1 2011-09-21 02:23:52 <SeniorBob> hey fellas
  2 2011-09-21 02:33:05 <SeniorBob> is it still worth investing in GPU rigs to mine?
  3 2011-09-21 02:33:18 <SeniorBob> even if i dont pay the power bill?
  4 2011-09-21 02:34:26 <gobiligo> wtf is it so hard to buy bitcoins :(
  5 2011-09-21 02:34:37 <gobiligo> im tired of readin wiki pages
  6 2011-09-21 02:34:51 <AAA_awright> gobiligo: Create an account on Mt Gox, follow directions to deposit
  7 2011-09-21 02:35:57 <gobiligo> lol, google link #2... MtGox Sucks - Learn Why You Should Not Use Mt Gox Bitcoin ...
  8 2011-09-21 02:36:22 <AAA_awright> Also, I think there's some serious oppertunities for arbitrage on there
  9 2011-09-21 02:36:47 <AAA_awright> gobiligo: Well out of the markets they're still the best, I trust it more than anything else
 10 2011-09-21 02:36:51 <gobiligo> its ironic, i work for commodities traders... the LAST thing i want to deal with is trading.....
 11 2011-09-21 02:38:12 <SeniorBob> ehh,,, only getting 750mhash
 12 2011-09-21 02:38:17 <SeniorBob> i miss being at home :/
 13 2011-09-21 02:38:19 <AAA_awright> Mining is probably going to over around the cheapest electric rate, there's people shutting off mining operations for costs that are too big now, we're seeing the difficulty fall
 14 2011-09-21 02:38:53 <SeniorBob> so GPU mining isnt worthless yet?
 15 2011-09-21 02:39:05 <SeniorBob> seems FPGA is the new big boy in town
 16 2011-09-21 02:40:00 <AAA_awright> SeniorBob: And an account with Dwolla which you should have anyways since PayPal is evil
 17 2011-09-21 02:40:20 <gobiligo> lol, i wanted to deposit with paypal
 18 2011-09-21 02:41:13 <SeniorBob> i used to mine at home, but built a new PC
 19 2011-09-21 02:41:15 <SeniorBob> and sadly
 20 2011-09-21 02:41:23 <SeniorBob> my wallet went with it :/
 21 2011-09-21 02:41:39 <SeniorBob> 12 coins floating in space now
 22 2011-09-21 02:41:41 <gobiligo> arrrrrgggggghhhhhhh, all i want is to buy some viagra from silk road... i hate the internetz
 23 2011-09-21 04:49:55 <osmosis> Why is it that 1 hour is enough to make reversal computationally impractical?
 24 2011-09-21 04:54:10 <ifungus> Hi. Do you know where gribble is on bitcoin-otc?
 25 2011-09-21 05:01:00 <neofutur> iffreenode is broken in at least 4 separated networks
 26 2011-09-21 05:01:08 <luke-jr> v4|5
 27 2011-09-21 05:01:20 <neofutur> ah ifungus is no more here
 28 2011-09-21 05:09:51 <mrb_> osmosis: see the bitcoin whitepaper for the math
 29 2011-09-21 05:12:42 <mrb_> 6 blocks reduce the probabily of outpacing the legit chain by an attacker having 10% of the computational power to less than 0.1%
 30 2011-09-21 07:12:32 <AlexWaters> anyone know how I can blockexplore test-net-in-a-box?
 31 2011-09-21 07:21:12 <sipa> is there an open-source blockexplorer-like system yet?
 32 2011-09-21 07:27:04 <AlexWaters> ;;seen bluematt
 33 2011-09-21 07:27:05 <gribble> bluematt was last seen in #bitcoin-dev 4 days, 18 hours, 21 minutes, and 16 seconds ago: <BlueMatt> jgarzik: ping
 34 2011-09-21 07:27:28 <doublec> sipa: like abe?
 35 2011-09-21 07:27:29 <AlexWaters> =/
 36 2011-09-21 07:27:45 <doublec> sipa: https://github.com/jtobey/bitcoin-abe
 37 2011-09-21 07:27:53 <AlexWaters> doublec: nice, thanks
 38 2011-09-21 07:27:58 <doublec> np
 39 2011-09-21 07:37:24 <diki> as much as i like windows, i dont like the fact that windows 8 will REQUIRE UEFI firmware
 40 2011-09-21 07:38:04 <diki> meaning those of us with the standard BIOS will not be able to install win8
 41 2011-09-21 07:40:22 <diki> AlexWaters:just so you know
 42 2011-09-21 07:40:25 <diki> importing takes days
 43 2011-09-21 07:40:28 <diki> at least it is for me
 44 2011-09-21 07:40:50 <diki> the first 115 or so thousand blocks import for an hour or two
 45 2011-09-21 07:40:54 <diki> after that, its slow as hell
 46 2011-09-21 07:47:21 <AlexWaters> diki: why do you think I will use windows 8? what kind of system are you running that it takes days?
 47 2011-09-21 07:50:42 <AlexWaters> diki: and, why wouldn't there be a work-around for not using uefi?
 48 2011-09-21 08:26:37 <diki> AlexWaters:this is not related to win8
 49 2011-09-21 08:26:43 <diki> i was just saying that abe takes a lot of time]
 50 2011-09-21 08:27:07 <diki> to import the blockchain into mysql
 51 2011-09-21 08:33:10 <AlexWaters> diki: ohhh, I thought you were talking about the bitcoin client
 52 2011-09-21 08:33:47 <AlexWaters> diki: I only need like 50 blocks to work with, or less even
 53 2011-09-21 08:33:53 <AlexWaters> so no big deal
 54 2011-09-21 08:43:23 <AlexWaters> diki: http://bit.ly/q3Iotj
 55 2011-09-21 09:14:17 <DeeTah> anyone here familiar with the uint256 class in Bitcoin sources?
 56 2011-09-21 09:16:58 <DeeTah> nvm, it was just a syntax error problem
 57 2011-09-21 09:25:40 <cjdelisle> I read it and it is awesome and full of win
 58 2011-09-21 09:26:49 <cjdelisle> it could be reimplemented with SSE2 __m128i types and it would be faster
 59 2011-09-21 09:27:38 <DeeTah> you mean uint256?
 60 2011-09-21 09:28:03 <cjdelisle> yes
 61 2011-09-21 09:28:18 <DeeTah> what exactly would it speed up?
 62 2011-09-21 09:28:52 <cjdelisle> it would be written using a pair of __m128i fields instead of 8 ints
 63 2011-09-21 09:29:06 <DeeTah> you mean 32.
 64 2011-09-21 09:29:15 <cjdelisle> err
 65 2011-09-21 09:29:15 <DeeTah> oh, 8 ints, 32 u8's
 66 2011-09-21 09:29:17 <DeeTah> my bad
 67 2011-09-21 09:29:27 <DeeTah> anyway, i mean, what in the client would gain more speed?
 68 2011-09-21 09:30:25 <cjdelisle> you'd have to look around at the operations on uint256s
 69 2011-09-21 09:30:45 <cjdelisle> maybe it would not be much but addition, shifting, multiplication of uint256s would be faster
 70 2011-09-21 10:44:55 <DeeTah> sipa: you here?
 71 2011-09-21 10:45:49 <DeeTah> or anyone else who recently explained me how does transaction signing work?
 72 2011-09-21 11:16:37 <DeeTah> anyone familiar with bitcoin cryptography here?
 73 2011-09-21 11:17:07 <helo> here, yes... awake, no :)
 74 2011-09-21 11:17:21 <DeeTah> i heard that transaction inputs are signed by ECDSA
 75 2011-09-21 11:17:38 <DeeTah> i wanted to see what data are sent to ECDSA_sign
 76 2011-09-21 11:17:51 <DeeTah> i found only one reference to this function in bitcoin sources and hacked it up like this:
 77 2011-09-21 11:18:19 <DeeTah> http://wklej.org/id/597409/
 78 2011-09-21 11:18:39 <DeeTah> then I recompiled and ran bitcoind on testnet and launched a sample transaction
 79 2011-09-21 11:18:45 <DeeTah> nothing was printed out, any ideas why?
 80 2011-09-21 11:19:31 <sipa> DeeTah: what did you do to invoke it?
 81 2011-09-21 11:19:44 <sipa> signing is only done to spend coins
 82 2011-09-21 11:20:03 <sipa> so you'd have to create a transaction to test it
 83 2011-09-21 11:20:07 <DeeTah> i did
 84 2011-09-21 11:20:16 <sipa> did you check debug.log?
 85 2011-09-21 11:20:20 <DeeTah> not really, thanks
 86 2011-09-21 11:20:20 <sipa> printf is redirected there
 87 2011-09-21 11:20:50 <DeeTah> nothing there
 88 2011-09-21 11:22:00 <DeeTah> got it
 89 2011-09-21 11:22:04 <sipa> you're printing the private key as a string?
 90 2011-09-21 11:22:05 <DeeTah> i read the wrong debug.log
 91 2011-09-21 11:22:09 <DeeTah> thanks sipa
 92 2011-09-21 11:22:17 <DeeTah> honestly, i had no idea how to print it properly
 93 2011-09-21 11:22:29 <sipa> you can try HexStr
 94 2011-09-21 11:22:33 <sipa> it's in util.h
 95 2011-09-21 11:23:03 <DeeTah> HexStr(pkey)?
 96 2011-09-21 11:27:11 <sipa> pkey is a blackbox type, so you can't inspect it directly
 97 2011-09-21 11:27:34 <sipa> you can extract a serialized private key, or a serialized public key, or just the secret parameter
 98 2011-09-21 11:27:49 <DeeTah> how do I?
 99 2011-09-21 11:27:57 <sipa> well, what do you want?
100 2011-09-21 11:28:08 <DeeTah> i want something real and reproductible
101 2011-09-21 11:28:25 <sipa> what for?
102 2011-09-21 11:28:36 <DeeTah> i mean, i want to see what is signed so I could build my own app using ECDSA_sign and get the same effects
103 2011-09-21 11:28:51 <DeeTah> then I could test AVR ECDSA code I got from someone42
104 2011-09-21 11:29:04 <DeeTah> and see if I can get it to have the same results on Arduino
105 2011-09-21 11:29:20 <sipa> try .GetSecret()
106 2011-09-21 11:29:28 <DeeTah> i thought it's a C type
107 2011-09-21 11:29:29 <DeeTah> ok
108 2011-09-21 11:29:31 <sipa> it'll give you a 32-byte secret parameter
109 2011-09-21 11:29:37 <sipa> in a CSecret
110 2011-09-21 11:29:55 <DeeTah> like pkey.GetSecret()
111 2011-09-21 11:30:00 <DeeTah> ?
112 2011-09-21 11:30:04 <sipa> no
113 2011-09-21 11:30:08 <sipa> just GetSecret()
114 2011-09-21 11:30:24 <DeeTah> like printf("%d", GetSecret()); ?
115 2011-09-21 11:30:38 <sipa> CSecret vchSecret = GetSecret();
116 2011-09-21 11:30:54 <sipa> printf("%s", HexStr(vchSecret.begin(), vchSecret.end()));
117 2011-09-21 11:31:07 <DeeTah> ok, thanks
118 2011-09-21 11:31:18 <sipa> printf("%s", HexStr(vchSecret.begin(), vchSecret.end()).c_str());
119 2011-09-21 11:31:19 <sipa> actually
120 2011-09-21 11:31:21 <DeeTah> is it the private key thingy, this secret?
121 2011-09-21 11:32:07 <DeeTah> key.h:143: error: CSecret was not declared in this scope
122 2011-09-21 11:32:11 <sipa> eh
123 2011-09-21 11:32:14 <DeeTah> perhaps you're talking of 0.4?
124 2011-09-21 11:32:36 <sipa> i'm talking about git head
125 2011-09-21 11:32:40 <DeeTah> sigh
126 2011-09-21 11:32:46 <DeeTah> which sounds like i'll need more hacking
127 2011-09-21 11:32:47 <sipa> what version are you using?
128 2011-09-21 11:32:55 <DeeTah> current stable, 0.3.24
129 2011-09-21 11:33:45 <sipa> hmm, i believed that was merged there already
130 2011-09-21 11:34:17 <DeeTah> will I be able to just copy the class?
131 2011-09-21 11:34:24 <sipa> just use 0.4
132 2011-09-21 11:34:41 <DeeTah> will its bitcoind compile flawlessly?
133 2011-09-21 11:34:53 <DeeTah> i mean, no new dependencies?
134 2011-09-21 11:34:56 <sipa> if 0.3.24 compiles on your system, 0.4 will too
135 2011-09-21 11:35:07 <DeeTah> ok
136 2011-09-21 11:35:56 <DeeTah> will it mess up my db?
137 2011-09-21 11:36:01 <DeeTah> or 100% compatible?
138 2011-09-21 11:36:53 <sipa> at least forward compatible
139 2011-09-21 11:37:29 <DeeTah> i'll backup the block chain just in case.
140 2011-09-21 11:37:50 <DeeTah> can I expect the output to be the same if it's all pulled from the same txin?
141 2011-09-21 11:38:31 <sipa> hmm?
142 2011-09-21 11:38:50 <DeeTah> the results of my printfs
143 2011-09-21 11:39:02 <sipa> yes
144 2011-09-21 11:39:04 <da2ce7> SomeoneWeird, there is a great series of js videos up on channel 9
145 2011-09-21 11:39:07 <DeeTah> is it possible for them to repeat when I call the same transaction twice, pulling the coins from the same input?
146 2011-09-21 11:39:10 <da2ce7> *finds linky*
147 2011-09-21 11:39:16 <DeeTah> /usr/include/miniupnpc/miniupnpc.h: In function void ThreadMapPort2(void*):
148 2011-09-21 11:39:20 <DeeTah> /usr/include/miniupnpc/miniupnpc.h:45: error: too many arguments to function UPNPDev* upnpDiscover(int, const char*, const char*, int)
149 2011-09-21 11:39:23 <DeeTah> net.cpp:1087: error: at this point in file
150 2011-09-21 11:39:26 <DeeTah> /usr/include/miniupnpc/upnpcommands.h:123: error: too many arguments to function int UPNP_AddPortMapping(const char*, const char*, const char*, const char*, const char*, const char*, const char*, const char*)
151 2011-09-21 11:39:35 <sipa> oh yes, miniupnpc was updated, i believe
152 2011-09-21 11:39:42 <sipa> if you don't need it, just disable it
153 2011-09-21 11:39:50 <sipa> USE_UPNP= on the make command line
154 2011-09-21 11:39:57 <DeeTah> -DUSE_UPNP=0 is set
155 2011-09-21 11:40:13 <sipa> make it =
156 2011-09-21 11:40:14 <sipa> not =0
157 2011-09-21 11:40:16 <DeeTah> kay
158 2011-09-21 11:40:22 <DeeTah> it's a bit confusing ;p
159 2011-09-21 11:40:23 <sipa> =0 means compile in but disabled by default
160 2011-09-21 11:40:27 <sipa> = means don't compile
161 2011-09-21 11:40:30 <DeeTah> kay
162 2011-09-21 11:40:36 <sipa> it's confusing indeed
163 2011-09-21 11:40:46 <DeeTah> good to have you around then
164 2011-09-21 11:41:16 <DeeTah> i hope the data will be enough to test ECDSA_sign
165 2011-09-21 11:41:28 <DeeTah> i wonder if the code I got works the same way
166 2011-09-21 11:41:49 <sipa> i expect it'll take some work
167 2011-09-21 11:41:59 <sipa> encodings differ between several implementations
168 2011-09-21 11:42:15 <DeeTah> work is not a problem, math is
169 2011-09-21 11:42:17 <sipa> you need to set the same curve (secp256k1)
170 2011-09-21 11:42:27 <DeeTah> so it's not a standard?
171 2011-09-21 11:42:32 <sipa> it is
172 2011-09-21 11:42:42 <sipa> the standard is called secp256k1 :)
173 2011-09-21 11:43:06 <DeeTah> oh
174 2011-09-21 11:43:08 <DeeTah> lame of me :d
175 2011-09-21 11:43:32 <sipa> you could also manually set the field prime and power, group order, curve parameters, ...
176 2011-09-21 11:44:20 <DeeTah> you seem to know it all very well
177 2011-09-21 11:44:54 <sipa> it took some time :D
178 2011-09-21 11:46:22 <DeeTah> okay
179 2011-09-21 11:46:24 <DeeTah> i think i've managed to get the same private key twice
180 2011-09-21 11:46:40 <DeeTah> now I'd love to get the same message twice
181 2011-09-21 11:47:10 <DeeTah> got it :D
182 2011-09-21 11:47:37 <DeeTah> i guess that means I was using the same input with the same privkey twice?
183 2011-09-21 11:48:42 <sipa> note that an ECDSA signature involves a random number
184 2011-09-21 11:49:01 <sipa> so a second signature with the same key on the same data won't be the same
185 2011-09-21 11:49:09 <Diablo-D3> unless you're sony
186 2011-09-21 11:49:23 <DeeTah> so how did I get the same result twice?
187 2011-09-21 11:49:36 <DeeTah> i guess the hash variable stores the signature, right?
188 2011-09-21 11:49:38 <Diablo-D3> DeeTah: you didnt change the random number
189 2011-09-21 11:49:45 <sipa> he did
190 2011-09-21 11:49:51 <sipa> ECDSA_sign does that internally
191 2011-09-21 11:50:13 <sipa> DeeTah: the hash variable is the data being signed
192 2011-09-21 11:50:16 <DeeTah> and when does it change the random number?
193 2011-09-21 11:50:26 <sipa> it doesn't "change" it
194 2011-09-21 11:50:35 <sipa> it just generates a new one every time you call it
195 2011-09-21 11:51:00 <DeeTah> sigh
196 2011-09-21 11:51:03 <DeeTah> that's messed up
197 2011-09-21 11:51:08 <DeeTah> what is it even used for?
198 2011-09-21 11:51:14 <sipa> what?
199 2011-09-21 11:51:33 <DeeTah> i mean, why would signatures need random numbers if they already utilize unknown private keys?
200 2011-09-21 11:51:54 <Diablo-D3> because thats how it works dude
201 2011-09-21 11:51:56 <sipa> if you have two signatures with the same key and the same random parameter, you can derive the private key from the signature
202 2011-09-21 11:52:07 <sipa> that's what ECDSA's security is based on
203 2011-09-21 11:52:09 <Diablo-D3> ECDSA is a black box, you dont NEED to know how it works
204 2011-09-21 11:52:20 <Diablo-D3> if you actually want to, the wikipedia page describes it well
205 2011-09-21 11:52:24 <sipa> you're right
206 2011-09-21 11:52:41 <sipa> don't try to delve into the details of crypto algorithms unless you know what you do
207 2011-09-21 11:53:07 <DeeTah> i don't exactly :p
208 2011-09-21 11:53:12 <Diablo-D3> heh
209 2011-09-21 11:53:21 <sipa> so if it uses a random number, you can be pretty well sure there is a reason for :)
210 2011-09-21 11:53:22 <Diablo-D3> all public/private key crypto works the same
211 2011-09-21 11:53:32 <sipa> rsa signing doesn't require a random number
212 2011-09-21 11:54:01 <DeeTah> before bitcoin i only heard of rsa when it came to public/private key
213 2011-09-21 11:54:04 <DeeTah> so, ok.
214 2011-09-21 11:54:13 <DeeTah> that means i won't be able to reproduce it all
215 2011-09-21 11:54:24 <sipa> you might try to verify it
216 2011-09-21 11:54:27 <DeeTah> can I somehow force ECDSA_sign to use a specific random seed?
217 2011-09-21 11:54:31 <Diablo-D3> DeeTah: reproduce what?
218 2011-09-21 11:54:39 <Diablo-D3> you're trying to write a ECDSA miner?
219 2011-09-21 11:54:44 <DeeTah> nah
220 2011-09-21 11:54:56 <Diablo-D3> because I've almost been considering that
221 2011-09-21 11:55:12 <sipa> DeeTah: if you're able to correctly verify the signatures bitcoin creates, you probably have a working EC stack implemented
222 2011-09-21 11:55:17 <DeeTah> i'm trying to move the signing part to an external device
223 2011-09-21 11:55:27 <Diablo-D3> DeeTah: dont
224 2011-09-21 11:55:33 <Diablo-D3> because it takes zero computation time to do it
225 2011-09-21 11:55:47 <DeeTah> Diablo-D3: not for performance
226 2011-09-21 11:55:49 <Diablo-D3> seriously, I can verify the entire chain in bitcoin in like 2-3 minutes
227 2011-09-21 11:55:58 <sipa> DeeTah: that would surprise me
228 2011-09-21 11:56:03 <sipa> eh, Diablo-D3
229 2011-09-21 11:56:24 <Diablo-D3> sipa: well, I assume bitcoind internally sucks at it
230 2011-09-21 11:56:37 <sipa> that may be a correct assumption
231 2011-09-21 11:56:45 <Diablo-D3> because it chews cpu time like a motherfucker when you're downloading the chain the first time
232 2011-09-21 11:56:51 <Diablo-D3> and all of its being soaked up by the crypto shit
233 2011-09-21 11:56:54 <sipa> but still: you need like 1ms for a signature verification
234 2011-09-21 11:56:55 <Diablo-D3> so something isnt right somewhere
235 2011-09-21 11:57:20 <Diablo-D3> although it'd be lol to opencl check it
236 2011-09-21 11:57:21 <DeeTah> so, the thing is to verify if the ECDSA_signature is correct, because it's the only way to reproduce stuff?
237 2011-09-21 11:57:54 <sipa> DeeTah: even better would be you generating signatures, and hacking bitcoind to verify them
238 2011-09-21 11:58:09 <Diablo-D3> I wonder why there isnt a universal version of bitcoin yet
239 2011-09-21 11:58:11 <Diablo-D3> it'd be useful
240 2011-09-21 11:58:14 <DeeTah> Diablo-D3: not that bad idea. bitcoin works extremely slowly on my laptop, i think there would be some point doing similar stuff
241 2011-09-21 11:58:26 <sipa> "universion version" ?
242 2011-09-21 11:59:20 <helo> the current version is just the earthly version
243 2011-09-21 11:59:31 <Diablo-D3> a framework just to handle network communications and tx handling
244 2011-09-21 11:59:34 <Diablo-D3> none of the btc shit
245 2011-09-21 11:59:58 <DeeTah> what new data would I need to verify a signature?
246 2011-09-21 12:00:18 <helo> Diablo-D3: shared mining will make that pretty awesome
247 2011-09-21 12:00:20 <sipa> you need a public key, data, and a signature :)
248 2011-09-21 12:01:49 <Diablo-D3> helo: shared mining is kinda pointless imo
249 2011-09-21 12:01:54 <DeeTah> can I get the pubkey from privkey?
250 2011-09-21 12:03:42 <DeeTah> (i wonder how would I get random data from Arduino)
251 2011-09-21 12:04:47 <sipa> DeeTah: yes
252 2011-09-21 12:05:15 <DeeTah> sipa: mind helping me?
253 2011-09-21 12:05:37 <DeeTah> i'd PM you the hashes and you'd help me code a program using ECDSA_verify
254 2011-09-21 12:06:03 <DeeTah> it's testnet, but i wouldn't want the coins to be stolen anyway
255 2011-09-21 12:06:17 <sipa> i can answer questions, but you'll have to do the coding :)
256 2011-09-21 12:06:45 <DeeTah> okay
257 2011-09-21 12:07:04 <DeeTah> i'll do some testing first
258 2011-09-21 12:10:13 <DeeTah> btw, why is even any code apart from the headers in key.h?
259 2011-09-21 12:16:02 <DeeTah> i wonder how to dump pchSig
260 2011-09-21 12:16:55 <sipa> vchSig?
261 2011-09-21 12:17:07 <sipa> HexStr(vchSig.begin(), vchSig.end())
262 2011-09-21 12:17:25 <DeeTah> that could work ;)
263 2011-09-21 12:17:48 <DeeTah> it's a pity i have to rebuild everything everytime
264 2011-09-21 12:29:47 <DeeTah> hey, sipa, wtf
265 2011-09-21 12:29:50 <DeeTah> i got the same pchsig twice
266 2011-09-21 12:29:57 <DeeTah> oh
267 2011-09-21 12:30:01 <DeeTah> not exactly the same, nvm
268 2011-09-21 12:30:28 <DeeTah> just the first five bytes
269 2011-09-21 12:31:06 <DeeTah> i wonder why anyway :P
270 2011-09-21 12:31:23 <sipa> the DER encoding used for signatures isn't very efficient
271 2011-09-21 12:31:34 <DeeTah> what do you men?
272 2011-09-21 12:31:36 <DeeTah> mean*?
273 2011-09-21 12:31:36 <sipa> i guess the first few bytes are always the same
274 2011-09-21 12:32:01 <edcba> so ?
275 2011-09-21 12:32:53 <DeeTah> yeah, the five bytes are always the same.
276 2011-09-21 12:36:43 <sipa> that's not too strange; DER encoded signatures are 72 bytes, but there are only 64 significant bytes in there
277 2011-09-21 13:20:26 <graingert> what version of TLS does bitcoind support?
278 2011-09-21 13:20:43 <sipa> for RPC?
279 2011-09-21 13:29:05 <edcba> suspicious question
280 2011-09-21 13:29:55 <graingert> yeah DW it was my mistake
281 2011-09-21 13:30:20 <graingert> the sploit in TLSv1 only works if you have JS injected in the target anyway
282 2011-09-21 13:51:47 <CIA-101> DiabloMiner: Patrick McFarland master * r6f4053e / src/main/java/com/diablominer/DiabloMiner/DiabloMiner.java : Fix crash in arg parsing - http://git.io/e3-8xg
283 2011-09-21 15:18:21 <b4epoche> I need a catchy title for this TEDx talk on Bitcoin I'm giving&  any suggestions?
284 2011-09-21 15:19:27 <AlexWaters> Bitcoin the musical
285 2011-09-21 15:19:40 <AlexWaters> lol, sorry
286 2011-09-21 15:20:41 <AlexWaters> Bitcoin: the premiere crypto-currency
287 2011-09-21 15:23:04 <da2ce7> Bitcoin and Buttcoin: the if's the butt's and the whatevers.
288 2011-09-21 15:23:42 <da2ce7> Bitcoin: naked finance.
289 2011-09-21 15:25:22 <da2ce7> hello Diablo-D3 :)
290 2011-09-21 15:25:53 <Diablo-D3> hello
291 2011-09-21 15:26:07 <da2ce7> what ya up2 today?
292 2011-09-21 15:27:01 <b4epoche> my best:  Bitcoin - Money to the People
293 2011-09-21 15:27:01 <Diablo-D3> trying to get consolekit to work properly under slim
294 2011-09-21 15:28:36 <gmaxwell> b4epoche: "Bitcoin: Platonic money"
295 2011-09-21 15:30:31 <b4epoche> hmm&  I like it
296 2011-09-21 15:32:00 <gmaxwell> (There is a good argument that bitcoin is 'pure' in a way that nothing has been before other than its use as money bitcoin is useless, but it has the important properties of money limited supply, fluidity, fungability, non-forgablity, etc more strongly than anything else we use as money)
297 2011-09-21 15:33:22 <gmaxwell> This article makes an argument supporting this position: http://www.libertariannews.org/2011/06/21/against-the-gold-standard/
298 2011-09-21 15:33:30 <b4epoche> I like it more
299 2011-09-21 15:35:57 <gmaxwell> b4epoche: and that also gives you a framework to talk about the specifics of the bitcoin system, e.g. how those properties arise.
300 2011-09-21 15:36:30 <gmaxwell> (also it tells the audience why it should care about money right up front)
301 2011-09-21 15:36:45 <gmaxwell> s/money/bitcoin/ heh
302 2011-09-21 15:37:37 <b4epoche> yea, I think I'm going to start the talk with a little discussion of 'money'
303 2011-09-21 15:37:53 <helo> yeah... discussions of bitcoin i've started tend to end because nobody cares about it
304 2011-09-21 15:38:57 <b4epoche> it takes a bit of a captive audience because most people will say, "What's the problem with the current system?"
305 2011-09-21 15:39:37 <gmaxwell> Yea, a key idea in giving a talk is that you need to create a connection between the audience and the subject.
306 2011-09-21 15:40:50 <gmaxwell> b4epoche: you can at least make the point that if nothing else undestanding bitcoin gives you a new perspective on the subject as money.  We take it for granted, having another kind of money available gives you a microscope to understand everything else.
307 2011-09-21 15:41:09 <b4epoche> and I like starting with 'money' in general because people think they understand it but really don't (or haven't thought much about it)
308 2011-09-21 15:41:29 <gmaxwell> Right. Money needs to have all these properties in order to work, and people never think about them.
309 2011-09-21 15:41:59 <b4epoche> in fact, that's a lot of the reason I like bitcoin.  It's forced me to think about some things.
310 2011-09-21 15:42:02 <gmaxwell> It's more than "money can be exchange for goods and services". Why should people trade you goods and services for inadequate toilet paper?
311 2011-09-21 15:42:06 <gmaxwell> yep.
312 2011-09-21 15:42:50 <nhodges> someone needs to convince lowtax (somethingawful admin/owner) to accept bitcoin for registration
313 2011-09-21 15:42:54 <nhodges> THAT would be funny
314 2011-09-21 15:51:33 <k9quaint> money should have pictures of nekkid women on it
315 2011-09-21 15:52:19 <ThomasV_> kekkid?
316 2011-09-21 15:52:50 <b4epoche> quit f'ing 'round copumpkin ;-)
317 2011-09-21 15:53:23 <b4epoche> and I'm sure tons have people have tried to figure out how to combine money with porn
318 2011-09-21 15:53:23 <gribbles> am I not mature and educated enough for you, b4epoche ?
319 2011-09-21 15:53:36 <b4epoche> where are you getting this?
320 2011-09-21 15:54:04 <b4epoche> did you get kicked out of the top-secret channel?
321 2011-09-21 15:54:16 <gribbles> nah, it's just invite-only and I have no way to get back in
322 2011-09-21 15:55:03 <helo> use up/down motion to generate current to power a miner
323 2011-09-21 17:52:49 <Optimo> rasengan: I run the bigboss repo :) everything looks okay but it's not every day that we get a non-cocoa app submitted. if you've done the appropriate testing on a jailbroken device it will probably be okay - and I will test it for myself prior to publication
324 2011-09-21 17:53:47 <b4epoche> Optimo:  wrong channel
325 2011-09-21 17:53:56 <b4epoche> I think
326 2011-09-21 17:53:56 <Optimo> why?
327 2011-09-21 17:54:00 <Optimo> nope
328 2011-09-21 17:54:00 <rasengan> Optimo:  Thank you so much, sir!
329 2011-09-21 17:54:08 <copumpkin> :o
330 2011-09-21 17:54:12 <Optimo> any reason you chose to use Appcelrator insted of something more native on teh whole?
331 2011-09-21 17:54:20 <rasengan> Optimo: Cross platform
332 2011-09-21 17:54:30 <Optimo> oh sure. of course
333 2011-09-21 17:54:50 <b4epoche> what's rasengan up to?
334 2011-09-21 17:55:02 <Optimo> a nice mtgox app, from the looks of it
335 2011-09-21 17:55:27 <b4epoche> web app?
336 2011-09-21 17:55:40 <b4epoche> not sure what you mean by non-cocoa
337 2011-09-21 17:55:46 <rasengan> www.appcelerator.com i think
338 2011-09-21 17:56:14 <b4epoche> ah, I see
339 2011-09-21 17:56:37 <Optimo> technically primarily it is cocoa, but not entirely through and through :)
340 2011-09-21 17:56:57 <Optimo> rasengan: cool I shot you a quick reply too
341 2011-09-21 17:57:36 <rasengan> Optimo: Cool, thank you :)
342 2011-09-21 17:58:06 <EskimoBob> Hi, do bitcoin-qt devs hang around here?
343 2011-09-21 17:58:26 <sipa> wumpus is bitcoin-qt's developer
344 2011-09-21 17:59:27 <Optimo> I suppose this chat is really for dev chat regarding the development of the bitcoin protocol and mainline app, so forgive me if I sputed off int eh wrong place
345 2011-09-21 17:59:37 <EskimoBob> wumpus: any idea what "i686-pc-linux-gnu/bin/ld: cannot find -lboost_s-1_42ystem " error means?
346 2011-09-21 17:59:39 <Optimo> probably should have mentioned it in #mtgoxlive or #mtgox
347 2011-09-21 17:59:54 <rasengan> :D
348 2011-09-21 18:00:24 <sipa> EskimoBob: it means you don't have a file libboost_s-1_42ystem.a
349 2011-09-21 18:00:33 <sipa> in you LD_LIBRARY_PATH
350 2011-09-21 18:02:13 <wumpus> probably means that your boost libraries have another name
351 2011-09-21 18:02:46 <wumpus> it sucks that boost has no pkgconfig support so we can only guess at the library names...
352 2011-09-21 18:03:43 <wumpus> especially as they embed version numbers and other funny stuff
353 2011-09-21 18:10:35 <EskimoBob> sipa: I have Boost Libraries for C++ installed
354 2011-09-21 18:10:46 <EskimoBob> version 1.42.0-r2
355 2011-09-21 18:11:01 <sipa> i assume you made a typo when copy-pasting to IRC?
356 2011-09-21 18:11:10 <sipa> there seems to be an 's' missing
357 2011-09-21 18:12:17 <EskimoBob> ?
358 2011-09-21 18:12:38 <tcatm> can bitcoin re-create blkindex.dat from blk0001.dat?
359 2011-09-21 18:13:21 <sipa> i don't think that's implementede
360 2011-09-21 18:13:25 <sipa> but it should be possible
361 2011-09-21 18:13:32 <wumpus> -lboost_s-1_42ystem  <- boost has funny library names, but I don't think that's right
362 2011-09-21 18:13:45 <sipa> EskimoBob: 1_42system instead of 1_42ystem
363 2011-09-21 18:14:01 <wumpus> did you change the bitcoin-qt.pro?
364 2011-09-21 18:14:30 <wumpus> on unix it links to  unix:!macx:LIBS += -lboost_system -lboost_filesystem -lboost_program_options -lboost_thread
365 2011-09-21 18:14:37 <EskimoBob> lboost_s-1_42ystem  < this is a copy / paste from error message
366 2011-09-21 18:14:37 <wumpus> change that line if the libraries are named differently
367 2011-09-21 18:15:15 <wumpus> it seems you added -1_42 in the middle of the lib name instead of at the end? or someone else did
368 2011-09-21 18:15:55 <EskimoBob> "/usr/lib/gcc/i686-pc-linux-gnu/4.4.5/../../../../i686-pc-linux-gnu/bin/ld: cannot find -lboost_s-1_42ystem"
369 2011-09-21 18:16:06 <EskimoBob> that is how it looks
370 2011-09-21 18:17:02 <wumpus> that's why I asked whether you changed the bitcoin-qt.pro file
371 2011-09-21 18:17:29 <wumpus> unchanged it links against -lboost-system, no 1_42 inbetween
372 2011-09-21 18:18:27 <EskimoBob> Makefile:LIBS          = $(SUBLIBS)  -L/usr/lib/qt4 -lssl -lcrypto -ldb_cxx -lboost_s-1_42ystem <---!
373 2011-09-21 18:18:57 <wumpus> yes...
374 2011-09-21 18:19:03 <EskimoBob> i have not changed anything. Looks like the error is right there
375 2011-09-21 18:19:08 <sipa> EskimoBob: where did yoi get the source?
376 2011-09-21 18:19:11 <sipa> you
377 2011-09-21 18:19:22 <wumpus> EskimoBob: is it also in bitcoin-qt.pro ?
378 2011-09-21 18:19:36 <wumpus> the makefile is generated from that by qmake
379 2011-09-21 18:19:59 <EskimoBob> git git
380 2011-09-21 18:20:50 <wumpus> here it looks like LIBS          = $(SUBLIBS)  -L/usr/lib -lssl -lcrypto -ldb_cxx -lminiupnpc -lboost_system -lboost_filesystem...
381 2011-09-21 18:22:17 <EskimoBob> wumpus: looks OK in bitcoin-qt.pro
382 2011-09-21 18:22:28 <wumpus> try re-running qmake
383 2011-09-21 18:22:47 <wumpus> you are on linux right?
384 2011-09-21 18:23:26 <EskimoBob> yes
385 2011-09-21 18:24:21 <wumpus> ok, and after rerunning qmake the makefile is still messed up?
386 2011-09-21 18:41:11 <EskimoBob> wumpus: sorry, I was afk for a moment . Makefile looks fine to me
387 2011-09-21 18:41:37 <EskimoBob> is it possible, that bitcoin-qt-9999.ebuild has errors ?
388 2011-09-21 18:42:08 <wumpus> I guess so
389 2011-09-21 18:42:18 <EskimoBob> http://pastebin.com/s1cRq2p3
390 2011-09-21 18:42:22 <wumpus> could be it does a find/replace :p
391 2011-09-21 18:42:48 <wumpus> and does it wrongly
392 2011-09-21 18:42:49 <wumpus> dunno
393 2011-09-21 18:48:17 <wumpus> yes it seems to do some sed thing...
394 2011-09-21 19:26:15 <tcatm> I think bitcoins' "load block index" could be improved. It shouldn't take > 30s when my simple python script can re-construct the index from blk0001.dat within 9 seconds (without verifying transactions)
395 2011-09-21 19:27:01 <Disposition> compare said script with bitcoin codebase and find bottle neck?
396 2011-09-21 19:27:38 <tcatm> well, the bitcoin codebase already lodas the pre-build index
397 2011-09-21 19:27:43 <tcatm> loads*
398 2011-09-21 20:27:55 <CIA-101> bitcoinjs/node-bitcoin-p2p: Stefan Thomas master * r027d3a0 / wscript : Better error message if pkg-config is missing. - http://git.io/9y2fjw
399 2011-09-21 20:38:20 <makomk> gmaxwell: bitcoin may be "pure" in the sense that it can't be used as anything except money, but it requires a vast amount of "impure" resources that could be used for other things to keep it running...
400 2011-09-21 20:39:25 <gmaxwell> makomk: so does everything else. Or do you think that magical farries handle, print, count, and transport dollar bills?
401 2011-09-21 20:40:44 <makomk> True, but right now the costs are quite large compared to dollar bills...
402 2011-09-21 20:41:55 <imsaguy> really?
403 2011-09-21 20:42:00 <imsaguy> transporting USD are cheap?
404 2011-09-21 20:42:04 <imsaguy> is*
405 2011-09-21 20:42:10 <imsaguy> how about securing USD?
406 2011-09-21 20:42:58 <makomk> imsaguy: I'm not sure how the cost of securing USD compares to the cost of securing Bitcoins; no-one's achieved the latter yet.
407 2011-09-21 20:43:49 <sipa> how do you mean?
408 2011-09-21 20:44:52 <makomk> Is there a major Bitcoin service that hasn't been robbed and had major security holes?
409 2011-09-21 20:45:04 <imsaguy> thats like saying banks get robbed
410 2011-09-21 20:45:11 <imsaguy> but securing the USD itself
411 2011-09-21 20:45:17 <imsaguy> all the technology and research that goes into it
412 2011-09-21 20:45:20 <donpdonp> makomk: lolz. hacking makes the exchange stronger!
413 2011-09-21 20:46:10 <makomk> imsaguy: is amortized over all the dollars in circulation.
414 2011-09-21 20:46:24 <sipa> makomk: i think storing bitcoins safely is almost trivial
415 2011-09-21 20:46:49 <sipa> the problem is providing safe webservices to interact with it
416 2011-09-21 20:47:30 <sipa> but really, send you coins to an address generated with an offline computer, encrypt the wallet with a strong password, and backup it on a few places
417 2011-09-21 20:47:55 <sipa> there are few safer stores of value around (assuming - and that's a big if of course - bitcoin retains it value)
418 2011-09-21 20:57:13 <luke-jr> makomk: Bitcoin costs a lot less resources to "run" than USD
419 2011-09-21 20:58:13 <sipa> bitcoin's economy is also massively smaller, so it is hard to compare
420 2011-09-21 20:58:51 <gmaxwell> makomk: it's easy to find examples of ecommerce sites that have been robbed, thats the same either way.
421 2011-09-21 21:00:22 <gmaxwell> The anti-counterfeiting costs for paper money are enormous.  Bitcoin's core security costs (e.g. the blockchain difficulty, not the computer security stuff which is the same for everything) are relatively high when bitcoin is small but scale better.
422 2011-09-21 21:02:59 <luke-jr> probably fair to say Bitcoin's current security level is acceptable for a global scale
423 2011-09-21 21:03:18 <luke-jr> unless maybe a big government were to attack it directly
424 2011-09-21 21:06:32 <gmaxwell> I think so too. Thats what I mean by scales better. The whole system needs to be strong enough but thats it. It's O(1)ish, while paper money has a significant O(N) component (enforcement and physical anti-counterfeiting measures)&
425 2011-09-21 21:24:17 <midnightmagic> it is not acceptable for a global scale, unless you are defining "global scale" as its current use.
426 2011-09-21 21:26:29 <luke-jr> actually, I suppose it's an arms race
427 2011-09-21 21:26:49 <luke-jr> for both Bitcoin (hashpower) and USD (defeating anti-counterfieting)
428 2011-09-21 21:27:04 <luke-jr> USD has "merely" progressed further
429 2011-09-21 21:28:51 <gmaxwell> luke-jr: The fundimental difference is that all hashpower put into bitcoin increases global security. Only the tech in dollars has conservation and thats only R&D conservation, special inks/special printing/etc cost per bill and feet on the street doing enforcement in florida doesn't prevent counterfieting in califonia.
430 2011-09-21 21:29:11 <luke-jr> hm
431 2011-09-21 21:32:43 <midnightmagic> it's only increased its security if you assume the security of the software contributing to the blockchain is not an actual risk.
432 2011-09-21 21:33:29 <gmaxwell> midnightmagic: true, but effort to improve that is also well conserved.
433 2011-09-21 21:34:39 <midnightmagic> imo, full-on high-level software engineering practices should probably be adopted the moment developer resources allow for it. I just don't think enough resources are being contributed to the development effort..
434 2011-09-21 21:35:06 <Namegduf> Ultimately the only safety IMO is through heterogenous systems
435 2011-09-21 21:35:09 <gmaxwell> Not perfectly, because hopefully people run software from multiple sources.... But it's still not the kind of effort fragmentation that you have with paper-money anti-counterfieting.
436 2011-09-21 21:35:10 <midnightmagic> .. to make it feasible to put full unit-testing, continuous integration, issue-tracking, and design phases..
437 2011-09-21 21:35:45 <gmaxwell> Sure, formal methods too.
438 2011-09-21 21:37:09 <midnightmagic> I can see Gavin making inroads on it, but.. it's a terrible task to tack it on after the fact to an already-existing project. I don't envy him!
439 2011-09-21 21:37:22 <gmaxwell> It's still strongly conserved resources expenditures.   One issue is that a lot of bitcoin operators are pretending that this is not an immature system.
440 2011-09-21 21:38:01 <midnightmagic> yes, there's way too much PR going on about it. I can see Gavin's statements months ago in my head clear as day, "This is experimental software," and so on.
441 2011-09-21 21:39:09 <gmaxwell> We should probably try courting academics doing work on formal methods and try to convince them to work on bitcoin. ... academics in other areas have had no luck getting grants for work on bitcoin, but the formals methods stuff may have better luck.
442 2011-09-21 21:39:26 <midnightmagic> any corporate software engineers could do it too.
443 2011-09-21 21:39:55 <genjix> do what?
444 2011-09-21 21:40:13 <midnightmagic> bring formal methods to the development of bitcoin.
445 2011-09-21 21:40:39 <CIA-101> libbitcoin: genjix * rf06a6f3e50fe / (5 files in 4 dirs): Delete branch method - only took a week to get right :/
446 2011-09-21 21:41:02 <gmaxwell> Well, industrial practices generally suck except in a few specific domains.
447 2011-09-21 21:41:05 <genjix> https://en.bitcoin.it/wiki/Bitcoin_Request_for_Comments
448 2011-09-21 21:41:12 <genjix> gmaxwell: au contraire!
449 2011-09-21 21:41:12 <midnightmagic> unit tests, continuous integration, tofu-scale development lines, proper issue tracking, and design phase documentation.
450 2011-09-21 21:41:20 <genjix> academic methods suck
451 2011-09-21 21:41:36 <gmaxwell> E.g. if you can get people that work on embedded controllers for flight criticial machines and life support equipment to work on bitcoin you might actually make some major improvements.
452 2011-09-21 21:41:41 <genjix> seen some of the spaghetti patchwork from academia?
453 2011-09-21 21:41:48 <midnightmagic> genjix: nah, software engineers are being properly trained now who could do all this.
454 2011-09-21 21:41:58 <gmaxwell> But the defect density in most commercial software would leave us dead if bitcoin was that bad.
455 2011-09-21 21:42:18 <gmaxwell> I see both genjix and midnightmagic don't actually know what I'm talking about when I said formal methods.
456 2011-09-21 21:42:27 <midnightmagic> That IMO is an indication of the current quality of seng practices.. that it isn't worse than that considering the economic pressures on a corporate team.
457 2011-09-21 21:42:40 <gmaxwell> https://secure.wikimedia.org/wikipedia/en/wiki/Formal_methods
458 2011-09-21 21:43:03 <midnightmagic> you're talking provably-correct software.
459 2011-09-21 21:43:19 <midnightmagic> I'm talking formal seng methods..
460 2011-09-21 21:43:41 <TuxBlackEdo> Does anyone have some TBCs?
461 2011-09-21 21:43:46 <gmaxwell> (which are not widely used in industry except in a few areas, e.g. aeronautics and of course, HDL 'software' intended for production)
462 2011-09-21 21:43:48 <midnightmagic> time-base-correctors?
463 2011-09-21 21:43:53 <genjix> gmaxwell: we're making an opensource project (think svn, apache, linux, firefox...)
464 2011-09-21 21:44:02 <TuxBlackEdo> well.. i was hoping test net bitcoins
465 2011-09-21 21:45:05 <gmaxwell> genjix: those are horrible comparison points, severe bugs in those platforms don't have the same consequence that bugs in bitcoin could have.
466 2011-09-21 21:45:26 <gmaxwell> Bitcoin's risk factors are a superset of those.
467 2011-09-21 21:45:40 <midnightmagic> formal methods as you describe them are uneconomic. :)
468 2011-09-21 21:46:29 <gmaxwell> midnightmagic: indeed, they're justified in some contexts, for example in bitcoin at least for the software doing blockchain validation.
469 2011-09-21 21:47:26 <midnightmagic> :-)
470 2011-09-21 21:48:51 <TuxBlackEdo> nobody has testnet bitcoins....?
471 2011-09-21 21:49:07 <midnightmagic> I have a bunch of them but I'm not running a testnet btc instance right now so I can't give any to you.
472 2011-09-21 21:49:18 <TuxBlackEdo> oh ok nvm
473 2011-09-21 21:49:29 <midnightmagic> I have a pile of namecoin testnet coins you can have.
474 2011-09-21 21:50:46 <gmaxwell> Unfortunately the blockchain validation is so complicated that I don't think it can be validated except by proving it in coq and then using extraction to generate the software. Not really something we'd want in the main client though... but someone could be running such a thing as an alert service.
475 2011-09-21 21:51:13 <midnightmagic> coq?
476 2011-09-21 21:51:32 <gmaxwell> https://secure.wikimedia.org/wikipedia/en/wiki/Coq
477 2011-09-21 21:51:33 <vsrinivas> gmaxwell: that bad?
478 2011-09-21 21:53:07 <gmaxwell> Well, maybe I just suck, but I've had bad luck e.g. proving a range coder correct from C code. Simple things are simple, but hard things get really hard to prove from procedural code very fast.
479 2011-09-21 21:53:36 <midnightmagic> hrm..
480 2011-09-21 21:53:43 <midnightmagic> well let's get on it! :-) lol
481 2011-09-21 21:55:26 <gmaxwell> (but people have written formally proven C compilers with coq+extraction, so thats evidence that complicated systems can be done that way)
482 2011-09-21 21:57:17 <copumpkin> I love doing that
483 2011-09-21 21:57:28 <copumpkin> if you have questions about how it works, I'd be happy to evangelize
484 2011-09-21 21:58:10 <sipa> gmaxwell: roconnor had blockchain validation written in haskell
485 2011-09-21 21:58:21 <midnightmagic> i love haskell
486 2011-09-21 21:58:25 <copumpkin> high five
487 2011-09-21 21:58:31 <midnightmagic> o !
488 2011-09-21 21:58:50 <sipa> that's already a much better candidate for proving
489 2011-09-21 21:58:51 <sipa> !hi65
490 2011-09-21 21:58:56 <copumpkin> :)
491 2011-09-21 21:59:02 <sipa> eh
492 2011-09-21 21:59:07 <sipa> !hi5
493 2011-09-21 22:01:27 <copumpkin> I'd imagine that most of the scary sort of bugs would be eliminated by using haskell for the client
494 2011-09-21 22:01:37 <copumpkin> not sure what kinds of things we'd want to prove about the code beyond that
495 2011-09-21 22:17:37 <doublec> Something like ATS could allow you to slowly move parts of the code into something that has proof checks
496 2011-09-21 22:17:46 <doublec> since it's fairly easy to mix C and ATS code
497 2011-09-21 22:21:49 <copumpkin> oh fuck haskell
498 2011-09-21 22:21:55 <copumpkin> agda 4eva
499 2011-09-21 22:21:57 <copumpkin> or as some say
500 2011-09-21 22:22:00 <copumpkin> agda SSSSZ eva
501 2011-09-21 22:22:39 <doublec> hehe
502 2011-09-21 22:23:13 <copumpkin> but if you mix C code into your ATS, you can't really be very sure that you're free of buffer overflows and the like, right?
503 2011-09-21 22:23:27 <copumpkin> it's not going to go and run a static analysis pass over your C is it? :P
504 2011-09-21 22:25:02 <doublec> right, it's a gradual process of ATS-ifying things
505 2011-09-21 22:25:18 <doublec> but you do provide annotations to C functions so you can be sure you're not passing bad data
506 2011-09-21 22:25:36 <doublec> and that you don't use pointers returned by the C function incorrectly, etc
507 2011-09-21 22:25:45 <gmaxwell> Right, you can make promises about the C code, and then make sure they're true with asserts.
508 2011-09-21 22:25:51 <doublec> yes
509 2011-09-21 22:26:35 <doublec> you can just write the core important stuff in ATS (verification, coin handling, etc)