1 2015-07-29 01:53:53 <leakypat> Is there research on block validation anywhere, eg average time to validate blocks at 1mb 8mb 20gb etc? Struggling to find...
  2 2015-07-29 03:19:23 <arix> hello, any brazilian here?
  3 2015-07-29 05:46:56 <Luke-Jr> grumble, apparently LibreSSL is adding the rand_egd function, which will break the detection
  4 2015-07-29 06:06:53 <wumpus> well BlueMatt made the point yesterday that it is way less important now that BIP66 is enforced. Any libressl/openssl differences would be *bugs* instead of e.g. BER decoding implementation differences.
  5 2015-07-29 06:33:48 <Luke-Jr> wumpus: yes, but bugs are a risk
  6 2015-07-29 06:35:25 <wumpus> but so are they for OpenSSL, there's nothing special there
  7 2015-07-29 06:36:32 <wumpus> this is 'just' a matter of implementing strict DER decoding + ECDSA correctly. We take a similar risk when we would switch verification to the secp256k1 library.
  8 2015-07-29 06:39:01 <wumpus> then again, I too feel better lifting the libressl restriction *after* completely severing OpenSSL from consensus code
  9 2015-07-29 06:46:01 <Luke-Jr> wumpus: OpenSSL is presently the network consensus.
 10 2015-07-29 06:46:29 <Luke-Jr> anyhow, we do have --with-libressl; IMO the important thing is to warn and have the user confirm they understand the risk
 11 2015-07-29 06:52:33 <wumpus> right.
 12 2015-07-29 10:04:10 <MrTratta> what is the right way of building core binaries with static boost and openssl other than using gitian? I've tried a couple of things but they are probably old suggestions I found around or I probably guessed wrong. I don't want to use gitian (I tried) because I don't care about Qt or wallet or upnp. I'm reasonably sure I can find out by looking at how gitian does it but I didn't how thus far.
 13 2015-07-29 10:05:07 <MrTratta> s/how/get how
 14 2015-07-29 10:05:11 <sipa> go to the depends directory and read the readme :)
 15 2015-07-29 10:05:27 <sipa> that way you'll do exactly the same as gitian, only without the vm
 16 2015-07-29 10:15:06 <MrTratta> thanks sipa. I did have a look there but it didn't occur to me to pass in the HOST flag. I tried to pass in to make HOST=x86_64_linux_cc but that didn't change, ldd still looks for boost externally. Maybe I got something wrong in my configure
 17 2015-07-29 10:18:21 <petertje> Hello
 18 2015-07-29 10:18:46 <wumpus> MrTratta: you need to pass the HOST flag to depends' make, then the right prefix to configure
 19 2015-07-29 10:19:19 <MrTratta> thanks wumpus, I was rereading the README and it seems I need to call make, then configure than make again
 20 2015-07-29 10:20:28 <wumpus> if bitcoin core's configure script keeps looking for boost elsewhere then the specified --prefix is wrong
 21 2015-07-29 10:21:06 <Luke-Jr> eh? --prefix controls output, not [just] input
 22 2015-07-29 10:24:37 <wumpus> yes, but it is used with the depends build as well to specify the input
 23 2015-07-29 10:25:59 <Luke-Jr> what if the user wants a normal prefix for output though?
 24 2015-07-29 10:26:05 <Luke-Jr> as gitian binaries should..
 25 2015-07-29 10:28:03 <wumpus> I don't know, I don't care, it's a shortcut that works because bitcoin doesn't actually use the prefix for anything
 26 2015-07-29 10:28:24 <wumpus> the other option is to specify the path per library - that also works, but is more work
 27 2015-07-29 10:29:51 <Luke-Jr> `make install` does use the prefix, but I guess not in a way that particularly is annoying to undo
 28 2015-07-29 10:41:41 <wumpus> right
 29 2015-07-29 12:41:35 <firelegend> Hi all. I'd like to ask a question that isn't related to bitcoin, but since some people here might have worked on pseudo-RNGs they might know something. I have a dozen thousand of sets of numbers that should be randomly generated, the numbers in each set are unique and they will not repeat within a set. What method can I use to test if there is a uniform distribution of these numbers? Or
 30 2015-07-29 12:41:35 <firelegend> to verify that there is or isn't any pattern to them, that some number in a set is or isn't likely to appear in another future set?
 31 2015-07-29 12:57:10 <unicodesnowman> firelegend, Hi! This is good background reading: https://en.wikipedia.org/wiki/Statistical_randomness#Tests
 32 2015-07-29 12:58:45 <unicodesnowman> firelegend, Dieharder will probs be useful for you. It's a series of tests for testing statistical randomness. http://www.phy.duke.edu/~rgb/General/dieharder.php
 33 2015-07-29 13:00:08 <sipa> the fact that all numbers in the set are different breaks the statistical indpendence of the numbers
 34 2015-07-29 13:00:28 <sipa> which needs different tests
 35 2015-07-29 13:39:29 <hearn> cfields: ping
 36 2015-07-29 13:52:06 <firelegend> sipa:The numbers generated do have one or two rules. They are limited to 20, and range from 1 through 80, but a number cannot be repeated within a set. If this requires different tests, what might they be?
 37 2015-07-29 13:53:33 <sipa> firelegend: well usually randomness testers are for independent uniform numbers
 38 2015-07-29 13:53:52 <sipa> those numbers don't have that distribution, as knowing some numbers in a set reveals some information about the others
 39 2015-07-29 14:03:48 <hearn> cfields: i thought gitian was supposed to cache builds of dependencies, these days? it doesn't do so for me, every time i build, it rebuilds every single dependency
 40 2015-07-29 14:03:50 <hearn> cfields: is there a way to fix this
 41 2015-07-29 15:27:35 <cfields> hearn: it does cache them
 42 2015-07-29 15:27:39 <wumpus> the caching works fine here; are you sure the dependencies haven't been updated between your build attempts?
 43 2015-07-29 15:27:54 <cfields> hearn: maybe your gitian is from before the cache feature was added?
 44 2015-07-29 15:29:08 <jonasschnelli> Also works here. Normal 3 platforms build with cached deps take about 35min: (example: https://builds.jonasschnelli.ch/pulls/6315/), ... with new (partial) deps build  it take mostly more then 60mins: https://builds.jonasschnelli.ch/pulls/6471/
 45 2015-07-29 15:30:26 <wumpus> (eg for master in the last weeks all the dependencies, including Qt, have been bumped, for 0.11 branch just Qt)
 46 2015-07-29 15:30:54 <hearn> cfields: ah
 47 2015-07-29 15:30:59 <hearn> cfields: what's "my gitian" in this context?
 48 2015-07-29 15:31:15 <hearn> cfields: my VM setup? i do see "Caching foo....." type messages printed to the build log. but then when i rerun gbuild, it does it all over again
 49 2015-07-29 15:31:17 <cfields> hearn: your gitian git clone
 50 2015-07-29 15:31:37 <hearn> hm
 51 2015-07-29 15:31:42 <hearn> what am i supposed to see when a dep is cached?
 52 2015-07-29 15:31:45 <hearn> in the build.log file i mean
 53 2015-07-29 15:32:02 <hearn> my gitian is from   "Date:   Tue Nov 25 11:30:07 2014 -0800"
 54 2015-07-29 15:32:06 <hearn> is that too old?
 55 2015-07-29 15:32:14 <wumpus> nothing - it will just not build it
 56 2015-07-29 15:33:07 <cfields> hearn: sounds like that's too old. in the release process doc it specifies that you need to be newer than gitian commit xyz
 57 2015-07-29 15:33:23 <hearn> ah ok
 58 2015-07-29 15:33:40 <cfields> In order to take advantage of the new caching features in gitian, be sure to update to a recent version (`e9741525c` or later is recommended)
 59 2015-07-29 15:34:35 <hearn> ah ha. that's in release-process.md, i was looking at gitian-building.md
 60 2015-07-29 15:34:50 <hearn> however i do have e9741525c in my gitian directory :(
 61 2015-07-29 15:34:58 <hearn> that commit was on Tue Nov 11 15:41:54 2014 -0800
 62 2015-07-29 15:35:27 <hearn> i'll update to latest gitian master when this build is done and see if it changes anything
 63 2015-07-29 15:45:46 <jonasschnelli> Copay is doing 1of1 multisig for a personal wallet. Are there any drawbacks if a wallet uses 1of1 multisig by default? advantages/disadvantages (general verification time on 1of1 instead of P2PKH)?
 64 2015-07-29 15:46:13 <sipa> nearly no difference
 65 2015-07-29 15:47:47 <wumpus> "1-of-1 multisig" is simply  a P2PKH through P2SH I suppose?
 66 2015-07-29 15:47:57 <tripleslash> slight difference in tx size
 67 2015-07-29 15:48:31 <sipa> wumpus: or maybe with OP_CHECKMULTISIG
 68 2015-07-29 15:48:32 <afk11> wumpus: no - 1 of 1 still can use op_checkmultisig IIRC
 69 2015-07-29 15:48:45 <sipa> but that's at most a few bytes difference iirc
 70 2015-07-29 15:49:21 <afk11> whereas, you can still have your redeemScript be a p2pkh script
 71 2015-07-29 15:49:22 <wumpus> ah of course
 72 2015-07-29 15:49:53 <sipa> afk11: or a raw op_checksig
 73 2015-07-29 15:50:37 <jonasschnelli> so it's unproblematic if a wallet uses 1of1 by default to avoid implementation complexity while supporting HDM?
 74 2015-07-29 15:52:05 <afk11> jonasschnelli: probably not a problem. copay does this IIRC
 75 2015-07-29 15:52:20 <sipa> 17:45:46 < jonasschnelli> Copay is doing 1of1 multisig
 76 2015-07-29 15:52:46 <hearn> jonasschnelli: it reduces compatibility with old software that doesn't understand p2sh and is a bit less efficient. i'm not sure it really reduces implementation complexity much ...
 77 2015-07-29 15:53:46 <jonasschnelli> You don't need to distinct between  normal bip32 wallets and hd multisig wallets in your code... not saying it dramatically reduces the code.
 78 2015-07-29 15:54:34 <jonasschnelli> but the way how you generate addresses, how you prepare for signing would be simpler than supporting P2PKH along to HDM.
 79 2015-07-29 15:55:11 <afk11> sipa: ah, should have scrolled
 80 2015-07-29 15:57:14 <jonasschnelli> hearn: are there any type of software that doesn't understand p2sh?
 81 2015-07-29 15:57:53 <jonasschnelli> I think a wallet only communicates with nodes or other wallets which could "speak" MS, and therefore must understand p2sh?
 82 2015-07-29 15:59:10 <hearn> jonasschnelli: i guess old software doesn't.
 83 2015-07-29 15:59:15 <hearn> jonasschnelli: how many people use such programs is anyone's guess.
 84 2015-07-29 15:59:47 <hearn> jonasschnelli: the issue is if somebody wants to send you money and their wallet doesn't do p2sh
 85 2015-07-29 15:59:56 <hearn> jonasschnelli: network nodes don't matter
 86 2015-07-29 16:00:38 <afk11> I haven't heard of a wallet that didn't support sending to p2sh in a while.
 87 2015-07-29 16:00:40 <jonasschnelli> ah.. i see you point. Sure. But this wallet app then would have serious problems with receiving any sending (signing p2sh unspents)
 88 2015-07-29 16:03:18 <hearn> no, why would it? wallets don't care about how money was sent *to* them
 89 2015-07-29 16:06:38 <jonasschnelli> hearn: i think they verify the transaction itself (no P2SH support = cannot verify). And signing P2SH unspents would also not possible. So using bitpay would be impossible (i assume same for coindesk)
 90 2015-07-29 16:07:06 <hearn> only fully verifying wallets verify transactions they are sent and they were forked out of the network a long time ago
 91 2015-07-29 16:07:36 <hearn> yes, obviously wallets need to receive outputs that they understand. my point is that how the sender signed for their own money doesn't matter
 92 2015-07-29 16:07:42 <hearn> sorry, i wasn't clear before
 93 2015-07-29 16:11:01 <afk11> jonasschnelli: do you see HDM making it's way into a later PR? :)
 94 2015-07-29 16:32:26 <hearn> hm, is there something somewhere that controls which libraries are statically linked into the bitcoind binary ?
 95 2015-07-29 16:32:35 <hearn> i can't see anything obvious in the make files or configure scripts
 96 2015-07-29 16:36:23 <afk11> hearn: try https://github.com/bitcoin/bitcoin/blob/master/src/Makefile.am
 97 2015-07-29 16:39:16 <hearn> yeah i looked at that
 98 2015-07-29 16:39:31 <hearn> e.g. where is the code that says openssl should be statically linked, or boost?
 99 2015-07-29 16:39:43 <hearn> i suspect it's some libtool thing
100 2015-07-29 18:16:29 <hearn> jonasschnelli: when you gbuild and get caching, are you using the -i switch?
101 2015-07-29 18:38:39 <tucenaber> cal16gustf
102 2015-07-29 18:38:53 <tucenaber> oops
103 2015-07-29 18:39:29 <arubi> this channel makes a good password dictionary.
104 2015-07-29 18:47:23 <jonasschnelli> hearn: (sorry was afk), no, not using the -i switch. My script follows the release-process.md pretty much.
105 2015-07-29 18:47:56 <jonasschnelli> afk11: yeah. HDM is on the way over the corewallet fork... so will take a couple of month till its ready.
106 2015-07-29 21:12:39 <go1111111> +1 one to the suggestions to add some more controls to which posts can show up on the dev mailing list, along with a clear policy on what is appropriate there. The recent "Personal opinion on the fee market from a worried local trader" is another example of something that seems out of place
107 2015-07-29 21:22:52 <riwi> quit
108 2015-07-29 21:22:57 <riwi> :(
109 2015-07-29 22:27:02 <jcorgan> regarding dev list issues, i suggest we add temporary 1MB limit to the amount that can be posted in a given thread :-)
110 2015-07-29 22:27:19 <kadoban> XD
111 2015-07-29 22:30:54 <thufir> anyone here a windows user, familiar with python 3? looking for a team member to join my open source project, asap :) need help to make the windows package while i finish coding for the release on the 31st... http://morph.is/description.html     i promise you the first payment from the donation pot for your time. (i won't be taking any money from the project, only directing its flow, and 100% transparent)
112 2015-07-29 22:33:34 <Luke-Jr> thufir: no spam here please; this channel is only for development of bitcoin
113 2015-07-29 22:33:58 <teward> ^ that
114 2015-07-29 22:34:00 <teward> (he ninja'd me)
115 2015-07-29 22:35:08 <thufir> ok, know that morphis is lightning network before that guy came about,
116 2015-07-29 22:35:32 <thufir> morhpis was originally conceived to provide a microtransaction network for bitcoin, and will, i calculated that a datastore is needed to enable it, and thus i built one :)
117 2015-07-29 22:36:12 <thufir> so i disagree completely it is spam, it is absolutely development of bitcoin, they are to be symbiants
118 2015-07-29 22:36:33 <thufir> imagine maid safe but using bitcoin and providing inherent microtransaction and mixing layer
119 2015-07-29 22:36:43 <jcorgan> thufir: not here
120 2015-07-29 22:37:38 <thufir> oh, no talking about developing bitcoin in the bitcoin-dev channel? no prob, I expected that here on your crazy ass run irc channels. i won't mention it again here then until asked
121 2015-07-29 22:46:30 <StephenM347> What version of JSON RPC does bitcoin core currently support?
122 2015-07-29 22:50:23 <StephenM347> Seems to be 2.0, but I'm using a 2.0 library and getting error: "Invalid Response (0): Missing member: jsonrpc"
123 2015-07-29 22:50:32 <StephenM347> Anyway, seems to be a problem with the library
124 2015-07-29 23:06:17 <moa> that signal:noise on bitcoin-dev just spiked again ... :( ... is it correlated with the spam attacks?