1 2013-09-24 00:20:57 <michagogo> It's about time...
 2 2013-09-24 00:21:11 <michagogo> FINALLY I have everything set up in the new VM
 3 2013-09-24 00:23:45 <nkuttler> nice. now make a snapshot if you haven't used used something more sophisticated to configure it
 4 2013-09-24 00:28:45 <warren> cfields: around?
 5 2013-09-24 00:29:22 <warren> cfields: after your small fix to boost-win32.yml, ./configure gets stuck at the next problem
 6 2013-09-24 00:29:34 <warren> checking whether the Boost::Unit_Test_Framework library is available... yes
 7 2013-09-24 00:29:34 <warren> configure: error: No working boost sleep implementation found
 8 2013-09-24 00:30:48 <michagogo> nkuttler: I'll do that after I make-base-vm
 9 2013-09-24 00:31:03 <michagogo> Or should I do that first?
10 2013-09-24 00:34:01 <michagogo> I guess I may as well make one before and after
11 2013-09-24 00:34:46 <warren> wumpus: sipa: regarding the earlier discussion about gitian for linux builds, it appears there is a way to explicitly request a particular glibc version.  http://stackoverflow.com/questions/4032373/linking-against-an-old-version-of-libc-to-provide-greater-application-coverage
12 2013-09-24 00:38:05 <warren> Hmm it's also the libsdc++ ABI, not sure how often that changes.
13 2013-09-24 01:18:05 <dizko> ive had a persistent problem with tons of connections in TIME_WAIT using bitcoinrpc.   just upgraded to the newest 8.5 beta but no improvement
14 2013-09-24 01:18:15 <dizko> anyone seen this?
15 2013-09-24 01:57:14 <Luke-Jr> dizko: don't open that many connections
16 2013-09-24 01:57:18 <Luke-Jr> dizko: it's a TCP limitation
17 2013-09-24 01:58:58 <cfields> warren: need to see a log
18 2013-09-24 01:59:14 <longcat> 
ACTION poops

19 2013-09-24 02:06:30 <warren> cfields: https://togami.com/~warren/temp/config.log
20 2013-09-24 02:07:16 <michagogo> Okay, so I just gave up on trying to figure out how to tell make-base-vm to use an ISO
21 2013-09-24 02:07:28 <michagogo> After wasting...
22 2013-09-24 02:07:33 <cfields> warren: you have libssl linked in twice
23 2013-09-24 02:07:44 <michagogo> 2-3 hours...
24 2013-09-24 02:07:49 <michagogo> on that issue alone, I think
25 2013-09-24 02:08:36 <warren> cfields: ok, that's likely happening in the bitcoins-deps*.zip ... looking
26 2013-09-24 02:08:48 <cfields> warren: ah sorry, the similar name threw me.
27 2013-09-24 02:08:56 <cfields> tls = thread local storage in this case
28 2013-09-24 02:09:07 <warren> this works with 10.04 gitian?  I never tried.
29 2013-09-24 02:09:27 <warren> 10.04 gitian didn't need that ranlib name?
30 2013-09-24 02:10:11 <cfields> boost was able to work it out. but that problem's solved, unrelated to this one
31 2013-09-24 02:13:13 <cfields> warren: that was fixed in boost 1.5.1
32 2013-09-24 02:13:20 <cfields> *1.51
33 2013-09-24 02:13:51 <warren> hm
34 2013-09-24 02:14:20 <warren> cfields: this being for bitcoin 0.9, should we just upgrade gitian to the latest boost?
35 2013-09-24 02:14:51 <cfields> warren: seems the most reasonable would be to use whatever version precise ships
36 2013-09-24 02:15:04 <cfields> since that's what linux will be building against
37 2013-09-24 02:16:25 <warren> cfields: "linux" being what?
38 2013-09-24 02:16:47 <cfields> heh, that'd be a downgrade to 1.48, so i suppose that's out of the question
39 2013-09-24 02:17:21 <warren> cfields: gitian linux builds against whatever you define in gitian...
40 2013-09-24 02:17:40 <warren> f*ck it, trying latest boost
41 2013-09-24 02:17:45 <cfields> no, see packages. it uses system packages
42 2013-09-24 02:19:02 <warren> cfields: 1.48 is not acceptable
43 2013-09-24 02:19:45 <cfields> warren: that's what linux will be using
44 2013-09-24 02:20:31 <warren> 10.04 had what version?
45 2013-09-24 02:21:09 <cfields> 1.40
46 2013-09-24 02:21:49 <warren> win32 gitian was using 1.50 for a while now
47 2013-09-24 02:25:23 <warren> cfields: meanwhile, the actual devs aren't building and testing bitcoin against the gitian version of libraries at all
48 2013-09-24 02:26:39 <warren> Fedora has only 37 patches on boost 1.54.  Sounds great.
49 2013-09-24 02:27:08 <warren> 
ACTION tries it anyway

50 2013-09-24 02:28:35 <cfields> warren: i'm aware. I've voiced my concerns.
51 2013-09-24 02:29:25 <gmaxwell> warren: would you please tone down the fud?
52 2013-09-24 02:29:44 <gmaxwell> This is inaccurate. People test the gitan builds for during RC time.
53 2013-09-24 02:32:34 <warren> gmaxwell: fine, s/at all/much/
54 2013-09-24 02:34:43 <gmaxwell> The alternative is testing less on other things that people use too. Being bug dependant on single old versions of boost would not be a virtue.
55 2013-09-24 02:44:37 <warren> gmaxwell: wouldn't it be better to make gitian equivalent to the version most devs are using daily?
56 2013-09-24 02:45:00 <warren> gmaxwell: that's easier to control than cat herding
57 2013-09-24 02:47:29 <gmaxwell> There is no such equivalence possible. People are on different systems, and it can't be concurrently the same as all of them.  Frequently updating the underlying libraries would also reduce the security assurances gitian provides. And then there is the issue of wanting to have binary compatiblity with older systems, which necessitates building against an older libc.
58 2013-09-24 02:51:41 <michagogo> gaining the ability to gitian-build in an Ubuntu VM with LXC (so I can do it from within Windows) is taking... much, much longer than I expected
59 2013-09-24 02:52:30 <michagogo> Even ignoring the 2-3 hours that I spent trying to figure out if I could speed up the make-base-vm with the ubuntu ISOs...
60 2013-09-24 02:53:21 <michagogo> I started with this 6 hours ago :-/
61 2013-09-24 02:57:50 <warren> michagogo: 'speeding up' is irrelevant when apt-cacher-ng does its job.
62 2013-09-24 02:58:09 <michagogo> apt-cacher-ng?
63 2013-09-24 03:00:48 <michagogo> warren: ?
64 2013-09-24 03:00:50 <cfields> warren / sipa: https://github.com/sipa/secp256k1/pull/14
65 2013-09-24 03:01:45 <warren> cfields: it has a way to force in the case of cross compiling?
66 2013-09-24 03:02:13 <warren> cfields: and you tested it on mac?
67 2013-09-24 03:02:43 <cfields> warren: it's autotools, so cross works the same as everything else
68 2013-09-24 03:03:24 <warren> cfields: on mac it does -arch i386 -mmacosx_version=10.5 ?
69 2013-09-24 03:03:33 <warren> hmm, the latter isn't needed
70 2013-09-24 03:17:09 <warren> boost-1.54 has trouble building, looking at this later.
71 2013-09-24 04:27:39 <warren> They added a bug to 1.54 that assumes you will never disable zlib.
72 2013-09-24 06:01:12 <dizko> Luke-Jr: sorry for delay in responding, thanks for your input
73 2013-09-24 06:01:22 <dizko> Luke-Jr: is there a way to kee pthe connection open then?
74 2013-09-24 06:03:19 <dizko> Luke-Jr: also, is there a method to close the connection instead of letting it go into TIME_WAIT?   wouldnt it be better if someone sent a FIN
75 2013-09-24 07:42:58 <warren> found bug in boost, reporting upstream.  It was broken in April 2013
76 2013-09-24 07:43:56 <warren> https://svn.boost.org/trac/boost/ticket/9156
77 2013-09-24 08:52:37 <warren> grrr... more build regressions in boost-1.54
78 2013-09-24 08:52:51 <warren> apparently mingw is not something they test.
79 2013-09-24 08:54:47 <warren> ...failed gcc.compile.asm bin.v2/libs/context/build/gcc-mingw-4.6/release/link-static/target-os-windows/threadapi-win32/threading-multi/asm/jump_i386_sysv_elf_gas.o...
80 2013-09-24 08:54:56 <warren> I think it shouldn't be trying to build this on win32 ...
81 2013-09-24 09:19:07 <warren> cfields: would it be dangerous to insert a persistent partition into gitian VM's containing ccache data?
82 2013-09-24 09:19:19 <warren> 
ACTION is tired of waiting.

83 2013-09-24 09:19:19 <warren> wumpus: thoughts?
84 2013-09-24 09:30:52 <warren> hmm, ccache might just fail entirely with the faketime
85 2013-09-24 09:35:49 <warren> https://svn.boost.org/trac/boost/ticket/7262
86 2013-09-24 09:36:09 <warren> crap.  some boost dev refuses to add a patch that would allow cross-compile builds...
87 2013-09-24 10:20:37 <runeks> Getting the following error when trying to build bitcoind from master:
88 2013-09-24 10:20:39 <runeks> leveldb/libleveldb.a: could not read symbols: Bad value
89 2013-09-24 10:20:39 <runeks> /usr/bin/ld: leveldb/libleveldb.a(db_iter.o): relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC
90 2013-09-24 10:20:55 <runeks> On Ubuntu Raring, 64 bit.
91 2013-09-24 10:21:03 <sipa> runeks: clean your working directory
92 2013-09-24 10:21:06 <sipa> (just once)
93 2013-09-24 10:21:12 <runeks> Will do.
94 2013-09-24 10:27:11 <runeks> Woo! That fixed it. Thanks sipa.
95 2013-09-24 10:29:53 <runeks> Err. Now I'm getting the ": Corrupted block database detected." error. The bitcoind version from the Ubuntu PPA works fine...
96 2013-09-24 10:31:00 <runeks> 2013-09-24 10:30:11 ERROR: DisconnectBlock() : added transaction mismatch? database corrupted
97 2013-09-24 10:31:00 <runeks> 2013-09-24 10:30:15 ERROR: DisconnectBlock() : added transaction mismatch? database corrupted
98 2013-09-24 10:31:00 <runeks> 2013-09-24 10:30:16 ERROR: VerifyDB() : *** coin database inconsistencies found (last 242 blocks, 11047 good transactions before that)