1 2012-12-23 00:00:01 <gmaxwell> rdymac: why would it?
2 2012-12-23 00:00:22 <phantomcircuit> rdymac, even at the maximum blocksize it wont be a serious issue for a very long time
3 2012-12-23 00:00:23 <rdymac> 128GB SSD?
4 2012-12-23 00:00:40 <phantomcircuit> and almost certainly disk space will drop in price faster than the blockchain grows in size
5 2012-12-23 00:00:58 <rdymac> good point
6 2012-12-23 00:01:06 <gmaxwell> rdymac: thats very small media by current standards, and sort of odd to store archival data on a slc ssd..., but even there??? do the math. It's not much of a concern.
7 2012-12-23 00:01:52 <MC1984> are 6tb disks out yet?
8 2012-12-23 00:02:14 <rdymac> ok - if it doesn't use the blockchain to validate new blocks, what's the method then?
9 2012-12-23 00:02:24 <rdymac> 3TB
10 2012-12-23 00:02:34 <gmaxwell> there are 4tb drives.
11 2012-12-23 00:02:48 <rdymac> 4*
12 2012-12-23 00:02:59 <phantomcircuit> the 3TB drives are the best $/GB currently
13 2012-12-23 00:03:03 <rdymac> even Hybrid
14 2012-12-23 00:03:22 <gmaxwell> rdymac: it uses a 'coins' database, which is a structure with all the spendable txouts. It's whats left if you throughly prune a blockchain as tightly as possible and pack the results.
15 2012-12-23 00:03:22 <phantomcircuit> either way you're talking like $2 to store the entire blockchain
16 2012-12-23 00:03:53 <gmaxwell> Right now that structure is 125MBytes, though its growing a fair bit because of the never-getting-spent 1e-8 txouts being produced by SD.
17 2012-12-23 00:03:53 <sipa> gmaxwell: aha, interesting
18 2012-12-23 00:04:26 <gmaxwell> sipa: I'm glad to have found a real bug. I've had bad testing luck this month! :P
19 2012-12-23 00:04:44 <sipa> gmaxwell: that check should be added before the flush when the cache size is extended
20 2012-12-23 00:05:18 <sipa> you can calculate how much it can grow basrd on the cache size settings
21 2012-12-23 00:05:33 <sipa> *exceeded
22 2012-12-23 00:06:24 <gmaxwell> it should have a couple other checks too. e.g. log writing should probably check every N seconds too.
23 2012-12-23 00:06:48 <MC1984> only 4tb
24 2012-12-23 00:06:55 <MC1984> seems like weve been stuck around there forever
25 2012-12-23 00:06:56 <gmaxwell> MC1984: ??? in a single disk.
26 2012-12-23 00:07:18 <sipa> gmaxwell: debug.log writing you mean?
27 2012-12-23 00:07:34 <MC1984> yeah single disk
28 2012-12-23 00:07:37 <MC1984> so what
29 2012-12-23 00:07:46 <gmaxwell> sipa: correct.
30 2012-12-23 00:08:01 <MC1984> theyve should be over 1tb per platter by now
31 2012-12-23 00:08:40 <sipa> "they" and "should" depend on the market
32 2012-12-23 00:09:49 <MC1984> you think the market is running out of demand for MOAR SPACE
33 2012-12-23 00:10:48 <gmaxwell> MC1984: there are a bunch of drives with 1TB platters now.
34 2012-12-23 00:11:49 <MC1984> cool
35 2012-12-23 00:12:23 <MC1984> i wonder when theyll be putting tiny cutle little electron force microscopes in them
36 2012-12-23 00:12:35 <MC1984> cant be far off
37 2012-12-23 00:12:41 <gmaxwell> How do you think they work now?
38 2012-12-23 00:12:54 <sipa> magic.
39 2012-12-23 00:13:41 <MC1984> they dont directly manipulate magnetism anymore afaik
40 2012-12-23 00:14:00 <gmaxwell> Seems that the 3TB drives got the higher density platters first, I assume to lower materials cost on the more price sensitive products.
41 2012-12-23 00:14:25 <MC1984> i wonder if they have to take account of quantum effects yet
42 2012-12-23 00:14:35 <MC1984> to the wikimobile
43 2012-12-23 00:14:52 <gmaxwell> ...
44 2012-12-23 00:15:04 <gmaxwell> yea, this is all answered, google giant magnetoresistance.
45 2012-12-23 00:15:32 <MC1984> giant? dont be silly, this stuff is very very small
46 2012-12-23 00:16:09 <gmaxwell> Don't look at me, I can't explain this physics stuff... and whos idea was it to give electrons _negative_ charge?!
47 2012-12-23 00:16:52 <MC1984> i know rite
48 2012-12-23 00:18:12 <MC1984> Worldwide revenues for HDDs shipments are expected to reach $38 billion in 2012, up about 19% from $32 billion in 2011.
49 2012-12-23 00:18:26 <MC1984> wow i thought disks were on the way out
50 2012-12-23 00:18:47 <gmaxwell> They don't mention that 80% of thats going to the NSA. :P
51 2012-12-23 00:19:03 <phantomcircuit> lolol
52 2012-12-23 00:19:04 <sipa> gmaxwell: given the choice, which would you give the negative cgarge... electrons or POSItrons?
53 2012-12-23 00:19:16 <MC1984> shiiiiiiit
54 2012-12-23 00:19:35 <gmaxwell> hahah.
55 2012-12-23 00:20:02 <gmaxwell> That just shows how mistakes compound over time. :P
56 2012-12-23 00:20:09 <MC1984> it would have helped if theyd got current flow direction right the first time
57 2012-12-23 00:20:43 <MC1984> i was taught something vaguely about "hole current" and then no one ever spoke it it again
58 2012-12-23 00:20:58 <Dagger2> MC1984: the lack of competition in the HDD space means that prices have gone up. 19% revenue increase could easily come from that
59 2012-12-23 00:21:26 <MC1984> it was the flood wasnt it?
60 2012-12-23 00:21:47 <MC1984> they decided they liked the new price level and the market was bearing it anyway
61 2012-12-23 00:22:26 <Dagger2> you'd expect one of their competitors to undercut them to get a bigger share of the market
62 2012-12-23 00:22:42 <Dagger2> but no, "one of their competitors" no longer applies in the desktop HDD market, it's just "their competitor" now
63 2012-12-23 00:23:00 <Dagger2> and two companies isn't enough competition
64 2012-12-23 00:23:34 <MC1984> yeah duopoly helps
65 2012-12-23 00:23:47 <MC1984> samsung spinpoints were the best too :(
66 2012-12-23 00:24:07 <MC1984> fuck seagate
67 2012-12-23 00:26:38 <MC1984> so anyway the moral of the story is dont worry about chain size vis a vis storage capacities
68 2012-12-23 00:27:36 <MC1984> dice pays fees right? But fees are flat rate according to each miner
69 2012-12-23 00:27:54 <MC1984> but its clear with the new DB that some txns are more onerous than others
70 2012-12-23 00:28:03 <Dagger2> ACTION has his block chain on an SSD
71 2012-12-23 00:28:03 <MC1984> like the shit coming out of dice
72 2012-12-23 00:28:15 <Dagger2> and it'd be really helpful if it was smaller
73 2012-12-23 00:28:19 <MC1984> what if these probematic txns were charged more?
74 2012-12-23 00:28:28 <MC1984> ones that cant be pruned, bit dust etc
75 2012-12-23 00:29:04 <gmaxwell> MC1984: we do require a fee for any txn that has very small outputs, but people pay it.
76 2012-12-23 00:29:05 <sipa> Dagger2: if you're on 0.8, only put the block and coin databased on fast storage
77 2012-12-23 00:29:58 <gmaxwell> MC1984: we need to teach the wallet to clean up its mess e.g. prefer to gobble up extra txouts as it can.
78 2012-12-23 00:30:04 <MC1984> gmaxwell its not even slowing dice down
79 2012-12-23 00:30:32 <gmaxwell> MC1984: why would it? their suckers^wcustomers pay for it.
80 2012-12-23 00:31:09 <Dagger2> I'm on whichever version was the last to ship with the wxwidgets interface
81 2012-12-23 00:32:21 <gmaxwell> ...
82 2012-12-23 00:32:27 <Dagger2> I don't have any other storage I can use apart from network drives though
83 2012-12-23 00:32:28 <gmaxwell> Dagger2: And that still runs??
84 2012-12-23 00:32:29 <gmaxwell> :P
85 2012-12-23 00:32:41 <MC1984> seem the 'leaderboards' on that site
86 2012-12-23 00:32:43 <Dagger2> I dunno, I haven't tried in a while :p
87 2012-12-23 00:32:48 <gmaxwell> I'm surprised you're not stuck.
88 2012-12-23 00:32:48 <MC1984> some people have a genuine problem
89 2012-12-23 00:33:09 <MC1984> spend 11,000 btc win 2,000
90 2012-12-23 00:33:21 <Dagger2> but still, the Qt interface irritated me so I figured I'd just ignore it
91 2012-12-23 00:33:54 <sipa> Dagger2: patches welcome :)
92 2012-12-23 00:34:38 <gmaxwell> Dagger2: well the old versions have fairly serious bugs that will leave them subject to getting isolated potentially maliciously.
93 2012-12-23 00:34:54 <Dagger2> the patch would be the wxwidgets interface, with maybe a fix for the toolbar background color over RDP :p
94 2012-12-23 00:35:25 <Dagger2> as a serious question, can I feasibly use that interface from 0.4 with more recent core code? or would that involve a fair bit of effort?
95 2012-12-23 00:36:05 <MC1984> doesnt luke maintain one with the old gui
96 2012-12-23 00:36:08 <sipa> no
97 2012-12-23 00:36:36 <sipa> the stable backports are bitcoind only, afaik
98 2012-12-23 00:36:55 <MC1984> whats wrong with qt anyway
99 2012-12-23 00:36:58 <sipa> Dagger2: are you serious?
100 2012-12-23 00:37:43 <gmaxwell> it would be an enormous amount of work??? and well, there was _no one_ who wanted to maintain the wxwidgets code. Lots of people are willing to hack on the QT stuff.
101 2012-12-23 00:38:06 <Dagger2> I know nothing about bitcoin's architecture, I have no idea if that should be easy or not
102 2012-12-23 00:38:09 <sipa> the wx interface was not written on top of the core (like qt), but intermingled with it
103 2012-12-23 00:38:43 <Dagger2> if the UI was just an interface to a bitcoind then it should just be a matter of upgrading the daemon while not touching the UI, but I don't know if it is or not
104 2012-12-23 00:38:48 <Dagger2> ... well, I guess I do now
105 2012-12-23 00:40:52 <sipa> i'm really surprised how much hatred an interface seems to cause for some people
106 2012-12-23 00:41:46 <sipa> except when you have some ideologic problem with Qt, I really don't understand it
107 2012-12-23 00:42:14 <gmaxwell> If there are specific issues with how the QT interface works- please open issues on them! (or better, submit patches??? it's now pretty easy to hack on)
108 2012-12-23 00:43:30 <Dagger2> no ideological problem with Qt per se... but it does produce a UI that's subtly wrong for my OS+theme, whereas the wxwidgets one was perfect
109 2012-12-23 00:44:35 <sipa> 'subtly wrong' sounds easily fixable if reported! and hardly a reason to expose yourself to security risks
110 2012-12-23 00:44:46 <Dagger2> the transaction list uses some weird Qt widget with shaded tall rows with pictures in them, vs the old regular (can't remember what Windows calls the widget -- "listbox" maybe) in the wxwidgets UI
111 2012-12-23 00:44:50 <Dagger2> and no status bar
112 2012-12-23 00:45:11 <Dagger2> instead some pictures that I have to mouse over to get the text that I used to be able to just read in the UI
113 2012-12-23 00:45:20 <Dagger2> (this is all from 0.5, it might have changed since then)
114 2012-12-23 00:45:44 <gmaxwell> it's been tweaked marginally some since 0.5... there were some larger proposed changes that petered out before actually getting done.
115 2012-12-23 00:50:24 <Dagger2> and a major part of the annoyance is not that I don't like the UI, but that somebody exerted effort to go from a UI I was completely happy with to one that I really wasn't
116 2012-12-23 00:51:32 <Dagger2> Firefox does that to me constantly, and then they have the gall to tell me that it's for my own good and that, if I would only just realize it, I actually secretly like the changes
117 2012-12-23 00:51:34 <Dagger2> which I don't
118 2012-12-23 00:51:40 <gmaxwell> Dagger2: But what you're not considering is that the change also moved it from software that no one was willing to maintain, to software that many people are willing to work on... and also a lot of people also prefer the qt interface too. (for reasons mostly different than what makes you dislike it)
119 2012-12-23 00:51:52 <Dagger2> though of course in this case "nobody was willing to hack on our old UI" is a reason I can understand for writing a new one
120 2012-12-23 00:52:21 <gmaxwell> And it was deeply interwoven into the core of the system??? a complete mess. So it had to be rewritten.
121 2012-12-23 00:52:47 <gmaxwell> There probably are some interface taste issues that can't be reconciled, but I expect most of what you dislike can be fixed.
122 2012-12-23 00:58:00 <Dagger2> a lot of it is a result of Qt using its own implementations over system-native widgets
123 2012-12-23 00:59:02 <Dagger2> which can't be so easily fixed by us or by Qt
124 2012-12-23 01:00:37 <sipa> just use bitcoind :p
125 2012-12-23 01:02:07 <gmaxwell> not like system native widgets has much meaning anymore in any case... with webaps and whatever the new windows 8 stuff is called.
126 2012-12-23 01:08:25 <Dagger2> quite, but "other people are ditching consistent UIs too" is no reason to do so
127 2012-12-23 01:08:49 <Dagger2> at least it's certainly not one that will make me happy to use the resulting software
128 2012-12-23 01:09:00 <rdymac> piles?
129 2012-12-23 01:11:25 <gmaxwell> Dagger2: I mean its a lost cause regardless of how you value it.
130 2012-12-23 01:13:24 <gmaxwell> Dagger2: (and QT does use the native libraries and widgets though I admit I don't know much about the details)
131 2012-12-23 01:19:58 <Dagger2> to some degree it must do, but some parts it definitely doesn't (e.g. the listview looks nothing like a Windows common control listview)
132 2012-12-23 01:20:42 <Dagger2> or dropdown option boxes. I can't figure out exactly what's wrong with them, but they're clearly not the native ones
133 2012-12-23 01:27:33 <Luke-Jr> [01:35:41] <Dagger2> as a serious question, can I feasibly use that interface from 0.4 with more recent core code? or would that involve a fair bit of effort? <-- 0.4.x is still maintained for bitcoind, yes
134 2012-12-23 01:28:48 <Luke-Jr> Dagger2: sounds like you should suggest changes to Bitcoin-Qt to make it more to your liking; but if you want to pick up wxBitcoin and maintain it, I'd be glad to assist a bit
135 2012-12-23 01:29:57 <Luke-Jr> Dagger2: I suspect the listview looks different because of styling rules more than Qt, if it does indeed look different
136 2012-12-23 01:36:23 <Dagger2> realistically I'd spend maybe a week hacking on it before promptly losing interest, so I'd have to get it to the point where it was an isolated UI that could cope with bitcoind upgrades without breaking
137 2012-12-23 01:36:56 <Dagger2> and from the sound of things that would be difficult, possibly to the point of it being easier to rewrite from scratch
138 2012-12-23 01:38:03 <Dagger2> it doesn't actually look like it would be too much effort to do that if it's just a UI on top of bitcoind... but software projects have a tendency to never be "not too much effort"
139 2012-12-23 01:40:02 <phantomcircuit> the time to load the blockchain is getting a lot worse :(
140 2012-12-23 01:41:50 <gmaxwell> phantomcircuit: by worse you mean a lot better?
141 2012-12-23 01:42:11 <phantomcircuit> i mean using the same version it's getting worse
142 2012-12-23 01:42:21 <gmaxwell> ha, well, duh. :P
143 2012-12-23 01:42:25 <phantomcircuit> 0.8 islot better
144 2012-12-23 02:26:41 <Luke-Jr> Dagger2: well, you could get started from the bitcoind 0.4.x code - I never removed the GUI code (that could have introduced bugs)
145 2012-12-23 02:27:04 <Luke-Jr> obviously you won't get LevelDB and such, but it's a step
146 2012-12-23 02:27:36 <gmaxwell> or a ton of other features. .. but at least it won't be outright busted and vulnerable (well, we hope)
147 2012-12-23 02:28:02 <Luke-Jr> gmaxwell: yeah, but I think LevelDB is the only thing that would be difficult to adapt ;p
148 2012-12-23 02:28:58 <gmaxwell> Luke-Jr: uh, there are a bunch of features with gui parts that have never been written as wx.
149 2012-12-23 02:29:11 <Luke-Jr> yeah, but most of them are trivial
150 2012-12-23 02:29:43 <Luke-Jr> (compared to LevelDB)
151 2012-12-23 02:30:25 <MC1984> if leveldb is designed to aplit the chain into storage and active use sections
152 2012-12-23 02:30:51 <MC1984> could it put those different sections on appropriate media in the machine, if available
153 2012-12-23 02:31:09 <MC1984> hybrid drives and stuff
154 2012-12-23 02:31:17 <Luke-Jr> MC1984: well, leveldb included ultraprune, which basically means you can just delete the block files when it's done I thin
155 2012-12-23 02:31:18 <Luke-Jr> think*
156 2012-12-23 02:31:40 <MC1984> yeah but i hope that will never be neccesary
157 2012-12-23 02:31:55 <gmaxwell> MC1984: you can do that just by symlinking the directories.. works fine.
158 2012-12-23 02:32:17 <MC1984> yes but i am john q dumbass user
159 2012-12-23 02:32:21 <gmaxwell> Luke-Jr: no, you can't delete them currently, not without breaking things (and the network as a whole)
160 2012-12-23 02:32:35 <gmaxwell> MC1984: splitting your system across multiple media is not a dumbass user thing to do.
161 2012-12-23 02:33:01 <MC1984> im nure there are machine sold with a small boot SSD and a hdd too
162 2012-12-23 02:33:07 <MC1984> and hybrid drives
163 2012-12-23 02:33:42 <gmaxwell> MC1984: the hybrid drive stuff is just automagic in windows
164 2012-12-23 02:33:56 <gmaxwell> and should actually do sane things for bitcoin 0.8. (hopefully)
165 2012-12-23 02:34:21 <MC1984> well if it couldnt be done automatically, then something in the gui to choose directories for chain storage and active use would be nice
166 2012-12-23 02:34:27 <MC1984> and the wallet for that matter
167 2012-12-23 02:35:03 <gmaxwell> I expect we'll get some improved wallet management stuff into 0.8.
168 2012-12-23 02:35:22 <MC1984> cool
169 2012-12-23 08:52:30 <cal_> hello
170 2012-12-23 12:58:23 <rebroad> my debug.log seems to be currently filled with getblocks requests... it seems unusually frequent and asking for the same blocks...
171 2012-12-23 13:01:28 <rebroad> does bitcoind accept orphan blocks with work less than the work of the latest checkpoint block?
172 2012-12-23 13:54:21 <JWU42> mem usage is much improved with 0.7.2 - appreciate that guys!
173 2012-12-23 13:54:37 <JWU42> thanks for all your work !!
174 2012-12-23 14:00:30 <cjd> Hi guys, I'm doing some windows porting and I'm curious whether it's a safe bet that the value of HANDLE will never exceed INT32_MAX
175 2012-12-23 14:01:56 <cjd> I printed the values of a few and they were <100, it's hard to imagine the windows kernel having a handle table larger than 2 billion so is it just a uintptr to be annoying and incompatable?
176 2012-12-23 14:02:56 <cjd> int fd = (int) handle; assert(handle < INT32_MAX || !"oops lost the bet"); :)
177 2012-12-23 14:13:19 <cjd> ahh it looks like the OpenSSL guys took the bet for winsock handles http://stackoverflow.com/questions/1953639/is-it-safe-to-cast-socket-to-int-under-win64
178 2012-12-23 17:51:36 <rebroad> pretty quiet on here today...
179 2012-12-23 17:59:59 <etotheipi_> shhhh! I was enjoying the peacefulness
180 2012-12-23 18:06:21 <sipa> ACTION just drove 600km
181 2012-12-23 18:07:13 <gmaxwell> sipa: wow, are you turning into an american? :P
182 2012-12-23 18:21:19 <sipa> gmaxwell: no, in that case i would driven 400 miles of 600 km :p
183 2012-12-23 20:22:09 <pjorrit> so you include some extra distance because of the cheap gas?
184 2012-12-23 20:41:53 <gmaxwell> hmph. python-jsonrpc changes all the 'float' outputs to Decimal but doesn't handle them on the input side... you have to pass float.
185 2012-12-23 20:50:32 <zooko> gmaxwell: I have a horrible kludge for that...
186 2012-12-23 20:51:24 <zooko> Wait, you're saying json.dumps({"foo": Decimal('1.0')}) raises an exception?
187 2012-12-23 20:51:34 <gmaxwell> Yes.
188 2012-12-23 20:52:06 <gmaxwell> well not json.dumps (haven't tried), but calls to the ServiceProxy methods for the bitcoin API calls
189 2012-12-23 20:52:28 <zooko> If it has the standard lib's json module inside...
190 2012-12-23 20:52:45 <zooko> Then I think there is nowadays some combination of options you can pass to json to do what I want...
191 2012-12-23 20:52:47 <zooko> ACTION investigates
192 2012-12-23 20:53:11 <zooko> Excuse my latency -- I'm on the other side of a satellite from you...
193 2012-12-23 20:53:18 <gmaxwell> Jeff did something smart to it so that all the values coming out are Decimal.
194 2012-12-23 20:53:26 <gmaxwell> It just won't take them going back in.
195 2012-12-23 20:54:27 <zooko> Here's the horrible kludge I mentioned: https://pypi.python.org/pypi/jsonutil
196 2012-12-23 20:54:47 <zooko> I think that's doing the same thing that Jeff already did for python-jsonrpc.
197 2012-12-23 20:55:01 <zooko> I'm struggling to remember -- I think the other direction should be even easier...
198 2012-12-23 20:55:14 <zooko> I'm also struggling to load web pages over satellite in order to spur my memory.
199 2012-12-23 20:56:13 <zooko> I think use_decimal=True argument to json might suffice for that direction.
200 2012-12-23 20:57:39 <zooko> Strange, http://docs.python.org/2/library/json.html doesn't mention "use_decimal".
201 2012-12-23 20:57:44 <gmaxwell> E.g. In this script http://pastebin.com/phJcj7iB on line 28 the float() is required there or createrawtransaction throws an exception.
202 2012-12-23 20:58:22 <zooko> And that call to float() could change the value!
203 2012-12-23 20:58:27 <gmaxwell> Yes.
204 2012-12-23 20:59:01 <gmaxwell> It double checks it at the end by showing you the actual decimal computed fee but ugh.
205 2012-12-23 20:59:35 <gmaxwell> (my own code does all this with integers, but I wanted to move out gmaxwell-hack crap in order to publish some simple scripts for people)
206 2012-12-23 20:59:52 <zooko> Well, it looks like my solution in https://tahoe-lafs.org/trac/pyutil/browser/trunk/pyutil/jsonutil.py is to use simplejson instead of json (the Python standard library's version) and to pass "use_decimal=True".
207 2012-12-23 21:00:08 <zooko> That causes it to encode Decimal instances into decimal strings in json, which is obviously good and correct.
208 2012-12-23 21:00:41 <zooko> I can recommend doing whatever jsonutil.py does, or even using jsonutil.py itself, on the grounds that it has unit tests and benchmarks.
209 2012-12-23 21:04:37 <zooko> https://zooko.com/uri/URI:DIR2-RO:d73ap7mtjvv7y6qsmmwqwai4ii:tq5tqejzulg7yj4h7nxuurpiuuz5jsgvczmdamcalpk2rc6gmbsq/klog.html#pyutil.jsonutil
210 2012-12-23 22:52:54 <sipa> it's indeed quiet here...
211 2012-12-23 22:53:55 <phantomcircuit> sipa, turns out most bitcoiners aren't loser shutins
212 2012-12-23 22:54:03 <phantomcircuit> and actually have somewhere to be during the holidays
213 2012-12-23 22:54:06 <phantomcircuit> who would have thought
214 2012-12-23 22:54:21 <sipa> oh please, it's not even december 24th in the US yet!
215 2012-12-23 22:59:13 <phantomcircuit> sipa, today is travel day silly
216 2012-12-23 22:59:43 <phantomcircuit> alternatively everybody got raided by the CIA and are now in an undisclosed location
217 2012-12-23 22:59:49 <phantomcircuit> s/location/hole/
218 2012-12-23 23:00:25 <etotheipi_> sipa: have you built binaries for your 0.8-testing version?
219 2012-12-23 23:00:31 <etotheipi_> or whatever it is
220 2012-12-23 23:02:49 <phantomcircuit> water proofing my shoes and i just realized im super lightheaded
221 2012-12-23 23:03:05 <sipa> etotheipi_: they're not very current, but yes
222 2012-12-23 23:03:11 <midnightmagic> "too quiet"
223 2012-12-23 23:03:13 <sipa> i can build new ones, actually
224 2012-12-23 23:03:15 <etotheipi_> I just need anything that has the new blk file behavior
225 2012-12-23 23:03:25 <sipa> what OS?
226 2012-12-23 23:03:27 <etotheipi_> Armory is 100% broken with those builds
227 2012-12-23 23:03:32 <etotheipi_> I'm in Ubuntu 64-bit
228 2012-12-23 23:03:50 <sipa> can't you build them yourself?
229 2012-12-23 23:04:01 <etotheipi_> I never have before... how much effort is it?
230 2012-12-23 23:04:06 <maaku> phantomcircuit: I've got a spider-hole; they won't find me!
231 2012-12-23 23:04:31 <etotheipi_> since I've never done any dev on bitcoind/-qt I never had a reason to compile it
232 2012-12-23 23:05:31 <sipa> etotheipi_: if you just need bitcoind... libssl-dev, build-essential, libboost-all-dev, libdb5.1++-dev
233 2012-12-23 23:05:45 <t7> twas easy on debian64 so ubuntu should be the same
234 2012-12-23 23:05:57 <t7> oh not sipas :P
235 2012-12-23 23:06:32 <sipa> etotheipi_: and then cd src; make -f makefile.unix bitcoind
236 2012-12-23 23:07:56 <etotheipi_> sipa: where's the source?
237 2012-12-23 23:08:01 <etotheipi_> oh wait, I think I know
238 2012-12-23 23:08:37 <sipa> git clone git@github.com:bitcoin/bitcoin.git; cd bitcoin/src; make -f makefile.unix bitcoind
239 2012-12-23 23:10:28 <etotheipi_> wait, I don'tI assume there's a "git checkout turbo" in there
240 2012-12-23 23:10:36 <etotheipi_> whoops
241 2012-12-23 23:10:42 <sipa> in my repo there is
242 2012-12-23 23:10:45 <etotheipi_> ignore the first 11 chars
243 2012-12-23 23:10:57 <sipa> let me update my turbo branch
244 2012-12-23 23:12:23 <etotheipi_> libdb5.1++-dev?
245 2012-12-23 23:13:27 <etotheipi_> gah, not in the 10.04 repo
246 2012-12-23 23:13:35 <sipa> oh, 10.04
247 2012-12-23 23:13:44 <sipa> then you can just use libdb4.8++-dev
248 2012-12-23 23:14:37 <etotheipi_> ahh, that works
249 2012-12-23 23:15:30 <etotheipi_> thanks
250 2012-12-23 23:16:21 <sipa> hmmm, unit tests fail in turbo
251 2012-12-23 23:19:22 <etotheipi_> "net.cpp:1027: error: ???upnpDiscover??? was not declared in this scope"
252 2012-12-23 23:19:36 <etotheipi_> about a dozen more related errors
253 2012-12-23 23:19:48 <sipa> make -f makefile.unix bitcoind USE_UPNP=
254 2012-12-23 23:22:11 <sipa> etotheipi_: new turbo branch pushed
255 2012-12-23 23:22:38 <etotheipi_> how did I get conflicts ?!?
256 2012-12-23 23:22:51 <etotheipi_> does the build process change committed files?
257 2012-12-23 23:22:59 <sipa> it shouldn't
258 2012-12-23 23:23:10 <sipa> what command did you use that reported conflicts?
259 2012-12-23 23:23:25 <etotheipi_> just did a "git pull"
260 2012-12-23 23:23:38 <sipa> git pull tries to merge with the remote head
261 2012-12-23 23:23:55 <sipa> so depending on what branch you had checked out, that may very well result in conflicts
262 2012-12-23 23:24:02 <etotheipi_> why doesn't it do a fastforward?
263 2012-12-23 23:24:12 <sipa> because that is not what pull does
264 2012-12-23 23:24:19 <sipa> it merges your current branch with the remote head
265 2012-12-23 23:24:39 <sipa> if the remote head is a direct descendent from your current state, then that merge results in a fastforward
266 2012-12-23 23:24:45 <etotheipi_> I thought it does a fetch, and then merges that into your current state
267 2012-12-23 23:24:51 <sipa> yes, excactly
268 2012-12-23 23:25:07 <etotheipi_> right, if I didn't change my current state, why would it not be a fast-forward
269 2012-12-23 23:25:33 <sipa> because the new remote is not a descendent of the old remote
270 2012-12-23 23:25:54 <etotheipi_> okay, what should I do? I'll figure it out when you tell me that
271 2012-12-23 23:26:00 <stealth222> reset?
272 2012-12-23 23:26:04 <sipa> git reset --hard sipa/turbo
273 2012-12-23 23:26:06 <etotheipi_> I just did a hard reset to "leveldb17"
274 2012-12-23 23:26:29 <sipa> turbo is now hal+parallel+leveldb17+minor fixes
275 2012-12-23 23:27:14 <etotheipi_> so reset takes me to the "fetch"d state?
276 2012-12-23 23:27:18 <sipa> yes
277 2012-12-23 23:27:38 <etotheipi_> man, so much for thinking I understood git
278 2012-12-23 23:27:51 <sipa> "reset --hard X" is basically "forget what this branch pointed to before, just make it point to X. kthxbye"
279 2012-12-23 23:27:55 <etotheipi_> I guess I have a very narrow list of commands I use, and it's mostly just myself, so I don't get into trouble
280 2012-12-23 23:28:08 <stealth222> we all do, etotheipi...I think :)
281 2012-12-23 23:28:11 <stealth222> or most of us
282 2012-12-23 23:28:33 <sipa> yeah, you gradually expand the number of commands you use, i guess
283 2012-12-23 23:28:39 <stealth222> git is brilliant - but you really do need a bigger picture understanding of it to use it right
284 2012-12-23 23:28:44 <sipa> uhu
285 2012-12-23 23:28:50 <etotheipi_> I thought I did...
286 2012-12-23 23:28:55 <stealth222> I still don't use it right - lol
287 2012-12-23 23:29:00 <sipa> it has several high-level commands that try to be smart in many cases
288 2012-12-23 23:29:06 <sipa> being smart is nice as long as it works
289 2012-12-23 23:29:13 <etotheipi_> so how is what you pushed not a direct descendant of my branch?
290 2012-12-23 23:29:27 <etotheipi_> it's the same branch, right?
291 2012-12-23 23:29:31 <sipa> yes
292 2012-12-23 23:29:51 <sipa> turbo is created by a script that merges several pull requests into bitcoin/bitcoin master
293 2012-12-23 23:30:06 <sipa> and then overwrites the old turbo
294 2012-12-23 23:30:38 <etotheipi_> so... this is an abnormal condition?
295 2012-12-23 23:30:55 <sipa> unless it's only an extra commit that got added to the very last branch being merged, that will not result in something that is a descendant from a previous run
296 2012-12-23 23:31:06 <sipa> in general, it's branch that's continuously rebased
297 2012-12-23 23:31:13 <etotheipi_> oh, it rewrites the whole branch?
298 2012-12-23 23:31:17 <sipa> yes
299 2012-12-23 23:31:23 <etotheipi_> okay, so I'm not going crazy
300 2012-12-23 23:31:38 <sipa> which is something you shouldn't do for public branches
301 2012-12-23 23:31:46 <sipa> but i want a clean history :)
302 2012-12-23 23:31:53 <sipa> bitcoin/bitcoin is never rebased
303 2012-12-23 23:32:00 <etotheipi_> gotcha
304 2012-12-23 23:32:10 <etotheipi_> I really thought that everything I knew about git went out the window there...
305 2012-12-23 23:32:23 <etotheipi_> once the word "rebase" pops up, I know things are going to get silly
306 2012-12-23 23:32:26 <maaku> sipa: what do the numbers mean in dnsstats.log?
307 2012-12-23 23:33:34 <sipa> maaku: timestamp and some weighted average of the number of active nodes at that point, i think
308 2012-12-23 23:34:13 <maaku> thanks!
309 2012-12-23 23:34:22 <sipa> probably the sum over rows for the several columns in the dump file
310 2012-12-23 23:38:39 <etotheipi_> wtf... my debug.log is up to 6 GB
311 2012-12-23 23:39:07 <stealth222> is there any way to undo a push made to a pull request?
312 2012-12-23 23:39:18 <stealth222> I typed the wrong remote by accident
313 2012-12-23 23:41:55 <sipa> stealth222: no, but you have a local reflog that remembers old branche
314 2012-12-23 23:41:57 <sipa> s
315 2012-12-23 23:42:42 <stealth222> oh well - I guess the commit will appear in the repo
316 2012-12-23 23:43:37 <etotheipi_> sipa: does 0.8 leave the original blk files, and then just start adding more block data in blocks dir?
317 2012-12-23 23:43:38 <andytoshi> stealth222, you can repush the old changeset (using reflog to find it as sipa suggests)
318 2012-12-23 23:47:08 <stealth222> another thing - I'm trying to figure out exactly what the automatic sanity tester does so I can do it locally before pushing to a public repo. Is it pretty much exactly what I see in the test.log?
319 2012-12-23 23:47:52 <stealth222> and can I get that script so that I can run it locally?
320 2012-12-23 23:49:03 <zooko> gmaxwell: I would be interested in what you ultimately did about the json+Decimal issue.