1 2012-08-29 00:03:12 <jgarzik> ACTION wonders why Fedora ships emacs-nox (no-X11). Other core system dependencies were built with X11 libs, so you're not escaping X11 libs with a stripped-down system.
2 2012-08-29 00:03:59 <Luke-Jr> lol
3 2012-08-29 00:29:53 <sipa> pirate will certainly have defaulted at block 241300
4 2012-08-29 00:29:55 <jgarzik> 1531 jgarzik 20 0 826m 29m 2836 S 88.1 0.5 47164:22 bitcoind
5 2012-08-29 00:30:03 <jgarzik> guess which one is mainnet
6 2012-08-29 00:30:13 <jgarzik> and which one is mining on testnet
7 2012-08-29 00:30:51 <sipa> at that point his debt will be more than the amount of BTC in circilation
8 2012-08-29 00:31:15 <sipa> jgarzik: eh, i hope the first is testnet
9 2012-08-29 00:31:27 <sipa> but if it was, you wouldnt ask
10 2012-08-29 00:32:22 <jgarzik> correct
11 2012-08-29 00:32:38 <jgarzik> ACTION watches top, to make sure -flto does not cause swapping
12 2012-08-29 00:32:46 <jgarzik> 1.4g RSS is quite a lot for mainnet
13 2012-08-29 00:33:37 <sipa> my node on my vps is at 450 MB RSS
14 2012-08-29 00:33:47 <sipa> after running a few days
15 2012-08-29 00:34:35 <jgarzik> this is us4.exmulti.net public node, ~30 day uptime
16 2012-08-29 00:34:46 <jgarzik> always listed in bitseed.xf2.org
17 2012-08-29 00:35:07 <jgarzik> (though usage may go down with 0.7 release, thanks to gmaxwell's DNS seed list reordering)
18 2012-08-29 00:35:08 <sipa> hmm, only 16 connections
19 2012-08-29 00:35:15 <jgarzik> >100 here
20 2012-08-29 00:35:37 <gmaxwell> 1.4g sounds like a leak.
21 2012-08-29 00:35:40 <gmaxwell> how many connections?
22 2012-08-29 00:35:41 <sipa> maybe i enabled the canonical sigs check there
23 2012-08-29 00:36:34 <jgarzik> OK, post-build, number of connections down to 78. << gmaxwell
24 2012-08-29 00:36:51 <jgarzik> get a lot of bursty BitcoinJ activity on the public nodes
25 2012-08-29 00:37:12 <gmaxwell> jgarzik: thats busted. crap.
26 2012-08-29 00:37:16 <gmaxwell> I wonder how to trigger it.
27 2012-08-29 00:38:15 <gmaxwell> none of my nodes have rss >200MB right now, but none have been up for more than 12 hours or so. (well, one has, but it has only one connection and it's burried behind the others)
28 2012-08-29 00:38:31 <jgarzik> gmaxwell: we have a lot of caches that can really grow. they are all size-limited... but if you max size all of them, it is still quite a lot
29 2012-08-29 00:39:13 <gmaxwell> well, and also almost all of my nodes have additional patches to drop dice txn now. So if mempool size is a part of it....
30 2012-08-29 00:40:52 <sipa> maybe the mining code leaks?
31 2012-08-29 00:41:03 <gmaxwell> lemme see.
32 2012-08-29 00:41:35 <gmaxwell> I have a testnet node up doing getwork mining, and two of my regular nodes are p2pool miners.
33 2012-08-29 00:41:48 <gmaxwell> so I don't think it's mining code leaking.
34 2012-08-29 00:42:30 <jgarzik> note the difference in CPU times, in the paste above
35 2012-08-29 00:42:38 <jgarzik> testnet + mining == very low memory usage
36 2012-08-29 00:42:48 <jgarzik> mainnet has low CPU usage but hugemem
37 2012-08-29 00:43:42 <gmaxwell> jgarzik: well you could attach gdb and check some of the cache sizes.
38 2012-08-29 00:44:05 <jgarzik> too late
39 2012-08-29 00:44:29 <gmaxwell> in any case, if we're at risk of >1GB RSS on a non-listening node; then I think thats a release blocker. :( There are a lot of people who run bitcoind in fairly modest VPSes.
40 2012-08-29 00:45:22 <phantomcircuit_> recent increase in memory usage?
41 2012-08-29 00:45:30 <jgarzik> 1.4g mainnet node was git HEAD as of...
42 2012-08-29 00:45:41 <jgarzik> ('dirty' is merely the makefile... its vanilla code)
43 2012-08-29 00:45:56 <gmaxwell> phantomcircuit_: I dunno, I haven't seen that on my nodes.
44 2012-08-29 00:45:56 <jgarzik> and the version snarfed from build.h is broken too.
45 2012-08-29 00:46:06 <jgarzik> the git commit is correct though (ga9d...)
46 2012-08-29 00:47:12 <phantomcircuit_> very long running node here is at 800 MB on a 8GB system (with nothing else running on it...)
47 2012-08-29 00:47:33 <jgarzik> gmaxwell: just reappeared on testnet3, incl IRC, with current git HEAD
48 2012-08-29 00:47:48 <phantomcircuit_> of course that has a ton of active transactions and such
49 2012-08-29 00:47:49 <gmaxwell> My nodes right now have RSS: 165m, 121m, 35m(testnet), 121m, 122m, 122m, 91m (testnet), 130m
50 2012-08-29 00:47:59 <jgarzik> trying connection 203.122.247.149:18333 lastseen=0.9hrs
51 2012-08-29 00:48:00 <jgarzik> connect() failed after select(): No route to host
52 2012-08-29 00:48:04 <jgarzik> getting that every few seconds
53 2012-08-29 00:48:07 <jgarzik> grump
54 2012-08-29 00:48:10 <gmaxwell> jgarzik: yes. expected.
55 2012-08-29 00:48:51 <jgarzik> one hopes my ISP doesn't think I'm trying to DoS somebody :)
56 2012-08-29 00:48:52 <gmaxwell> Go look at the IRC list you will only get _two_ nodes with :18333, one is me which you've connected to. The other is 203.122.247.149 which it keeps retrying.
57 2012-08-29 00:49:06 <gmaxwell> Oh every few seconds is normal.
58 2012-08-29 00:49:31 <gmaxwell> The important thing is that once everyone else updates you'll be able to connect to them.
59 2012-08-29 00:50:13 <gmaxwell> trying connection 96.241.176.56:18333 lastseen=0.8hrs
60 2012-08-29 00:50:23 <gmaxwell> oh nevermind that.
61 2012-08-29 00:50:44 <gmaxwell> jgarzik: woot; I learned your address over IRC.
62 2012-08-29 00:50:54 <gmaxwell> fix works.
63 2012-08-29 00:51:34 <gmaxwell> midnightmagic: please update your public testnet3 node(s) to git head.
64 2012-08-29 01:14:23 <jgarzik> bah. -flto poops itself on this laptop.
65 2012-08-29 01:14:42 <jgarzik> oh well.. with just 1GB RAM, might as well turn it off anyway.
66 2012-08-29 01:16:04 <gmaxwell> jgarzik: "offline wallet"
67 2012-08-29 01:16:48 <jgarzik> I'm certainly offline... had to exit Firefox to get some RAM for building ;p
68 2012-08-29 01:32:28 <jgarzik> ass!
69 2012-08-29 01:32:46 <jgarzik> we _require_ the json_spirit template stuff. no lib for us, I think.
70 2012-08-29 01:33:46 <gmaxwell> Whats the killer reason?
71 2012-08-29 01:36:48 <jgarzik> gmaxwell: bitcoinrpc.cpp uses read_string and write_string, which only exists in the templates
72 2012-08-29 01:43:59 <jgarzik> gmaxwell: can you try this? I'm curious about your results, if you have a moment. https://github.com/bitcoin/bitcoin/pull/1749
73 2012-08-29 01:44:14 <jgarzik> it builds bitcoinrpc as normal, but rpc*.cpp with libjson
74 2012-08-29 01:44:30 <jgarzik> not optimal, but the best that can be done with just include changes
75 2012-08-29 01:51:54 <gmaxwell> wumpus: okay. how the @#$@# do I get the quoting right in the QT console.
76 2012-08-29 01:51:57 <gmaxwell> I thought I tested this.
77 2012-08-29 01:52:14 <gmaxwell> addmultisigaddress 2 '["03a2a4381813bfc8e6774c40c353a8686dbfa78c2808b253a47d516a7093f9d915","022d04e128f94d8edbe806d4ff506c63644d183aa81a9f4532ba8f025afb133447"]'
78 2012-08-29 01:52:45 <gmaxwell> is how I quote at the CLI. I've tried every @#$@# mixture I can think of. HALP.
79 2012-08-29 01:55:34 <jgarzik> gmaxwell: why would ui console need quoting?
80 2012-08-29 01:55:41 <jgarzik> seems like that's a shell thing
81 2012-08-29 01:56:51 <gmaxwell> jgarzik: because it doesn't work without any either.
82 2012-08-29 01:57:15 <jgarzik> sad
83 2012-08-29 02:13:00 <sunshinehappy> gmaxwell, lets talk?
84 2012-08-29 02:13:56 <gmaxwell> sunshinehappy: I'll unban you in #bitcoin if you agree to chill out some. Okay?
85 2012-08-29 02:14:03 <sunshinehappy> yes
86 2012-08-29 03:42:02 <wumpus> gmaxwell: it shouldn't need quoting, at least as long as there are no spaces inside the argument
87 2012-08-29 03:43:16 <wumpus> gmaxwell: but maybe something is going wrong in RPCExecutor::request when separating the arguments
88 2012-08-29 03:46:35 <jgarzik> wumpus, gmaxwell: RPCExecutor only supports simple list of arguments, and not complex JSON Object arguments that some RPCs require, AFAICS
89 2012-08-29 03:46:44 <jgarzik> the parsing is IMO quite broken
90 2012-08-29 03:46:51 <jgarzik> well, s/broken/incorrect/
91 2012-08-29 03:47:07 <wumpus> it does, it simply extracts strings and parses them as the command-line tool would
92 2012-08-29 03:47:27 <wumpus> the idea is that it would work the same as a command-line
93 2012-08-29 03:47:40 <wumpus> of course, boost::escaped string probably doesn't emulate bash 100%
94 2012-08-29 03:48:23 <wumpus> so if there are specific problems please report them, I cannot do anything with 'is quite broken'
95 2012-08-29 03:48:38 <jgarzik> there should not need to be any escaping at all
96 2012-08-29 03:49:00 <jgarzik> there is no shell interpreting anything
97 2012-08-29 03:49:19 <wumpus> well, yeah there should... *currently* it emulates a command line with bash.. I know there are other options, but this was easiest to implement for now
98 2012-08-29 03:49:38 <wumpus> and resulted in the least extra code (ie, special handling of json arguments and such)
99 2012-08-29 03:50:23 <wumpus> or do you mean it should simply interpret json arrays instead of space separated arguments?
100 2012-08-29 03:50:30 <gmaxwell> wumpus: well, I opened an issue with an example; in case I didn't catch you before I went to bed (should have happened an hour ago :) )
101 2012-08-29 03:50:47 <wumpus> it's more typing work at least... then input would be ['getinfo'] etc
102 2012-08-29 03:50:55 <gmaxwell> yuck
103 2012-08-29 03:50:59 <wumpus> exactly
104 2012-08-29 03:51:12 <gmaxwell> Really I think it should emulate the cli as much as reasonably possible; then instructions work both ways.
105 2012-08-29 03:51:16 <wumpus> so I think trying to behave like bash is least evil...
106 2012-08-29 03:51:23 <gmaxwell> I resisted suggesting you embed a python shell. :)
107 2012-08-29 03:51:42 <jgarzik> <shrug> oh well, it remains a mess of quoting then
108 2012-08-29 03:52:08 <wumpus> gmaxwell: yes then you can copy/paste interchangably
109 2012-08-29 03:52:15 <gmaxwell> well, if it worked with less quoting than the shell but also as much quoting as the shell I think that would be okay.
110 2012-08-29 03:52:51 <wumpus> so how would it separate the arguments, if not by space-separated (or escaped) strings?
111 2012-08-29 03:52:59 <gmaxwell> But yes, I want to be able to write examples for the forums/wikis/docs only once. Not "the cli way" and "the gui console way". But regardless, just working is most important. :)
112 2012-08-29 03:53:31 <wumpus> I guess space-separated json expressions would work
113 2012-08-29 03:53:39 <wumpus> *but* then you can only give strings in "
114 2012-08-29 03:54:31 <gmaxwell> wumpus: thats how it works from bash.
115 2012-08-29 03:54:35 <wumpus> so you'd have to use getbalance "account" instead of getbalance account
116 2012-08-29 03:54:42 <wumpus> no, it's not, bash has no knowledge of json
117 2012-08-29 03:54:47 <gmaxwell> hm. even if account has no spaces?
118 2012-08-29 03:54:57 <wumpus> a first argument ["bla bla"] would be split into two
119 2012-08-29 03:55:12 <wumpus> or not?
120 2012-08-29 03:55:33 <wumpus> I have no idea how bash handles " if it's not at the beginning of the string
121 2012-08-29 03:55:59 <jgarzik> gmaxwell: ...said examples on forum/wiki/docs will be wrong for anyone using python-bitcoinrpc or any other "normal JSON" solution
122 2012-08-29 03:56:26 <wumpus> if bash handles " within strings correctly then yeah the bash-emulation is wrong :-)
123 2012-08-29 03:56:41 <jgarzik> The cli shell quoting is a huge mess, even more so once we start accepting Object arguments.
124 2012-08-29 03:56:43 <wumpus> within arguments, I mean
125 2012-08-29 03:56:56 <jgarzik> Carrying that mess into a new arena ... just spreads the infection
126 2012-08-29 03:57:45 <wumpus> ok let's switch to space separated json expressions then.. it means having to quote every string (except the command, that can be special-cased), but so be it
127 2012-08-29 03:58:21 <jgarzik> examples on the wiki are presented using normal JSON, not cli:
128 2012-08-29 03:58:23 <jgarzik> https://en.bitcoin.it/wiki/API_reference_%28JSON-RPC%29#Command_line_.28cURL.29
129 2012-08-29 03:58:40 <wumpus> addmultisigaddress 2 ["03a2a4381813bfc8e6774c40c353a8686dbfa78c2808b253a47d516a7093f9d915","022d04e128f94d8edbe806d4ff506c63644d183aa81a9f4532ba8f025afb133447"]
130 2012-08-29 03:58:42 <wumpus> would then simply work
131 2012-08-29 03:59:15 <gmaxwell> wumpus: what works in bash is addmultisigaddress 2 '["03a2a4381813bfc8e6774c40c353a8686dbfa78c2808b253a47d516a7093f9d915","022d04e128f94d8edbe806d4ff506c63644d183aa81a9f4532ba8f025afb133447"]' which is close enough.
132 2012-08-29 03:59:20 <wumpus> but JSON strings would have to be correctly put in quotes
133 2012-08-29 03:59:39 <wumpus> but spaces in JSON expressions would still work and not cause an argument split
134 2012-08-29 03:59:52 <wumpus> gmaxwell: that should work in the ui console as well
135 2012-08-29 04:00:03 <gmaxwell> wumpus: it doesn't work currently.
136 2012-08-29 04:00:10 <wumpus> ok, then that's a bug
137 2012-08-29 04:01:11 <wumpus> as I said, it's supposed to emulate bash, if it doesn't it's wrong... but seemingly jgarzik wants to do it completely different
138 2012-08-29 04:01:22 <wumpus> anyway, let him submit a patch then :-)
139 2012-08-29 04:03:00 <gmaxwell> wumpus: I don't think we should let jgarzik's desire for different get in the way of fixing what we've got.
140 2012-08-29 04:03:40 <gmaxwell> (I don't think being more RPCish is at all acceptable without, e.g. sticking in a python interpreter; and I think adding an email client to bitcoin is higher on the todo list :) )
141 2012-08-29 04:04:03 <wumpus> we could add a js interpreter (qt has some support for scripting) :p
142 2012-08-29 04:04:30 <freewil> embed v8 ;)
143 2012-08-29 04:04:41 <wumpus> QtScript
144 2012-08-29 04:05:14 <wumpus> https://en.wikipedia.org/wiki/QtScript
145 2012-08-29 04:05:17 <gmaxwell> wumpus: or that. Right. In any case, as is ??? all these awesome features we expected power users to get to via the gui console? 1/4 don't work. Doh.
146 2012-08-29 04:06:22 <wumpus> yes, I wonder why no one tried this out *before* 0.7.0rc1 release
147 2012-08-29 04:08:04 <wumpus> I do remember trying some testcases like that which worked, but maybe something changed later...
148 2012-08-29 04:11:34 <gmaxwell> wumpus: I have a checklist here that says I tested sendmany.
149 2012-08-29 04:11:48 <gmaxwell> I think the raw txn stuff wasn't merged when I did that testing, however.
150 2012-08-29 04:51:22 <midnightmagic> holy crap, look at all those 1dice txn
151 2012-08-29 12:47:07 <sebicas> ;;seen justmoon
152 2012-08-29 12:47:08 <gribble> justmoon was last seen in #bitcoin-dev 15 hours, 5 minutes, and 18 seconds ago: <justmoon> cu
153 2012-08-29 13:08:43 <MC-Eeepc> wow hal posted yesterday
154 2012-08-29 13:12:20 <jgarzik> yep, pretty cool
155 2012-08-29 13:41:22 <BlueMatt> nice
156 2012-08-29 13:41:41 <sipa> gmaxwell: added more tests to canonical
157 2012-08-29 13:44:23 <BlueMatt> sipa: oh, you're just checking sigs itself, not txes with sigs?
158 2012-08-29 13:44:29 <sipa> BlueMatt: yes
159 2012-08-29 13:44:42 <sipa> it's even independent from them being valid in some context or not
160 2012-08-29 13:45:02 <BlueMatt> would it not be easier to put it in full tx form and add a non-standard-but-valid tx data file to make it more extensible?
161 2012-08-29 13:45:06 <BlueMatt> s/easier/better/
162 2012-08-29 13:46:22 <sipa> how is that more extensible?
163 2012-08-29 13:46:35 <BlueMatt> you can add other IsStandard checks
164 2012-08-29 13:46:46 <sipa> it has nothing to do with IsStandard
165 2012-08-29 13:47:00 <sipa> as it require script execution to find out
166 2012-08-29 13:47:02 <BlueMatt> I was under the impression canonical sig checking would be a part of IsStandard
167 2012-08-29 13:47:22 <BlueMatt> or would otherwise be implemented in a similar fashion
168 2012-08-29 13:47:40 <sipa> no, it's a flag set when script-validating incoming mempool transactions
169 2012-08-29 13:47:45 <sipa> but not blockchain transactions
170 2012-08-29 13:48:12 <BlueMatt> and would that flag not be handled similarly to how IsStandard's return is handled?
171 2012-08-29 13:48:19 <sipa> right, sure
172 2012-08-29 13:48:22 <sipa> the effect is similar
173 2012-08-29 13:48:30 <BlueMatt> ok, well thats my point
174 2012-08-29 13:48:56 <sipa> but this allows you to have a testcase for a IsCanonicalSignature function anyone can implement
175 2012-08-29 13:49:15 <BlueMatt> people should be implementing all of IsStandard not just IsCanonicalSignature
176 2012-08-29 13:49:37 <BlueMatt> and, for a new implementation, they are effectively the same, and could be in a part of the same function
177 2012-08-29 13:49:42 <sipa> depends, they don't have to agree with IsStandard
178 2012-08-29 13:50:09 <sipa> but if IsCanonicalSignature ever becomes required as a network rule, it must be exactly implemented
179 2012-08-29 13:50:45 <BlueMatt> well when it does, those test cases which check for is standard through is canonicalsig would be moved to tx_invalid.json from tx_validbutnonstandard.json
180 2012-08-29 13:51:46 <sipa> no reason we can't have both
181 2012-08-29 13:52:03 <BlueMatt> well, its simpler for other clients to implement if we dont ;)
182 2012-08-29 13:52:12 <gmaxwell> 08:49 < BlueMatt> people should be implementing all of IsStandard not just IsCanonicalSignature
183 2012-08-29 13:52:23 <BlueMatt> if they are implementing it at all
184 2012-08-29 13:52:24 <gmaxwell> But they don't have to duplicate it exactly.
185 2012-08-29 13:52:25 <BlueMatt> (for now)
186 2012-08-29 13:52:41 <BlueMatt> meh
187 2012-08-29 13:52:52 <sipa> and again, i disagree IsCanonicalSignature is part of IsStandard
188 2012-08-29 13:53:00 <sipa> it's a more strict mode of script validation
189 2012-08-29 13:53:04 <sipa> like P2SH is
190 2012-08-29 13:53:13 <gmaxwell> This IsCanonicalSignature is a clear candidate for becoming a network rule.
191 2012-08-29 13:53:27 <BlueMatt> eh, ok thats not how I see it, but...whatever its not a huge deal either way
192 2012-08-29 13:54:01 <gmaxwell> If we'd done it in IsStandard a year ago, I'd have suggested considering it as part of the blockv2 change.
193 2012-08-29 13:54:29 <BlueMatt> ehhhh
194 2012-08-29 13:54:32 <sipa> maybe we can extend the script_*.json format to take some flags: fP2SH, fCanonical
195 2012-08-29 13:54:49 <BlueMatt> we already do
196 2012-08-29 13:54:58 <BlueMatt> or...no thats just for tx_*.json
197 2012-08-29 13:55:08 <sipa> i see something like that in the tx tests, yes
198 2012-08-29 13:55:20 <sipa> but it's ultimately a property of the script validation
199 2012-08-29 13:55:27 <sipa> but yes, there too, i guess
200 2012-08-29 13:56:04 <BlueMatt> well since you cant do any sig checking stuff in script_*.json... but, yea
201 2012-08-29 13:56:05 <sipa> anyway, i hope the cases I added are sufficient to have 100% coverage of IsCanonicalSignature
202 2012-08-29 13:56:06 <Luke-Jr> would be nice if IsStandard had 2 levels, so I could (easily) accept custom scripts, but reject those that use OP_NOPn or non-canonical signatures
203 2012-08-29 13:56:14 <BlueMatt> sipa: check it ;)
204 2012-08-29 13:56:29 <sipa> Luke-Jr: note that the implementation is completely independent from IsStandard
205 2012-08-29 13:56:36 <BlueMatt> gmaxwell: ok, maybe running coverage of test_bitcoin sooner rather than later would be nice
206 2012-08-29 13:56:48 <sipa> it's just a flag passed to EvalScript or VerifySignature
207 2012-08-29 13:57:45 <gmaxwell> Luke-Jr: I agree. If people like you are going to gut IsStandard, then we ought to guide them on more or less agressive ways to do it.
208 2012-08-29 13:59:42 <sipa> i could add some code to try to detect send-to-noncanonical-pubkey cases, and *that* would be something IsStandard()-like
209 2012-08-29 14:00:01 <sipa> but this is really just a property of script execution
210 2012-08-29 14:07:23 <gmaxwell> https://people.xiph.org/~greg/bitcoin_coverage/new_coverage/coverage/home/gmaxwell/src/bcm/bax/src/script.cpp.gcov.html hm.. doesnt seem to have increased coverage. I wonder if I goofed up the measurement.
211 2012-08-29 14:16:10 <jgarzik> yeah. While I think luke-jr's transaction acceptance policy is quite extreme, a luke-jr-lite version is IMO reasonable. It seems reasonable for a miner to want to compete by permitting a limited selection of !IsStandard transactions for additional fees or whatever.
212 2012-08-29 14:16:17 <jgarzik> Difficult defining "limited" though
213 2012-08-29 14:17:58 <gmaxwell> Well I don't even know about exposing it. Just restructring the code so that a Luke-Jr' who attempted to open up the policy wouldn't turn off the stuff thats more important might be good enough.
214 2012-08-29 14:19:10 <gavinandresen> meh. adding new templates to script.cpp is easy, if there's a demand for a specific type of new transaction.
215 2012-08-29 14:19:57 <gmaxwell> gavinandresen: perhaps, but I'm pretty sure at the moment that a miner wanting to allow one would simply gut the isstandard check entirely.
216 2012-08-29 14:21:32 <gavinandresen> gmaxwell: okey dokey. We should tell them to add a new template instead, it is much safer.
217 2012-08-29 14:22:08 <gmaxwell> then perhaps thats just what we should do. E.g. add a comment that recommends that, and tells them how.
218 2012-08-29 14:23:50 <gavinandresen> gmaxwell: a comment like "if you're thinking of gutting this because..." at the top of IsStandard is a good idea.
219 2012-08-29 14:24:41 <Luke-Jr> you're missing the "Let people experiment without explicit involvement from miners" option
220 2012-08-29 14:25:36 <gavinandresen> testnet is for experimenting
221 2012-08-29 14:29:01 <jgarzik> well, there's experimenting with a software change, then there's experimenting with a business idea that needs real-money testing :)
222 2012-08-29 14:29:19 <jgarzik> i.e. to see if real customers want the service
223 2012-08-29 14:29:40 <jgarzik> but that's more on the miner side, and not "people...without miner involvement" side
224 2012-08-29 15:05:12 <Diapolo> Luke-Jr: You saw my change to the 1stclass pull?
225 2012-08-29 15:10:54 <justmoon> sebicas: you were looking for me?
226 2012-08-29 15:11:23 <sebicas> Hi justmoon: yes..
227 2012-08-29 15:11:38 <justmoon> what's up?
228 2012-08-29 15:12:02 <justmoon> ah got your message
229 2012-08-29 15:16:37 <jgarzik> sebicas: hey. JFYI I'm at the beach all week, and not bothering with pynode for the most part
230 2012-08-29 15:17:24 <sebicas> jgarrzik: No worries :) Thanks so much for all your help!
231 2012-08-29 15:17:29 <sebicas> Your code helped me a lot!
232 2012-08-29 15:21:03 <sebicas> jgarrzik: I released v0.2 where you can actually see the output addresses and amounts - https://github.com/sebicas/bitcoin-sniffer
233 2012-08-29 15:24:26 <sebicas> And in v0.3 I will include a plug-in where you can publish notification to Redis SubPub
234 2012-08-29 15:25:54 <Eliel> sebicas: I think I'm going to put this to use soon :)
235 2012-08-29 15:26:13 <Eliel> sebicas: thank you for building it.
236 2012-08-29 15:26:20 <sebicas> Eliel: Great I glad it could help!
237 2012-08-29 15:26:52 <sebicas> Eliel: Let me know of any upgrade you think it may be helpfull
238 2012-08-29 15:27:21 <Eliel> I will, although, it might be in the form of a patch :)
239 2012-08-29 15:27:42 <sebicas> Sure, fork it and publish your changes if you like..
240 2012-08-29 15:33:48 <sebicas> I just build bitcoind 0.7.0-rc1 and run ./bitcoind -upgradewallet and still running after 30 minutes, is that ok?
241 2012-08-29 15:44:39 <devrandom> sipa: did lxc work out?
242 2012-08-29 16:54:14 <gmaxwell> wumpus: I wonder if we should have a one-time-ever popup dialog during the initial block download thats says something like "Woah! Why is my computer so busy?\\nWhen a new bitcoin install starts up for the first time it synchronizes itself with the Bitcoin network and verifies for itself that all of the rules of the system are being followed. This process is part of what makes Bitcoin secure. This can take several hours to overnight."
243 2012-08-29 16:57:55 <Wormik> It's true. When any market or shop in any town and city will get bitcoins for product, in the current form of online wallet is expensive for ordinary people. It will take tens and then hundreds of gigabytes!
244 2012-08-29 16:58:10 <_dr> while you're at it, an ETA in the message would be nice :)
245 2012-08-29 17:00:07 <gmaxwell> _dr: remotely reliable estimates of that will not be easy until we totally rework how the fetch happens. And once we've done that it'll probably be easy to fully background the fetch, making an ETA kind of unimportant.
246 2012-08-29 17:00:42 <_dr> sounds great
247 2012-08-29 17:02:18 <BlueMatt> hmm...my brain is broken, why can I not think of a scenario that ever hits "Postponing %i reconnects" ?
248 2012-08-29 17:02:25 <Luke-Jr> gmaxwell: well, showing "5 hours until full node" in the debug window wouldn't hurt at that point either ;p
249 2012-08-29 17:02:42 <BlueMatt> sipa: ?
250 2012-08-29 17:02:57 <gmaxwell> Luke-Jr: true true
251 2012-08-29 18:16:04 <sipa> sebicas: -upgradewallet is instantaneous; after that, bitcoin just runs as normal
252 2012-08-29 18:16:11 <sipa> devrandom: yes, works perfectly
253 2012-08-29 18:16:45 <sebicas> sipa: Ahh ok,thanks!
254 2012-08-29 18:17:13 <sebicas> I just stopped it and started as daemon and everything worked perfectly??? thanks
255 2012-08-29 18:17:48 <sebicas> sipa: Any place where I can find examples of https://en.bitcoin.it/wiki/Raw_Transactions ?
256 2012-08-29 18:18:12 <sipa> sebicas: i'm afraid they're too brand new for that
257 2012-08-29 18:18:28 <sebicas> sipa: Ok, I am testing them..
258 2012-08-29 18:18:35 <sebicas> So I guess I could create some..
259 2012-08-29 18:18:54 <sipa> sebicas: please report if you have difficulties; i'm sure examples on the wiki would also be useful for others if you like
260 2012-08-29 18:19:12 <sebicas> sipa: ok
261 2012-08-29 18:19:21 <sipa> BlueMatt: you don't know why that postponing happens, or you can't imagine what caused it?
262 2012-08-29 18:33:44 <Diapolo> jgarzik: Is that space and tab removing skript still targeted before 0.7 get's final ;)?
263 2012-08-29 18:35:27 <Diapolo> luke-jr: Are you here?
264 2012-08-29 18:49:55 <jgarzik> Diapolo: sure, just remind me :)
265 2012-08-29 18:51:03 <Diapolo> jgarzik: that's a word I just need to take care to listen when core devs talk about a relase date ^^
266 2012-08-29 19:28:05 <Guest5847> does anyone know where to get a copy of the mtgox (mtgox accounts.csv) database that was leaked around 6/20/2011? I wanted to test out hash cat on real data.
267 2012-08-29 20:16:21 <gmaxwell> https://plus.google.com/115547683951727699051/posts/g1E6AxVKtyc < I think we know how to make a cryptographic protocol with better properties than the ones described here.
268 2012-08-29 20:16:58 <gmaxwell> oops wrong channel.
269 2012-08-29 20:18:18 <midnightmagic> gmaxwell, might wanna ban Guest5847
270 2012-08-29 20:18:54 <Luke-Jr> -rw-r--r-- 1 bitcoinpool bitcoinpool 118G Aug 29 22:18 debug.log
271 2012-08-29 20:19:00 <Luke-Jr> <.<
272 2012-08-29 20:20:14 <Eliel> now that's big.
273 2012-08-29 20:20:27 <Eliel> I guess you found out what's eating up all the space :)
274 2012-08-29 20:20:42 <Luke-Jr> XD
275 2012-08-29 20:24:37 <Guest5847> midnightmagic: what? don't ban me!
276 2012-08-29 20:25:44 <Luke-Jr> hmm
277 2012-08-29 20:25:51 <Luke-Jr> Eligius is still claiming protocol 70000
278 2012-08-29 20:25:56 <Luke-Jr> will 0.7 handle that poorly?
279 2012-08-29 20:26:45 <sipa> no
280 2012-08-29 20:26:58 <sipa> it doesn't use the 60002 protocol feature itself
281 2012-08-29 20:27:17 <sipa> (bip 35, i mean)
282 2012-08-29 20:27:59 <Luke-Jr> phew
283 2012-08-29 20:28:16 <Luke-Jr> we're getting close to being able to upgrade to 0.7-ish, but not that close yet :/
284 2012-08-29 20:28:37 <Luke-Jr> (Eligius runs a 0.6.0.x-based next-test)
285 2012-08-29 20:29:12 <Luke-Jr> sipa: btw, if you can, 0.4.8rc1, 0.5.7rc1, 0.6.0.10rc1, and 0.6.4rc1 could use more builds <.<
286 2012-08-29 20:30:45 <sipa> how long are you going to keep doing backports for those?
287 2012-08-29 20:31:57 <slush> any issues with 0.7?
288 2012-08-29 20:32:51 <slush> I thought it is already stable but it is still rc2
289 2012-08-29 20:33:38 <Luke-Jr> sipa: 0.4.x doesn't have any anti-DoS measures, including the alert ones, so I'll probably drop it soonish unless someone tells me they need it for some good reason
290 2012-08-29 20:33:51 <Luke-Jr> 0.6.0.x probably not long after Eligius upgrades, unless <same thing>
291 2012-08-29 20:33:59 <sipa> ACTION would rather not maintain them, until someone asks...
292 2012-08-29 20:34:11 <Luke-Jr> 0.6.x probably 3 years for Debian
293 2012-08-29 20:35:38 <Luke-Jr> sipa: well, I know there are still people using 0.4.x for wxBitcoin - I kinda wish someone would step up to maintain that and bring it to 0.7.x level
294 2012-08-29 20:36:51 <Luke-Jr> it seems wx has users, but no interest from developers
295 2012-08-29 20:46:02 <Eliel> ACTION wonders if it would be enough for those people to implement a very close lookalike with the -qt version.
296 2012-08-29 20:46:42 <Eliel> ACTION no longer remembers what the wx-version looked like
297 2012-08-29 20:46:51 <kjj_> heh. I still have 0.4.0-beta running on my home box. never seen the QT gui
298 2012-08-29 20:47:12 <kjj_> no real reason, just never bothered to upgrade this box
299 2012-08-29 20:47:32 <Luke-Jr> kjj_: ???
300 2012-08-29 20:47:39 <Luke-Jr> using vulnerable version = fail
301 2012-08-29 20:47:56 <Luke-Jr> Eliel: I barely saw it, since i tried to keep wx off my system :P
302 2012-08-29 20:48:05 <kjj_> meh
303 2012-08-29 20:48:27 <Luke-Jr> kjj_: at least update to 0.4.8 ;)
304 2012-08-29 20:49:03 <kjj_> is there something other than a potential DOS that I'm supposed to be worried about? I think I read all of the changelogs
305 2012-08-29 20:49:42 <sipa> you may want to have a look at https://en.bitcoin.it/wiki/Common_Vulnerabilities_and_Exposures
306 2012-08-29 20:49:47 <Luke-Jr> kjj_: ^
307 2012-08-29 20:52:15 <kjj_> CVE-2012-3789 is the only one that looks even slightly worrisome, and that doesn't have any details
308 2012-08-29 20:52:42 <Luke-Jr> CVE-2012-2459 seems the most worrisome to me
309 2012-08-29 20:56:14 <Eliel> I have to say I'm quite impressed that there haven't been any vulnerabilities of remote code execution type in bitcoind so far :)
310 2012-08-29 21:00:09 <kjj_> and third trip to the colo. either I'm incredibly unlucky with power supplies, or I was right on the edge and the extra DIMM is drawing too much current
311 2012-08-29 21:00:50 <[Tycho]> Luke-Jr: why you aren't adding those fixes to your supported old versions ?
312 2012-08-29 21:01:00 <Luke-Jr> [Tycho]: ?
313 2012-08-29 21:01:19 <sipa> [Tycho]: he is, but if someone is running 0.4.0 vanilla, there's not much we can do about it, right?
314 2012-08-29 21:01:45 <Luke-Jr> sipa: well, 0.4.x isn't getting the alert fixes, simply because they're considered a feature from 0.4.x's viewpoint
315 2012-08-29 21:01:48 <[Tycho]> Oh. He said "Luke-Jr: sipa: 0.4.x doesn't have any anti-DoS measures"
316 2012-08-29 21:02:04 <Luke-Jr> [Tycho]: the anti-DoS framework was a new feature in 0.5.x
317 2012-08-29 21:02:33 <Luke-Jr> 0.4.x and earlier had nothing to handle DoS attacks
318 2012-08-29 21:03:10 <Luke-Jr> that is, the only fix for this is to say "you're being rude; imma ban you!"
319 2012-08-29 21:14:36 <BitcoinBaltar> Notice that one of the faucets put up a tor node address in case anyone needs one to start with: pqosrh6wfaucet32.onion
320 2012-08-29 22:12:58 <BlueMatt> sipa: I cant think of a test case which exercises that code path
321 2012-08-29 22:13:19 <sipa> BlueMatt: which?
322 2012-08-29 22:13:36 <BlueMatt> the reorg postpone
323 2012-08-29 22:14:09 <sipa> it occurs, and you don't understand why, or you want to build a testcase for it?
324 2012-08-29 22:14:42 <BlueMatt> I understand it, I wrote the same thing in bitcoinj, but I cant figure out a test case which will trigger it
325 2012-08-29 22:14:53 <BlueMatt> or...I have test cases which should and it doesnt seem to be
326 2012-08-29 22:15:36 <sipa> well it's something that can't occur in normal operation
327 2012-08-29 22:15:42 <BlueMatt> ah, well no wonder
328 2012-08-29 22:16:06 <sipa> it requires a all-of-a-sudden new best chain which is a lot longer than the old active one
329 2012-08-29 22:16:22 <sipa> well, actually, it can happen
330 2012-08-29 22:16:53 <sipa> in a tree: root block A, two children B and B', and then further C, D, E, ... from B
331 2012-08-29 22:17:19 <sipa> first have A, send B' to it, then send D,E,F,G,H,... to it but skipping C
332 2012-08-29 22:17:25 <sipa> and then suddenly send C?
333 2012-08-29 22:17:45 <BlueMatt> why does b12 not trigger it: http://jenkins.bluematt.me/pull-tester/files/FullBlockTestGenerator.java ?
334 2012-08-29 22:17:48 <sipa> actually, even then I think they are connected one by one
335 2012-08-29 22:18:11 <BlueMatt> b5, b6, b13 based on b12, b14 based on b13, then b12 based on b5
336 2012-08-29 22:18:42 <BlueMatt> yea, so I really dont think it will ever get triggered in normal op because b13 and 14 are orphans
337 2012-08-29 22:18:43 <sipa> nope, even with orphans you shouldn't be able to trigger it
338 2012-08-29 22:19:01 <BlueMatt> yea, thought so
339 2012-08-29 22:19:11 <BlueMatt> no wonder I couldnt think of why i couldnt do it
340 2012-08-29 22:19:15 <sipa> it needs something like a corrupted block chain db, or changed block validity rules
341 2012-08-29 22:19:36 <BlueMatt> well, in that case, I have pretty much full line coverage of the block acceptance except for one todo and some of the MoneyRange checks
342 2012-08-29 22:20:16 <sipa> nice!
343 2012-08-29 22:22:20 <BlueMatt> hopefully i can get it mined tomorrow and put it in a data-driven format
344 2012-08-29 22:22:44 <BlueMatt> though it does check some of the network handling as is, so putting it in test_bitcoin may be difficult
345 2012-08-29 22:26:39 <sipa> ACTION -> bed
346 2012-08-29 22:27:08 <BlueMatt> gnight
347 2012-08-29 23:14:12 <osmosis> trying to use decoderawtransaction fail http://dpaste.com/793426/
348 2012-08-29 23:14:48 <gmaxwell> osmosis: er, whatever that is you're giving it, it's not the right input.
349 2012-08-29 23:15:03 <gmaxwell> The raw transaction data is hex. that looks like base64?
350 2012-08-29 23:15:15 <gmaxwell> oh a second line.
351 2012-08-29 23:15:33 <osmosis> gmaxwell, im using tx's from https://mtgox.com/api/0/bitcoin_tx.php
352 2012-08-29 23:15:42 <gmaxwell> welp, I have no clue what it outputs.
353 2012-08-29 23:16:21 <gmaxwell> what txid is that supposted to be?
354 2012-08-29 23:16:54 <osmosis> im just grabbing any random one from that url. none of them work
355 2012-08-29 23:17:13 <gmaxwell> osmosis: "then stop using that url"
356 2012-08-29 23:17:29 <gmaxwell> osmosis: if you want one you can get the raw transaction for any txid w/ getrawtransaction
357 2012-08-29 23:17:39 <gmaxwell> if you need some txids, try a getblock
358 2012-08-29 23:39:02 <sebicas> I am getting this error 'Expected type obj, got str' from sendmany & createrawtransaction
359 2012-08-29 23:39:09 <sebicas> Any ideas?
360 2012-08-29 23:39:52 <gmaxwell> add more {}
361 2012-08-29 23:40:20 <sebicas> I try both '{"1AACtZp65EtF4mZ2fDHC2Fj5h1dGc8zu91":5000000}'
362 2012-08-29 23:40:24 <sebicas> And {"1AACtZp65EtF4mZ2fDHC2Fj5h1dGc8zu91":5000000}
363 2012-08-29 23:40:34 <gmaxwell> what platform?
364 2012-08-29 23:40:34 <sebicas> If I try '{"1AACtZp65EtF4mZ2fDHC2Fj5h1dGc8zu91":5000000}'
365 2012-08-29 23:40:38 <sebicas> Python
366 2012-08-29 23:40:56 <gmaxwell> oh in python you shoul be sending a dict
367 2012-08-29 23:41:59 <sebicas> I am using https://github.com/jgarzik/python-bitcoinrpc
368 2012-08-29 23:42:35 <sebicas> Yes <type 'dict'>
369 2012-08-29 23:43:08 <sebicas> Still getting value is type str, expected obj
370 2012-08-29 23:43:16 <sebicas> Debug Post Data: {"version": "1.1", "params": [["{\\"txid\\":\\"4abe438bbdea5d367dac3f58d9f9b6551d86568c5d5e02b81d76ce17bb8f11a8\\",\\"vout\\":0}"], {"1AACtZp65EtF4mZ2fDHC2Fj5h1dGc8zu91": 5000000}], "method": "createrawtransaction", "id": 1}
371 2012-08-29 23:43:56 <gmaxwell> worry about the sendmany first.
372 2012-08-29 23:44:34 <sebicas> Well if the same thing..
373 2012-08-29 23:44:44 <sebicas> https://github.com/bitcoin/bitcoin/blob/master/src/bitcoinrpc.cpp#L1136
374 2012-08-29 23:44:49 <sebicas> if (strMethod == "sendmany" && n > 1) ConvertTo<Object>(params[1]);
375 2012-08-29 23:44:54 <sebicas> if (strMethod == "createrawtransaction" && n > 1) ConvertTo<Object>(params[1]);
376 2012-08-29 23:45:56 <gmaxwell> you should work with the sendmany first.
377 2012-08-29 23:46:28 <gmaxwell> e.g. a={} a["1AACtZp65EtF4mZ2fDHC2Fj5h1dGc8zu91"]=5000000 and then using a as that argument should work
378 2012-08-29 23:48:18 <sebicas> Still "value is type str, expected obj"
379 2012-08-29 23:48:43 <gmaxwell> sebicas: putting in the account twice or something?
380 2012-08-29 23:48:49 <sebicas> It work from command line..
381 2012-08-29 23:48:55 <gmaxwell> pastebin your actual pyhton code.
382 2012-08-29 23:50:11 <sebicas> http://pastebin.com/CGeTenh8
383 2012-08-29 23:51:23 <gmaxwell> sebicas: Was I unclear when I asked you to work with sendmany first?
384 2012-08-29 23:51:40 <sebicas> Ok.. I will try that one then..
385 2012-08-29 23:51:43 <gmaxwell> In that case you've made a vector of strings for a, it probably needs to be a vector of dicts.
386 2012-08-29 23:52:09 <BlueMatt> alright, well afaict, I have every line which can actually occur in normal operation in the block connection logic covered except for tx.IsFinal and MoneyRange stuff, if someone wants MoneyRange, they can do it themselves...
387 2012-08-29 23:52:19 <gmaxwell> But the python json might actually be broken for createrawtransaction, I've never tested it with that; thats why I was suggesting sendmany, which I've used.
388 2012-08-29 23:52:39 <gmaxwell> BlueMatt: Is testing moneyrange even possible without patching?
389 2012-08-29 23:52:53 <sebicas> Ok, will try sandmany??? give me a sec
390 2012-08-29 23:52:54 <BlueMatt> sure
391 2012-08-29 23:53:15 <BlueMatt> you probably couldnt hit all of them, but a few of them are easy with negative/unrealistic values
392 2012-08-29 23:53:36 <gmaxwell> BlueMatt: hm, well I do think they're important to test.
393 2012-08-29 23:54:34 <BlueMatt> meh, spending a few days writing test after test kinda sucks your energy...
394 2012-08-29 23:55:20 <gmaxwell> ::nods:: well no rush, but it would make sense to complete it before bothering to mine it. How many blocks does it take?
395 2012-08-29 23:55:49 <BlueMatt> currently its at 61
396 2012-08-29 23:57:36 <BlueMatt> should probably check coverage of bitcoinj while Im at it, but...meh, I still need to go back and put in some infrastructure to get this bloom/block-as-v<tx hash> stuff done