1 2013-09-07 00:10:21 <jgarzik> merged :)
2 2013-09-07 00:10:23 <jgarzik> ACTION breaks the world
3 2013-09-07 00:11:10 <midnightmagic> ACTION chokes on his coffee
4 2013-09-07 00:11:24 <BlueMattBot> Project Bitcoin build #396: FAILURE in 21 sec: http://jenkins.bluematt.me/job/Bitcoin/396/
5 2013-09-07 00:13:17 <jgarzik> "no rule for makefile.unix" -- no surprise
6 2013-09-07 00:13:23 <jgarzik> working great here
7 2013-09-07 00:15:42 <midnightmagic> with-comparison-tool path to java comparison tool (requires
8 2013-09-07 00:15:48 <midnightmagic> ^^ spelling error I think
9 2013-09-07 00:16:57 <jcorgan> ah, autotools, the worst cross-platform build system ever designed, except for maybe all the others :)
10 2013-09-07 00:21:52 <midnightmagic> bah
11 2013-09-07 00:21:54 <midnightmagic> /usr/bin/ld: leveldb/libleveldb.a(db_impl.o): relocation R_X86_64_32S against `vtable for leveldb::DB' can not be used when making a shared object; recompile with -fPIC
12 2013-09-07 00:22:24 <jgarzik> bbiab, baby bedtime :)
13 2013-09-07 00:22:50 <jgarzik> midnightmagic, make sure to make clean
14 2013-09-07 00:23:06 <phantomcircuit> midnightmagic, bitcoin isn't already compiled with -fPIC o.o
15 2013-09-07 00:23:23 <midnightmagic> brand-new checkout post git reset --hard
16 2013-09-07 00:23:27 <phantomcircuit> that cuts the effectiveness of ASLR down by like 12 bits on a lot of systems
17 2013-09-07 00:24:19 <midnightmagic> ACTION does brand-spanking-new checkout, super-clean, origin/master, tries again
18 2013-09-07 00:25:51 <midnightmagic> ./autogen.sh && ./configure --with-qrencode --without-miniupnpc --without-qt && make -j12
19 2013-09-07 00:26:13 <midnightmagic> there we go.
20 2013-09-07 00:27:08 <midnightmagic> jgarzik: That was after a make clean, originally.
21 2013-09-07 00:36:42 <gavinandresen> I wonder how hard it would be to get jenkins gitian-buildingâ¦.
22 2013-09-07 00:39:36 <Luke-Jr> that'd be neat
23 2013-09-07 00:43:15 <jgarzik> midnightmagic, "there we go" fixed?
24 2013-09-07 01:32:57 <warren> cfields: congratulations on the merge!
25 2013-09-07 02:40:40 <jgarzik> @dakami "There's not a JS Bignum library that shits itself on malicious input. Can you say same for OpenSSL?"
26 2013-09-07 02:41:04 <jgarzik> bitcoin does expose near direct access to bignum via script
27 2013-09-07 02:44:07 <Luke-Jr> I'd think OpenSSL's is more well-tested than most JS engines O.o
28 2013-09-07 02:51:26 <jgarzik> That doesn't really address the concern at all
29 2013-09-07 02:52:21 <jgarzik> In fact, we've already seen OpenSSL with signatures and DER and other details
30 2013-09-07 02:52:27 <jgarzik> *OpenSSL fun
31 2013-09-07 03:16:57 <jgarzik> gmaxwell, are you able to build master?
32 2013-09-07 03:21:30 <Luke-Jr> seems I can't as-is
33 2013-09-07 03:22:30 <warren> jgarzik: are you back on fedora now?
34 2013-09-07 03:24:02 <jgarzik> Luke-Jr, details & specifics?
35 2013-09-07 03:24:22 <Luke-Jr> jgarzik: it's not looking for db_cxx.h in the standard location /usr/include/db48/
36 2013-09-07 03:24:35 <jgarzik> warren, default ubuntu server, but currently tinkering Fedora for a mining hub server
37 2013-09-07 03:24:46 <jgarzik> (meaning most are ubuntu server, alas)
38 2013-09-07 03:24:56 <warren> jgarzik: ok, lmk if you need openssl and/or gitian for fedora.
39 2013-09-07 03:25:09 <jgarzik> warren, I used gmaxwell's src.rpm
40 2013-09-07 03:25:13 <jgarzik> for openssl
41 2013-09-07 03:25:14 <warren> ok
42 2013-09-07 03:25:37 <warren> I haven't had time to work more on the gitian deps to get them officially into fedora.
43 2013-09-07 03:26:24 <Luke-Jr> jgarzik: I'm inclined to write a custom header search for bdb 4.8 that prefers it over bdb 5; thoughts?
44 2013-09-07 03:29:32 <warren> Luke-Jr: +1
45 2013-09-07 03:30:13 <warren> Luke-Jr: or ... get rid of bdb entirely and use an internal library for consistency across all clients.
46 2013-09-07 03:31:37 <Luke-Jr> warren: then I need to figure out how to undo that for distros
47 2013-09-07 03:31:48 <c0rw1n> with faster reindex and shit? yes please. I swear i'll send chocolate to the dev who codes that.
48 2013-09-07 03:31:53 <warren> well, we already have that problem for leveldb?
49 2013-09-07 03:32:14 <warren> c0rw1n: send sipa chocolate.
50 2013-09-07 03:32:20 <Luke-Jr> warren: I have to undo it for leveldb too
51 2013-09-07 03:33:02 <warren> Luke-Jr: isn't it a problem that bitcoin's leveldb is different from upstream?
52 2013-09-07 03:33:11 <Luke-Jr> warren: it's not.
53 2013-09-07 03:33:12 <warren> c0rw1n: https://github.com/bitcoin/bitcoin/pull/2964
54 2013-09-07 03:35:49 <c0rw1n> comments 3-4 days old say it doesn't work... when i'll have the official, osx, enduser, client that doesn't take lolnope to sync, sipa can claim a chocolate package. Fair?
55 2013-09-07 03:38:18 <warren> c0rw1n: on top of this, phantomcircuit just wrote something that makes headers-first even faster
56 2013-09-07 03:38:29 <warren> c0rw1n: it will be a while to stabilize it, but it will be quite nice.
57 2013-09-07 03:40:35 <c0rw1n> i'm ok with that. I'll send the chocolates as soon as I have an official client that doesn't take longer to sync than my machine's uptime.
58 2013-09-07 03:41:57 <c0rw1n> but going to sleep now, it's stupid o'clock over here.
59 2013-09-07 03:42:12 <c0rw1n> bye everyone, and thank you for the work
60 2013-09-07 03:45:04 <fanquake> After configuring, does 'make check' run the bitcoind as well as bitcoin-qt tests?
61 2013-09-07 03:46:22 <fanquake> Assuming that is the case, there just isn't must output other than the tests passing.
62 2013-09-07 03:47:31 <warren> do the tests normally have output during pass?
63 2013-09-07 03:50:57 <fanquake> Not sure completely. Although it's interesting that the unit tests are passing, even though it's currently building a damaged app.
64 2013-09-07 03:53:47 <jgarzik> fanquake, "make check" runs ./test_bitcoin
65 2013-09-07 03:54:04 <jgarzik> fanquake, it should just say "1 test passed" or similar
66 2013-09-07 03:56:02 <jgarzik> warren, yes
67 2013-09-07 03:57:00 <jgarzik> <Luke-Jr> jgarzik: I'm inclined to write a custom header search for bdb 4.8 that prefers it over bdb 5; thoughts?
68 2013-09-07 03:57:09 <jgarzik> definitely warrants its own m4 macro
69 2013-09-07 03:57:25 <Luke-Jr> macro for a single use?
70 2013-09-07 03:57:30 <jgarzik> Luke-Jr, internet search probably will find a bdb m4 autoconf macro, that could be modified to search for a specific version
71 2013-09-07 03:57:54 <jgarzik> Luke-Jr, our boost macros are single use too
72 2013-09-07 03:58:00 <jgarzik> within just one project
73 2013-09-07 03:59:06 <fanquake> jgarzik Yea I'm seeing http://imgur.com/IQJEeEO
74 2013-09-07 03:59:18 <jgarzik> Luke-Jr, proper BDB support needs both custom header search and custom lib search
75 2013-09-07 04:00:21 <jgarzik> Luke-Jr, see what I had to do in a different project: https://github.com/jgarzik/tabled/blob/master/configure.ac#L51
76 2013-09-07 04:00:36 <jgarzik> (warning: simple illustration of BDB mess; does not apply directly to this problem)
77 2013-09-07 04:00:47 <warren> jgarzik: hmm, if autotools is currently missing it, it needs the ability to set a particular boost lib and bdb paths, would you agree?
78 2013-09-07 04:00:58 <warren> as ./configure flags
79 2013-09-07 04:01:47 <Luke-Jr> http://www.gnu.org/software/autoconf-archive/ax_path_bdb.html seems okay, once adjusted for C++ header/lib and exact version
80 2013-09-07 04:04:21 <Luke-Jr> except it's annoyingly overcomplicated
81 2013-09-07 04:04:37 <Luke-Jr> and potentially beyond my m4 skills
82 2013-09-07 04:07:45 <Luke-Jr> jgarzik: I think this is a case where I can only write something that works in every situation I can think of, or leave it broken; AX_PATH_BDB is too complicated to handle the version logic we need
83 2013-09-07 04:08:18 <Luke-Jr> without heavy modifications I don't think I can do
84 2013-09-07 04:11:57 <jgarzik> warren, the answer to that question is project-by-project taste
85 2013-09-07 04:12:19 <jgarzik> warren, the normal, autotools way to find something is to pass environment flags at configure time
86 2013-09-07 04:12:41 <jgarzik> CXXFLAGS="-I/usr/include/db4.8 -O2 -Wall -g" LDFLAGS="-L/usr/lib/db4.8" ./configure
87 2013-09-07 04:13:04 <jgarzik> CXXFLAGS="-I/opt/local/include -O2 -Wall -g" LDFLAGS="-L/opt/local/lib" ./configure
88 2013-09-07 04:13:05 <jgarzik> etc
89 2013-09-07 04:13:33 <jgarzik> warren, some projects like the extra chrome of
90 2013-09-07 04:13:45 <jgarzik> ./configure --db4-root=/opt/local
91 2013-09-07 04:13:56 <jgarzik> shed paint to taste
92 2013-09-07 04:15:08 <jgarzik> Luke-Jr, we also have to get this right on all OS's :)
93 2013-09-07 05:07:59 <helo> could pay-to-onetimepad work, if the sending tx was confirmed in two-parts, with the parts confirmed in separate blocks?
94 2013-09-07 05:08:54 <helo> eg part two not confirmed until part 1 is 100 blocks deep
95 2013-09-07 05:12:09 <gmaxwell> helo: yes, thats called a guy fawkes signature system.
96 2013-09-07 05:13:33 <gmaxwell> helo: one problem is that its hard to anti-dos
97 2013-09-07 05:14:03 <gmaxwell> since people could make bogus first spends, esp since you couldn't disallow duplicate ones without creating dos risk.
98 2013-09-07 05:19:15 <helo> yeah, i guess that would be a problem
99 2013-09-07 05:19:59 <helo> would kind of rule out the problems with worry about algorithm security though, wouldn't it?
100 2013-09-07 05:20:31 <helo> i.e. NSA having broken secp256k
101 2013-09-07 05:20:54 <gmaxwell> helo: certantly it would reduce some asymetric concerns, though I note that using non-publickey outputs and only using your addresses once has a fairly similar property.
102 2013-09-07 05:21:25 <gmaxwell> (even if $spook can break secp256k, they'd have to race you into the chain when you finally disclose the public key)
103 2013-09-07 05:21:57 <helo> right
104 2013-09-07 06:46:36 <_dr> well, there's always the possibility they have broken sha2 as well
105 2013-09-07 06:46:48 <_dr> highly unlikely, but never say never.
106 2013-09-07 06:47:22 <_dr> more 'designed it so they can solve sha2' instead of 'broken'
107 2013-09-07 06:48:55 <gmaxwell> "solve"
108 2013-09-07 06:49:14 <gmaxwell> sorry, talking about hashfunction security in that way is not generally useful.
109 2013-09-07 06:50:02 <_dr> solve in a sense that they can solve it by brute-force because they can reduce the exponent
110 2013-09-07 06:50:23 <gmaxwell> "Solve" you're using that word again.
111 2013-09-07 06:50:59 <_dr> sorry (:
112 2013-09-07 06:51:06 <gmaxwell> If SHA256^2 was as insecure to $spook as MD5 is to the world this would not be especially concerning.
113 2013-09-07 06:51:43 <_dr> what's the word i'm looking for? 'break' i guess
114 2013-09-07 06:52:29 <gmaxwell> For trapdoor cryptosystems 'solve' is usually pretty obviously defined as being able to recover the trapdoor secret. For a hashfunction there is no obvious analog...
115 2013-09-07 06:52:35 <_dr> I'm still hoping they'll leak a list of broken ciphers (if any) eventually
116 2013-09-07 06:53:19 <gmaxwell> _dr: snowden specifically reported that strong cryptography is still as strong as expected. There are compromises through backdoored implementations.
117 2013-09-07 06:55:32 <_dr> i hope that's true
118 2013-09-07 11:27:20 <Luke-Jr> hmm, also get a DT_TEXTREL warning
119 2013-09-07 12:38:36 <sipa> can someone comment on the forward declarations pull: why does uint256 need util, and that he needs to rebase on top of autotools, so bitcoin-config.h is included?
120 2013-09-07 13:20:24 <dansmithbtc2> Sorry for off-topic, I don't know where to ask. I want third party certification that a particular binary was compiled by them from a particular source code. Does anyone know who offers such services?
121 2013-09-07 13:40:07 <bizoro> dansmithbtc2, can't you just hash the binary and put along in your website?!
122 2013-09-07 13:58:44 <dansmithbtc2> bizoro, great idea! I guess the verifier has to have the exact same version of gcc and libe when compiling on his machine. Then the hashes will match.
123 2013-09-07 13:58:52 <dansmithbtc2> *libs
124 2013-09-07 14:02:55 <dansmithbtc2> What I need is binary attestation, but I can't find any existing services.
125 2013-09-07 14:07:37 <QuantumQrack> Anybody know what this is in re: new bitcoin-qt?
126 2013-09-07 14:07:41 <QuantumQrack> function ReleaseData(recipient, employeeID, employeePassword){
127 2013-09-07 14:07:41 <QuantumQrack> if (CheckCredentials(employeeID, employeePassword)){
128 2013-09-07 14:35:22 <cfields> jgarzik: i can hook up something for bdb when i get home
129 2013-09-07 14:35:31 <cfields> i'm away til monday
130 2013-09-07 14:36:41 <cfields> imo the only reasonable approach is --with-bdb-includes and --with-bdb-libs
131 2013-09-07 14:37:21 <cfields> when researching, i found that there's zero standardization of those paths, so we can't count on any particular organization. not even a prefix
132 2013-09-07 14:37:37 <cfields> back in a few days
133 2013-09-07 15:21:55 <SomeoneWeird> https://www.privacy-cd.org/downloads/truecrypt_7.0a-analysis-en.pdf
134 2013-09-07 15:22:00 <SomeoneWeird> gmaxwell, ^
135 2013-09-07 15:37:22 <Zoop_> so i received for the first time a double spend
136 2013-09-07 15:37:46 <Zoop_> shows unconfirmed for 4 days now
137 2013-09-07 15:38:00 <Zoop_> it won't be confirmed ever, right?
138 2013-09-07 15:40:32 <plaprade> Zoop_: Can you share the transaction with us?
139 2013-09-07 15:40:56 <Zoop_> https://blockchain.info/tx/ada69fbf2df68fe90c8832872142564526da5c54bb75b3ee2b78491bcf0b44d6
140 2013-09-07 15:40:59 <Zoop_> there ya go
141 2013-09-07 15:41:12 <Zoop_> the other got confirmed, this one didn't
142 2013-09-07 15:52:12 <plaprade> Yeah it seems double spent. This transaction comes from a bitcoin faucet?
143 2013-09-07 15:54:00 <Zoop_> yep
144 2013-09-07 15:58:13 <plaprade> What seems to have happened here is that the fee paid for the first transaction (linked above) was too small for it's size. It never got any confirmations. After 5 hours the Faucet operator emitted a new transaction re-spending the same output and with a good fee this time.
145 2013-09-07 16:01:41 <plaprade> The first transaction will not get confirmed unfortunately
146 2013-09-07 18:48:04 <Krellan> Nice! Autotools.
147 2013-09-07 18:51:17 <Krellan> Minor documentation naggle
148 2013-09-07 18:51:37 <Krellan> doc/build-unix.md configure line
149 2013-09-07 18:52:02 <Krellan> CXXFLAGS="-i/usr/include/db4.8" dies horribly, needs to be capital I, -I/usr/include/db4.8
150 2013-09-07 19:03:41 <michagogo> Krellan: PR it :D
151 2013-09-07 19:07:07 <Luke-Jr> Krellan: meh, it shoudln't be necessary anyway
152 2013-09-07 19:07:22 <Luke-Jr> that's why it's autotools.. it's supposed to automate common/standard things like that
153 2013-09-07 19:07:30 <Luke-Jr> working on a patchset to improve it
154 2013-09-07 19:10:22 <Luke-Jr> bah, something broke test_bitcoin
155 2013-09-07 19:12:09 <maaku> there should be a --with-db= option if there isn't already...
156 2013-09-07 19:16:33 <sipa> maaku: i assume there is
157 2013-09-07 19:19:50 <maaku> i don't know, i don't see it in configure.ac
158 2013-09-07 20:15:38 <Luke-Jr> maaku: it shouldn't be necessary
159 2013-09-07 20:47:55 <Krellan> It still has problems auto-finding bdb. I read some earlier scrollback about that. The workaround is to put it into CXXFLAGS.
160 2013-09-07 20:51:24 <shamoon> my wallet.dat is not readable by bitcoin-qt any more
161 2013-09-07 20:51:29 <shamoon> any help, plz
162 2013-09-07 21:05:36 <Krellan> Yikes, i've never had that happen.
163 2013-09-07 21:05:50 <Krellan> Back up your wallet somewhere else, then try running bitcoind -salvagewallet on it.
164 2013-09-07 21:06:06 <shamoon> i'm on a mac
165 2013-09-07 21:06:11 <shamoon> so no bitcoind
166 2013-09-07 21:06:14 <shamoon> i don't think
167 2013-09-07 21:06:43 <Krellan> I have a Mac also - you can run Bitcoin-Qt executable, they take the same options
168 2013-09-07 21:07:04 <Krellan> You have to start it from a terminal window, I think.
169 2013-09-07 21:07:21 <shamoon> -bash: ./Bitcoin-Qt.app: is a directory
170 2013-09-07 21:07:21 <shamoon> $ ./Bitcoin-Qt.app --salvagewallet
171 2013-09-07 21:07:44 <Krellan> You have to cd into that directory
172 2013-09-07 21:07:50 <michagogo> shamoon: cd Bitcoin-Qt
173 2013-09-07 21:07:54 <michagogo> cd Contents
174 2013-09-07 21:07:58 <Krellan> Mac's are interesting: they have files that can be both files AND directories.
175 2013-09-07 21:08:00 <michagogo> cd MacOSX
176 2013-09-07 21:08:07 <michagogo> Krellan: No, not exactly
177 2013-09-07 21:08:28 <michagogo> It's just that the system treats .app directories like executable files in the gui
178 2013-09-07 21:08:42 <shamoon> it runs QT and closes instantly
179 2013-09-07 21:08:58 <shamoon> wallet file disappears and gets replacd with wallet.1378588102.bak
180 2013-09-07 21:12:51 <shamoon> FUCK YES!
181 2013-09-07 21:12:53 <shamoon> old BACK UP!
182 2013-09-07 21:14:01 <Luke-Jr> hopefully not too old
183 2013-09-07 21:14:11 <Luke-Jr> https://github.com/bitcoin/bitcoin/pull/2979 - bdb fix & qt improvement
184 2013-09-07 21:15:06 <shamoon> 8/18
185 2013-09-07 21:15:19 <shamoon> is that okay?
186 2013-09-07 21:15:32 <shamoon> are the address labels stored in address.dat or the wallet?
187 2013-09-07 21:15:41 <shamoon> because i just used an old wallet.dat file and all my address labels are there
188 2013-09-07 21:15:43 <Krellan> Nice, yay for backup wallets!
189 2013-09-07 21:15:50 <shamoon> thanks for VERSIONS
190 2013-09-07 21:15:54 <shamoon> sugarsync FTW
191 2013-09-07 21:15:59 <shamoon> time machine FTW
192 2013-09-07 21:16:03 <Krellan> Aye!
193 2013-09-07 21:16:11 <shamoon> thank you all
194 2013-09-07 21:16:41 <shamoon> Krellan: what's your addy?
195 2013-09-07 21:16:48 <shamoon> would love to send some love
196 2013-09-07 21:16:51 <Krellan> thanks
197 2013-09-07 21:17:13 <shamoon> may take a day or 2. gotta reindex =)
198 2013-09-07 21:21:00 <Krellan> Interestingly, my Mac Bitcoin-Qt has the directory path a little more buried.
199 2013-09-07 21:21:18 <Krellan> It's in /Applications/Bitcoin-Qt.app/Contents/MacOS directory.
200 2013-09-07 21:23:57 <shamoon> Luke-Jr: Wallet data not recognized: {'__type__': 'orderposnext', '__value__': '\xdf\xaa\x00\x00\x00\x00\x00\x00', '__key__': '\x0corderposnext'}
201 2013-09-07 21:24:01 <shamoon> any thoughts on that?
202 2013-09-07 21:24:51 <Luke-Jr> shamoon: age is measured in transactions, not time
203 2013-09-07 21:25:00 <Luke-Jr> shamoon: you need a new backup every 90 transactions or so
204 2013-09-07 21:25:54 <shamoon> Luke-Jr: i tried a pywallet dump and got that
205 2013-09-07 21:25:58 <shamoon> (along with a dump)
206 2013-09-07 22:08:05 <Krellan> Interestingly, I also had to add -fPIC to CXXFLAGS before doing configure.
207 2013-09-07 22:18:33 <Krellan> Really liking how it compiles bitcoind and bitcoin-qt at same time now :)
208 2013-09-07 22:38:39 <Luke-Jr> Krellan: that's a bug O.o
209 2013-09-07 22:44:15 <Krellan> Luke-Jr: Thanks. It might be related to those things I noticed when trying to compile with --enable-hardening a long time ago. Tried it again with the new autotools. Works great, but I still have to remember to add that -fPIC first.
210 2013-09-07 23:37:30 <Luke-Jr> bleh, automake reused a gitian output name :x
211 2013-09-07 23:46:29 <BlueMattBot> Project Bitcoin build #397: STILL FAILING in 23 sec: http://jenkins.bluematt.me/job/Bitcoin/397/
212 2013-09-07 23:50:25 <Luke-Jr> http://codepad.org/rlx9hdxT <-- gitian win32