1 2013-03-30 00:00:27 <TheLordOfTime> owowo, i've already dropped a note to nanotube about the bot not working with ;;ticker and such
 2 2013-03-30 00:03:26 <Luke-Jr> yay, Bitcoin-Qt using over 1 GB RAM :/
 3 2013-03-30 00:03:35 <TheLordOfTime> owowo, and lo-and-behold, nanottube wakes up after i say that :P
 4 2013-03-30 00:03:42 <Diablo-D3> gmaxwell: shouldnt you be doing the multisig in here and not #bitcoin?
 5 2013-03-30 00:05:37 <helo> multisig is clearly bitcoin usage, rather than development
 6 2013-03-30 00:06:40 <helo> i'd prefer that the noisy people be silenced, rather than the entire channel :/
 7 2013-03-30 00:06:57 <sipa> Luke-Jr: the bitcoind on my VPS is using 275 MiB RES, 63 connections, running for an hour or so
 8 2013-03-30 00:07:08 <sipa> Luke-Jr: with several of my and BlueMatt's pullreqs
 9 2013-03-30 00:09:13 <BlueMatt> sipa: would you merge tiny wrapper code around your secp256k1 thing for bitcoinj?
10 2013-03-30 00:09:41 <sipa> BlueMatt: what would be needed?
11 2013-03-30 00:09:55 <BlueMatt> its 1 cpp 1 h and a few lines in Makefile
12 2013-03-30 00:10:13 <BlueMatt> (not built by default because you need jni headers)
13 2013-03-30 00:10:38 <sipa> i'm in the process of converting it to C; would that simpify things?
14 2013-03-30 00:10:46 <BlueMatt> wouldnt change, really
15 2013-03-30 00:11:09 <BlueMatt> I suppose it could remove the namespace secp256k1 line...
16 2013-03-30 00:11:16 <sipa> meh
17 2013-03-30 00:11:47 <sipa> it's fine, i guess
18 2013-03-30 00:12:20 <sipa> the API will probably be extended with a few more functions to generate/import/export keys, and sign messages
19 2013-03-30 00:12:23 <sipa> but that's about ti
20 2013-03-30 00:12:40 <BlueMatt> well, verification is the only really time-limited thing in bitcoinj
21 2013-03-30 00:12:47 <BlueMatt> yea, signing isnt fast, but its rare enough...
22 2013-03-30 00:13:00 <sipa> signing as such is very easy
23 2013-03-30 00:13:14 <sipa> the hard part is making it not leak key data via timing
24 2013-03-30 00:13:16 <Scrat> 
ACTION shakes fist at gmaxwell

25 2013-03-30 00:13:25 <BlueMatt> sipa: isnt that always the hard part?
26 2013-03-30 00:13:36 <sipa> for verification that doesn't matter :D
27 2013-03-30 00:13:41 <BlueMatt> well, yes
28 2013-03-30 00:13:55 <sipa> verification is mathematically a lot more complex
29 2013-03-30 00:14:30 <BlueMatt> 
ACTION punts on timing and just writes it reasonably and makes gcc compile certain function -O0

30 2013-03-30 00:14:31 <BlueMatt> https://github.com/TheBlueMatt/shamirs/blob/master/shamirssecret.c#L68
31 2013-03-30 00:15:33 <sipa> BlueMatt: oh, but it's certainly doable; for most of low-level code i already have constant-time functions
32 2013-03-30 00:15:57 <BlueMatt> yea, when gcc doesnt get smart and optimize out your constant-time functions...
33 2013-03-30 00:15:57 <sipa> anyway, the reason to add signing is to be able to drop OpenSSL-EC entirely
34 2013-03-30 00:16:14 <BlueMatt> yea
35 2013-03-30 00:16:18 <sipa> even if it optimizes away my constant-time functions, they'll still be constant time :)
36 2013-03-30 00:16:29 <sipa> (they're branchless)
37 2013-03-30 00:16:53 <BlueMatt> well, ok, as long as you can write it branchless
38 2013-03-30 00:16:55 <DBordello> Does this transaction look valid to you guys:  52eb8e70428dbcfc3c93089d35919efe96a8d27a402350e6d6e79c352b00384b
39 2013-03-30 00:17:25 <DBordello> It looks reasonable to me, but blockchain.info doesn't like the input
40 2013-03-30 00:17:44 <sipa> how would i know what the transaction is?
41 2013-03-30 00:18:01 <DBordello> sipa, I believe that is the txid
42 2013-03-30 00:18:16 <CodeShark> any 256-bit number is a "reasonable" txid
43 2013-03-30 00:18:16 <sipa> a txid is not a transaction
44 2013-03-30 00:18:37 <sipa> and the txid surely looks fine, but i have no way of saying anything useful about the transaction without seeing it
45 2013-03-30 00:18:49 <DBordello> well, has your client seen it?
46 2013-03-30 00:19:09 <CodeShark> yes
47 2013-03-30 00:19:28 <sipa> oh, apparently it ha
48 2013-03-30 00:19:29 <sipa> s
49 2013-03-30 00:19:33 <DBordello> Does it think it is valid?
50 2013-03-30 00:20:10 <sipa> must be
51 2013-03-30 00:20:32 <CodeShark> unconfirmed, but satisfies the protocol rules
52 2013-03-30 00:20:51 <DBordello> I wonder why the network hates it
53 2013-03-30 00:21:13 <CodeShark> it has a long chain of transactions dependent on it
54 2013-03-30 00:21:13 <DBordello> Or coincidence
55 2013-03-30 00:21:42 <DBordello> CodeShark, I had a hunch that was the case, but neither blockexplorer or blockchain wanted to help me visualize it
56 2013-03-30 00:21:46 <BlueMatt> sipa: ahh, wait
57 2013-03-30 00:21:48 <CodeShark> is it a deliberate double-spend attempt?
58 2013-03-30 00:21:58 <BlueMatt> sipa: (dont merge secp256k1 thinggy)
59 2013-03-30 00:22:04 <DBordello> CodeShark, I hope not.  If so, we are not falling for it
60 2013-03-30 00:23:19 <BlueMatt> sipa: ok, sorry
61 2013-03-30 00:23:23 <CodeShark> DBordello: I traced it back to f058c655a42af9bceb0e4a09e096cc406c5e7fab2b560d4cd4303cf020005b93