1 2011-10-03 00:15:33 <phantomcircuit> shadders, you shouldn't be sending it
 2 2011-10-03 00:16:43 <shadders> yeah I thought it could be sent as an empty message to query to other end's inv.. like getaddr... but I see I was wrong.
 3 2011-10-03 02:24:05 <devrandom> gmaxwell: I'm not sure that the argument of valid->invalid vs invalid->valid makes sense.  In either case, the clients that did not upgrade will be on the wrong fork until they upgrade.
 4 2011-10-03 02:24:36 <gmaxwell> Not true.
 5 2011-10-03 02:25:13 <gmaxwell> They'll switch to the right fork once it has a longer chain, which if a super majority of the miners have upgraded in advance will always be true.
 6 2011-10-03 02:26:04 <gmaxwell> This isn't true for invalid->valid.
 7 2011-10-03 02:26:10 <gmaxwell> (as they'll never move)
 8 2011-10-03 02:28:35 <AlexWaters> anyone have success with a windows QT build?
 9 2011-10-03 02:32:01 <devrandom> gmaxwell: the assumption was that someone maliciously mines a tx that doesn't validate using the new rules, but validates under the old rules
10 2011-10-03 02:32:18 <gmaxwell> Yes.
11 2011-10-03 02:32:26 <gmaxwell> And the assumption is the miners have upgraded.
12 2011-10-03 02:32:28 <devrandom> oh, wait
13 2011-10-03 02:33:44 <devrandom> never mind, both forks are valid for old clients, so they'll get on the longer one
14 2011-10-03 02:35:32 <devrandom> gmaxwell: thanks for the handholding
15 2011-10-03 02:35:49 <gmaxwell> devrandom: No problem. I'm a dingbat too.
16 2011-10-03 02:38:17 <gmaxwell> There could even be some crazy rule on the upgraded nodes: "New txn will be permitted and the rules will begin being enforced at the first block number after 12345 when the prior 1000 blocks have not had the opcode, prior to that point I will not mine any txn with that opcode"
17 2011-10-03 02:38:41 <gmaxwell> And then you _intentionally_ inject txn that use the opcode.. and the network will activate only after almost all the miners are smart enough to reject it
18 2011-10-03 02:38:58 <gmaxwell> and everyone can agree on when it activated (because its an objective fact of the chain) so no splits.
19 2011-10-03 02:39:26 <gmaxwell> This would avoid the risk that you don't quite get a super majority of the miners and you end up with old clients on a seperate fork for days at a time.
20 2011-10-03 02:40:06 <gmaxwell> Actually I think I'll post about that, it's a useful thing to remember.
21 2011-10-03 02:40:13 <imsaguy> has anyone considered adding a simple 'update available' thing?
22 2011-10-03 02:40:34 <imsaguy> I think some people are just ignorant that a new version is out
23 2011-10-03 02:40:48 <Folklore> you run bitcoin now gmax?
24 2011-10-03 02:54:01 <devrandom> gmaxwell: maybe it makes sense to include a client version # in mined blocks, so that we can judge the % of power that has upgraded
25 2011-10-03 02:54:25 <gmaxwell> Version number are meaningless. Any large scale miner is on highly patched up code.
26 2011-10-03 02:55:33 <gmaxwell> Also, for this, the fact that aren't mining invalid txn is whats important not their version number.  E.g. the stock software won't mine these txn already, but some people have modified it to relax the rules and will mine them.
27 2011-10-03 02:56:43 <devrandom> malicious people could keep mining invalid txns to delay adoption
28 2011-10-03 02:59:03 <gmaxwell> Yep.
29 2011-10-03 02:59:10 <devrandom> with some kind of version number, which doesn't need to be same as software version, you could say "if 90% of past 1000 blocks have validation version n + 1, activate it"
30 2011-10-03 02:59:14 <gmaxwell> If they have enough hash power to always get a TXN in during the window.
31 2011-10-03 02:59:49 <devrandom> you are basically checking that 90% of hash power has the right validation capability before turning it on
32 2011-10-03 03:00:50 <gmaxwell> devrandom: but how do you count, e.g. nodes that don't mine any txn? It would be sad if your deployment failed because 10% who mine no txn at all also don't emit the message.
33 2011-10-03 03:00:52 <imsaguy> devrandom, screw 90, go 75
34 2011-10-03 03:01:21 <imsaguy> people will uninstall/reininstall/upgrade and solve their problem
35 2011-10-03 03:01:35 <gmaxwell> Well, it just has to be enough so that it's unlikely that the part that doesn't could outpace the part that does for more than 6 blocks at the transition time.
36 2011-10-03 03:01:36 <devrandom> gmaxwell: right, it only helps with gauging mining power on board, which is an important metric
37 2011-10-03 03:02:26 <devrandom> 0.25 ^ 6 = 1 in 4096
38 2011-10-03 03:02:43 <devrandom> reasonable
39 2011-10-03 03:03:32 <devrandom> gmaxwell: or wait, I think I don't understand your question
40 2011-10-03 03:03:57 <devrandom> the version # would be in the block, not the txn
41 2011-10-03 03:04:07 <gmaxwell> devrandom: I think the more important metric is the mining power on whatever side would make a split out of this. E.g. I think we could deploy eval _today_ because only a couple percent of the hashpower is willing to mine txn with nops... if we got one of the top four miners to deploy the patch then the part of the network that would mine a fork wouldn't be big enough to make a six deep split.
42 2011-10-03 03:04:13 <gmaxwell> devrandom: version numbers are really meaningless.
43 2011-10-03 03:04:35 <gmaxwell> No version will mine the conflicting txn, only modified software will.
44 2011-10-03 03:04:53 <gmaxwell> I have nodes running .21+patches which won't, luke has nodes with .24 with patches that will.
45 2011-10-03 03:05:15 <devrandom> gmaxwell: but the old miners will mine on top of the conflicting txn block, so that's no good
46 2011-10-03 03:05:31 <shadders> script opcode is 1 byte?
47 2011-10-03 03:05:46 <gmaxwell> devrandom: You only need to significantly outpace the conflicting miner, not the whole network.
48 2011-10-03 03:05:47 <devrandom> gmaxwell: you really do need >>50%
49 2011-10-03 03:05:53 <gmaxwell> No.
50 2011-10-03 03:06:50 <devrandom> if you deploy eval today and only 10% validate it, 90% of the network will mine on top of a conflicting block and the 10% will be swamped
51 2011-10-03 03:06:55 <gmaxwell> well, okay, I'll grand you that you'll lose money without it.
52 2011-10-03 03:07:21 <gmaxwell> devrandom: no the 10% will just move the goalpost further up and try again.
53 2011-10-03 03:07:39 <devrandom> well, that doesn't count as "deployment" ;)
54 2011-10-03 03:08:44 <gmaxwell> Okay, fair, I'm assuming gavin's discouragement patch.
55 2011-10-03 03:08:45 <devrandom> oh, you are saying deploy but don't recommend use... well, the minoring will lose money on hash power when a conflict arises
56 2011-10-03 03:09:01 <devrandom> ah, I see
57 2011-10-03 03:09:11 <devrandom> I don't know much about that one, will read up
58 2011-10-03 03:09:19 <devrandom> I have to run, talk to you later
59 2011-10-03 03:10:09 <gmaxwell> It's pretty simple: If you get a block you don't like you remember it but don't forward it, and don't mine against it, until you hear that its been extended, then you switch to the extended branch so you don't fall behind.
60 2011-10-03 03:35:35 <luke-jr> devrandom: basically, it's collusion
61 2011-10-03 03:35:55 <luke-jr> devrandom: and yes, unless the colluders get 50% hashpower total, their blocks will mostly be invalid
62 2011-10-03 03:36:51 <luke-jr> at least I oppose any kind of endorsed 50% attack without very good reason (like gmaxwell's block chain waster filling blocks with bs)
63 2011-10-03 06:12:21 <louigi> luke-jr, yo
64 2011-10-03 07:11:15 <Diablo-D3> Occupy Wall Street Protesters Accepting Donations in #Bitcoin; Big Test for the Controversial Virtual Currency - http://bit.ly/pC489i
65 2011-10-03 07:12:10 <AAA_awright> Aren't they kinda the antithesis of Bitcoin? At least, the progressives among them?
66 2011-10-03 07:12:15 <AAA_awright> I have no clue what they even represnet
67 2011-10-03 07:12:40 <AAA_awright> At least a few are anarchists, hence the Bitcoin maybe, but not of the peaceful sort
68 2011-10-03 07:12:58 <Diablo-D3> AAA_awright: they want wallstreet to shove it up their gold plated asses.
69 2011-10-03 07:13:04 <Diablo-D3> so Im pretty sure they're our kind of people
70 2011-10-03 07:13:35 <AAA_awright> End The Fed? Sure. Death to Capitalism... What?
71 2011-10-03 07:13:51 <Diablo-D3> dude, capitalism is just another name for slavery.
72 2011-10-03 07:14:14 <Diablo-D3> the only way for true captialism to survive is for government regulation on ALL threats to society.
73 2011-10-03 07:14:25 <epscy> human soceity is just another name for slavery
74 2011-10-03 07:14:28 <AAA_awright> Uh what?
75 2011-10-03 07:14:33 <Diablo-D3> AAA_awright: you heard me.
76 2011-10-03 07:14:36 <Diablo-D3> eat the bankers.
77 2011-10-03 07:14:51 <AAA_awright> Quite the opposite, capitalism implies you have /no/ right to another person's labor or property
78 2011-10-03 07:15:01 <epscy> Diablo-D3: you should watch "All watched over by machines of loving grace"
79 2011-10-03 07:15:12 <AAA_awright> That's something of a preresiquite
80 2011-10-03 07:15:30 <Diablo-D3> AAA_awright: not at all
81 2011-10-03 07:15:40 <Diablo-D3> goldman sachs paid a lot of money to redefine what capitalism means
82 2011-10-03 07:15:53 <AAA_awright> So if you're trying to end slavery in all it's forms, the only way you're headed is to one that's capitalist in some capacity, be it anarchocapitalism, liberalism, etc
83 2011-10-03 07:16:05 <Diablo-D3> thus, the destruction of any entity that prevents OTHERS from profiting from their labor must be destroyed
84 2011-10-03 07:16:08 <AAA_awright> ... Well then you're arguing against a by definition argument and you lose
85 2011-10-03 07:16:10 <Diablo-D3> thus the end of capitalism as it so exists.