1 2015-09-11 00:42:16 <alpalp> cfields: reusing libconsensus could work, but dont see it happening.  Other implementations do exist though
  2 2015-09-11 00:43:20 <cfields> alpalp: the question was about this particular implementation. of course there are others :)
  3 2015-09-11 00:43:48 <alpalp> well if you switch to another language, it pretty much is another implementation, so why bother?
  4 2015-09-11 01:05:09 <extor> any testnet support channels on freenode?
  5 2015-09-11 01:39:05 <Luke-Jr> extor: here?
  6 2015-09-11 01:50:34 <jgarzik> phantomcircuit,
  7 2015-09-11 01:50:39 <jgarzik> Arrive in Montreal tomorrow night
  8 2015-09-11 02:19:10 <Luke-Jr> williamcotton: no development decisions should ever be based on the current state of that
  9 2015-09-11 02:24:21 <extor> Luke-Jr, yes this is freenode
 10 2015-09-11 03:15:18 <gmaxwell> If you're dust-b-goning known-to-the-public private keys, you should use my insecure signer for considerably smaller and more compressible signatures, perhaps?
 11 2015-09-11 03:15:45 <gmaxwell> they're ~10 bytes smaller IIRC, and if you also use the sighash single bug they're all identical in the transaction.
 12 2015-09-11 04:08:31 <Luke-Jr> gmaxwell: is that public now?
 13 2015-09-11 04:27:21 <Luke-Jr> BlueMatt: I think cfields is somehow confused thinking you're trying to build with Intel's own compiler
 14 2015-09-11 04:27:33 <cfields> eh?
 15 2015-09-11 04:27:46 <Luke-Jr> well, what is ICE? :/
 16 2015-09-11 04:28:28 <cfields> internal compiler error. I guess I've spent too much time working with busted gcc versions :)
 17 2015-09-11 04:29:04 <Luke-Jr> ah; but that's just an OOM - hardly G++'s fault
 18 2015-09-11 04:29:34 <Luke-Jr> BlueMatt: btw, you could of course disable tests.. not that I think we should encourage that :P
 19 2015-09-11 04:29:54 <cfields> Luke-Jr: sure it is, if gcc sucks up way more mem than necessary
 20 2015-09-11 04:30:02 <cfields> also, ICE != OOM
 21 2015-09-11 04:30:10 <Luke-Jr> Killed == OOM
 22 2015-09-11 04:30:49 <cfields> right. ICE is more like a BUG_ON or assert
 23 2015-09-11 04:30:51 <mr_burdell> https://bitcoin.stackexchange.com/questions/38513/the-shortest-ecdsa-signature
 24 2015-09-11 04:31:02 <cfields> i'd guess he's building with gcc 4.6/4.7 or so
 25 2015-09-11 04:31:05 <mr_burdell> is there a better way to get a k value to produce a shorter sig?
 26 2015-09-11 04:31:15 <mr_burdell> or just using the value in that example is good?
 27 2015-09-11 04:32:02 <Luke-Jr> cfields: the error BlueMatt posted is what G++ does when OOM occurs.
 28 2015-09-11 04:36:54 <cfields> Luke-Jr: yes i see now, cc1 was indeed killed. I still contend that it's gcc's problem, though
 29 2015-09-11 04:37:36 <Luke-Jr> cfields: GCC intentionally leaks memory unless you tell it not to. ;)
 30 2015-09-11 04:39:00 <cfields> Luke-Jr: heh, sadly, i'm not sure if you're joking or if there's really some obscure flag i don't know about :)
 31 2015-09-11 04:39:11 <Luke-Jr> cfields: those flags I posted to the bug ;)
 32 2015-09-11 04:42:55 <cfields> heh
 33 2015-09-11 05:21:44 <Arnavion> It's desirable behavior for a short-lived process that needs to work quickly to let the OS clean its memory up
 34 2015-09-11 05:21:54 <Arnavion> IIRC the canonical D compiler also does it
 35 2015-09-11 07:17:21 <dcousens> Arnavion: aye it does
 36 2015-09-11 08:21:04 <paveljanik> jonasschnelli, Hi, do you have a minute (with main network bitcoind running)?
 37 2015-09-11 08:22:54 <paveljanik> jonasschnelli, Try running bitcoin-cli getrawmempool true and immediately after entering this command press Ctrl+C. Repeat it several times. My bitcoind sometimes exits 8) Can you repeat it?
 38 2015-09-11 08:23:26 <extor> What's a decent windows client that can run testnet? Evidently bitcoin-qt cannot
 39 2015-09-11 08:24:51 <paveljanik> extor, ?
 40 2015-09-11 08:25:01 <paveljanik> it can, of course
 41 2015-09-11 08:25:21 <paveljanik> why do you think it cannot?
 42 2015-09-11 08:25:37 <extor> because I see no option to switch
 43 2015-09-11 08:25:46 <extor> I dont even see an ini file
 44 2015-09-11 08:26:11 <extor> I'd like to zap my 40 gig blockchain too and run the smaller testnet one and start playing and testing
 45 2015-09-11 08:26:49 <paveljanik> http://bitcoin.stackexchange.com/questions/21149/cannot-get-bitcoin-qt-to-run-in-testnet-mode
 46 2015-09-11 08:27:18 <extor> so I created bitcoin.conf
 47 2015-09-11 08:27:36 <extor> So this is what software has come to in 2015. You have to create config files out of thin air.
 48 2015-09-11 08:29:09 <paveljanik> no, you can use and editor to do so...
 49 2015-09-11 08:30:12 <extor> I dont have a bitcon directory in Roaming which is good I suppose. I must have wisely refused that option during install.
 50 2015-09-11 08:33:44 <extor> seems to be working, it says no block source and five hours behind
 51 2015-09-11 08:35:09 <extor> Although it is giving me my old "real" btc balances
 52 2015-09-11 08:37:10 <extor> ok bitcoin.conf did not work but adding the -testnet switch did work
 53 2015-09-11 08:40:32 <drazisil> i was not aware that testnet was expected to work from .conf, the instructions do say to use the switch.
 54 2015-09-11 08:42:19 <paveljanik> jonasschnelli, bitcoind exits with code 141, ie. SIGPIPE...
 55 2015-09-11 08:42:48 <extor> the page I was linked to said conf would work
 56 2015-09-11 08:42:51 <paveljanik> anyone else be able to reproduce it?
 57 2015-09-11 08:43:10 <paveljanik> (on current master)
 58 2015-09-11 08:44:25 <extor> How can one test on testnet when it takes 3 hours to confirm a transaction lol
 59 2015-09-11 08:47:03 <drazisil> paveljanik: why are you trying to ctrl-c getrawmempool?
 60 2015-09-11 08:47:29 <paveljanik> drazisil, well, doing so on the mainnet right now takes a lot of time, you know? ;-)
 61 2015-09-11 08:47:57 <paveljanik> so it is quite common to stop the running task with ctrl+c.
 62 2015-09-11 08:48:04 <paveljanik> and you do not expect the server to stop 8)
 63 2015-09-11 08:48:52 <paveljanik> Imagine stopping the Firefox with ctrl+c while browsing google.com 8)
 64 2015-09-11 08:49:00 <drazisil> do you do it more then once, or it crashes after just once?
 65 2015-09-11 08:49:27 <paveljanik> drazisil, ctrl+c imediatelly stops bitcoin-cli. And sometimes, it also stops bitcoind...
 66 2015-09-11 08:49:50 <paveljanik> I expect that new libevent based server side is culprit here\
 67 2015-09-11 08:50:23 <drazisil> i have not run into that, what version are you running, and did you build or binaries?
 68 2015-09-11 08:51:05 <paveljanik> current master, self built
 69 2015-09-11 08:51:36 <paveljanik> I'm able to reproduce ~1-in-20 tries
 70 2015-09-11 08:51:38 <paveljanik> mac os x
 71 2015-09-11 08:52:32 <paveljanik> it looks easier once mempool gets filled with the spam transactions again.
 72 2015-09-11 08:52:42 <drazisil> ifaik, the devs dont do support in this channel, i would post as much info as you can in an issues. make sure you check your versions against the requested ones, like libdb and such.
 73 2015-09-11 08:53:37 <paveljanik> drazisil, sure. I'm looking for confirmation from other dev so I can PR the fix...
 74 2015-09-11 08:53:57 <drazisil> you have a fix?
 75 2015-09-11 08:54:13 <paveljanik> not yet, but will find it ;-0
 76 2015-09-11 08:54:49 <paveljanik> as it is sigpipe ;-)
 77 2015-09-11 08:54:50 <drazisil> make the issue, then if you find a fix you can always make and link the PR
 78 2015-09-11 08:56:16 <drazisil> i think the btcd guys may have run into a similer issues, might want to ask over there as well. its late and my memory went to sleep.
 79 2015-09-11 08:57:33 <paveljanik> right, I'll do so
 80 2015-09-11 09:04:53 <paveljanik> #6660
 81 2015-09-11 09:05:09 <paveljanik> 6 more to #6666 ;-)
 82 2015-09-11 09:36:24 <jonasschnelli> paveljanik: i'm online now... but only short.
 83 2015-09-11 09:36:56 <jonasschnelli> You saying a Ctrl-C in bitcoin-cli will shutdown bitcoind?
 84 2015-09-11 09:38:51 <paveljanik> yes
 85 2015-09-11 09:39:02 <jonasschnelli> paveljanik: Yeah. There is something wrong! Confirmed. Ctrl-C in bitcoin-cli shuts down bitcoind.
 86 2015-09-11 09:39:05 <paveljanik> at least here...
 87 2015-09-11 09:39:17 <paveljanik> jonasschnelli, great. Thanks for confirmation.
 88 2015-09-11 09:39:22 <jonasschnelli> Debian 8
 89 2015-09-11 09:40:17 <jonasschnelli> But how's this possible? Does bitcoin-cli passes the SIGPIPE cmd and bitcoind interprets that? I don't get it.
 90 2015-09-11 09:41:20 <paveljanik> iirc the HTTP client stops (after ctrl-c), bitcoind receives SIGPIPE and as there is no handler, nor it is ignored, it simply exits
 91 2015-09-11 09:42:00 <jonasschnelli> paveljanik: try to address it over a PR?
 92 2015-09-11 09:42:14 <paveljanik> I filled #6660
 93 2015-09-11 09:42:32 <paveljanik> I did not read the new webserver code
 94 2015-09-11 09:42:33 <jonasschnelli> Yes. I saw that. But are you up to write a patch?
 95 2015-09-11 09:43:09 <paveljanik> not yet, I have to leave soon. But will look at it next week.
 96 2015-09-11 09:43:31 <jonasschnelli> https://www.mail-archive.com/libevent-users@monkey.org/msg00998.html
 97 2015-09-11 09:43:31 <jonasschnelli> Okay... Maybe this helps:
 98 2015-09-11 11:24:20 <ThomasV> have recent versions of bitcoin merged some form of memorypool curator?
 99 2015-09-11 11:35:16 <dcousens> ThomasV: there have been quite a few PRs related to mempool recently,  I'm not sure about recent versioned releases though
100 2015-09-11 11:42:49 <gmaxwell> oh darn he left before I could get my response out.
101 2015-09-11 11:47:32 <nsh> what would it have been for my curiosity, gmaxwell?
102 2015-09-11 12:04:05 <gmaxwell> to point to the actual PRs!
103 2015-09-11 12:04:21 <gmaxwell> e.g. https://github.com/bitcoin/bitcoin/pull/6557
104 2015-09-11 12:10:15 <nsh> aha
105 2015-09-11 12:13:58 <dcousens> core dumps, core dumps everywhere
106 2015-09-11 12:14:36 <dcousens> Can't decide if its just bitcoind being fun, my SSD taking a crap, or the 2$ chinese USB3->SATA cable I'm using
107 2015-09-11 12:15:52 <gmaxwell> bitcoind crashing would be highly concerning and unusual.
108 2015-09-11 12:16:07 <gmaxwell> can you bring up the crashes in gdb and see where it failed?
109 2015-09-11 12:17:31 <dcousens> gmaxwell: finally got it going again with a re-index,  will see where that takes me first
110 2015-09-11 12:18:09 <dcousens> "terminate called after throwing an instance of 'boost::filesystem::filesystem_error'"
111 2015-09-11 12:18:19 <dcousens> hit an assertion this time :D
112 2015-09-11 12:18:31 <dcousens> yeah... looks like this SSD is dying :(
113 2015-09-11 12:19:23 <dcousens> if you want me to test anything while its on the way out, sounds like it might be a good test bed haha
114 2015-09-11 12:22:06 <dcousens> maybe not, whole directories juts dumped an io error, and I now have unmount at 100% with the drive already disconnect, lol
115 2015-09-11 12:24:57 <gmaxwell> hah okay
116 2015-09-11 12:27:10 <dcousens> brb reboot
117 2015-09-11 12:32:46 <dcousens_> heh, nothing was killing that process, infinitely blocking syscall... naturally even systemd couldn't shutdown.  Lets try drive no.2
118 2015-09-11 12:33:33 <nsh> there are some mount options for SSDs that are advisable for bitcoind's access/write patterns
119 2015-09-11 12:33:40 <nsh> i don't recall the specifics though
120 2015-09-11 12:34:51 <dcousens_> nsh: any recommendations on a fs?
121 2015-09-11 12:35:22 <dcousens_> I usually use ext4, though not sure if the journaling will be worth it
122 2015-09-11 12:37:03 <nsh> ext4 is definitely preferable to ext3 for SSD
123 2015-09-11 12:37:18 <nsh> that may be optimal for bitcoin, but i wouldn't venture to assert that without researching further
124 2015-09-11 12:37:36 <dcousens_> that was my thoughts
125 2015-09-11 12:37:49 <dcousens_> ah well, maybe that drive was too old? 4 years for an SSD?
126 2015-09-11 12:37:57 <dcousens_> (it was ext4)
127 2015-09-11 12:39:04 <nsh> the manufacturers' warranty should be longer than that, but they may assume patterns of usage
128 2015-09-11 12:39:21 <nsh> actually, it seems to vary from model to model: http://www.samsung.com/global/business/semiconductor/minisite/SSD/global/html/support/warranty.html
129 2015-09-11 12:39:48 <nsh> there are known more-troublesome and less-troublesoom manufacturers but again i don't know specifics
130 2015-09-11 12:40:37 <nsh> Total Bytes Written may be the one exceeded, rather than mean-time-to-failure in your case
131 2015-09-11 12:45:59 <nsh> cf. https://wiki.debian.org/SSDOptimization  t
132 2015-09-11 12:50:57 <jgarzik> in regards to Boost...
133 2015-09-11 12:50:58 <jgarzik> https://google-styleguide.googlecode.com/svn/trunk/cppguide.html#Template_metaprogramming
134 2015-09-11 12:51:21 <jgarzik> C++11 + boost removal would be nice
135 2015-09-11 12:52:18 <nsh> what's shiny about C++11?
136 2015-09-11 12:52:27 <dcousens_> nsh: readability
137 2015-09-11 12:52:33 <nsh> ah
138 2015-09-11 12:52:53 <nsh> is that mostly a style rewrite then a whole lot of regression testing then, i guess?
139 2015-09-11 12:53:13 <dcousens_> nsh: depends, jgarzik may be coming at from a point of view of reducing system dependencies
140 2015-09-11 12:53:33 <dcousens_> which would be great too
141 2015-09-11 12:53:36 <Luke-Jr> jgarzik: C++11 is impractical to use/deploy right now :/
142 2015-09-11 12:53:45 <dcousens_> Luke-Jr: how so?
143 2015-09-11 12:54:10 <nsh> could you make an programmer's text editor that doesn't allow you to make semantic changes
144 2015-09-11 12:54:10 <nsh> that's probably not generally possible
145 2015-09-11 12:54:11 <jgarzik> reducing dependencies, reducing non-standard code, reducing compile-time memory usage (#6658) and more
146 2015-09-11 12:54:13 <Luke-Jr> dcousens_: GCC breaks ABI regularly, and since distros (must?) build libraries with C++98, those libraries are incompatible with C++11 compiled programs
147 2015-09-11 12:54:25 <Luke-Jr> dcousens_: eg, we couldn't link to Qt
148 2015-09-11 12:54:34 <jgarzik> In general template usage should be reduced
149 2015-09-11 12:55:42 <Luke-Jr> (notably LLVM does not have this problem)
150 2015-09-11 12:55:52 <dcousens_> Luke-Jr: another thing solved by not having Qt in the core application? :P, but, I haven't had that issue in my own projects w/ C++11, granted none used Qt
151 2015-09-11 12:55:54 <sturles> I just compiled the latest version from git on x86_64, and it segfaults: http://pastebin.com/tfcrh3Nn
152 2015-09-11 12:56:12 <sturles> src/test/test_bitcoin segfaults.
153 2015-09-11 12:56:23 <jgarzik> Solving the immediate problem, the tests could use a less memory intensive harness
154 2015-09-11 12:56:29 <jgarzik> boost--
155 2015-09-11 12:56:48 <Luke-Jr> dcousens_: ABI incompatibilities may be subtle, like the 0.5 exploit ;)
156 2015-09-11 12:57:08 <dcousens_> Luke-Jr: not aware, any writeups around? :)
157 2015-09-11 12:57:38 <Luke-Jr> dcousens_: https://en.bitcoin.it/wiki/Common_Vulnerabilities_and_Exposures#CVE-2012-1910
158 2015-09-11 12:58:14 <Luke-Jr> sturles: latest version = 0.11.0 or git master?
159 2015-09-11 12:58:30 <jgarzik> In general C++ sum mangling ensures very little stealth ABI incompatibilities; far fewer than C or other languages
160 2015-09-11 12:58:37 <Luke-Jr> sturles: looks like it failed to find test data files, but .. I thought that was compiled in nowadays
161 2015-09-11 12:58:42 <jgarzik> an order of magnitude fewer than python, perl, and friends ;p
162 2015-09-11 12:58:58 <jgarzik> *sym mangling
163 2015-09-11 12:59:04 <jgarzik> stupid OSX autocorrect
164 2015-09-11 12:59:05 <sturles> Luke-Jr: git master
165 2015-09-11 12:59:26 <sturles> Segfault is in AlertApplies
166 2015-09-11 12:59:32 <nsh> sturles, can you enable coredump [rlimic -c] and upload the corefile somewhere?
167 2015-09-11 12:59:40 <sturles> bitcoind segfaults as well, btw
168 2015-09-11 13:01:17 <jgarzik> sturles, add
169 2015-09-11 13:01:22 <jgarzik> sturles, odd.  platform?
170 2015-09-11 13:01:34 <jgarzik> os/cpu/version
171 2015-09-11 13:01:56 <jgarzik> I would run bitcoind inside gdb
172 2015-09-11 13:02:49 <sturles> Linux lara 3.2.0-4-amd64 #1 SMP Debian 3.2.68-1+deb7u1 x86_64 GNU/Linux
173 2015-09-11 13:05:05 <jgarzik> sturles, how much ram?
174 2015-09-11 13:06:09 <sturles> Ran inside gdb: http://pastebin.com/HGyyd8B3
175 2015-09-11 13:06:13 <sturles> 2 GiB
176 2015-09-11 13:44:36 <dcousens_> hmmm
177 2015-09-11 13:45:02 <dcousens_> Whats the odds of 3 drives dying? I'm starting to think its more likely my 2 USB3/SATA cables are bung lol
178 2015-09-11 13:45:29 <Luke-Jr> dcousens_: same make/lot?
179 2015-09-11 13:46:26 <dcousens_> Luke-Jr: 4* drives? no, HDD WD2014, HDD WD2012, SanDisk 120G SSD, Corsair F60
180 2015-09-11 13:47:34 <dcousens_> heh, switched to another USB3/SATA,  all is well
181 2015-09-11 13:47:46 <dcousens_> for now anyway...
182 2015-09-11 13:48:26 <helo> hot glue gun patch