11-09 22:36:47 <D34TH> test/miner_tests.cpp:75: void miner_tests::CreateNewBlock_validity::test_method(): Assertion `ProcessBlock(__null, pblock)'
690 2012-11-09 22:36:49 <D34TH> hmm
691 2012-11-09 22:37:04 <sipa> D34TH: it's already fixed
692 2012-11-09 22:37:22 <D34TH> D:
693 2012-11-09 22:37:30 <D34TH> i was going to look into it too
694 2012-11-09 22:37:39 <D34TH> there goes my nice deed
695 2012-11-09 22:37:44 <sipa> sorry...
696 2012-11-09 22:37:45 <xisalty-otc> lol
697 2012-11-09 22:38:03 <D34TH> Build #135 is already in progress (ETA:6 hr 15 min) )
698 2012-11-09 22:38:06 <D34TH> DD:
699 2012-11-09 22:41:34 <xisalty-otc> 6 hrs
700 2012-11-09 22:41:44 <xisalty-otc> why does it take so long >
701 2012-11-09 22:42:09 <sipa> it runs in a virtual machine on an amazon instance
702 2012-11-09 22:42:30 <xisalty-otc> fair poitn
703 2012-11-09 22:42:31 <D34TH> i.e. -1 cores
704 2012-11-09 22:42:38 <D34TH> @ -55 THz
705 2012-11-09 22:42:47 <Joric> go boost