1 2014-12-22 00:02:08 <wallet42> hi what is the earliest estimate OP_CHECKLOCKTIMEVERIFY could be live?
  2 2014-12-22 01:04:22 <BlueMatt> wallet42: in exactly 11 minutes/hours/days/weeks/months/years
  3 2014-12-22 01:04:30 <BlueMatt> pick a unit, but it must be 11!
  4 2014-12-22 01:08:19 <wallet42> im going with swatch internet time
  5 2014-12-22 01:13:20 <Luke-Jr> BlueMatt: tonal days
  6 2014-12-22 01:34:59 <proserpine-> just to be SURE, a bitcoin address based from the same uncompressed and compressed public key DIFFERS, right?
  7 2014-12-22 01:35:15 <BlueMatt> yes
  8 2014-12-22 01:35:25 <proserpine-> k - ty
  9 2014-12-22 01:47:06 <Luke-Jr> proserpine-: but it is still address reuse, note
 10 2014-12-22 01:48:48 <proserpine-> Luke-Jr: yea because the other address can be constructed looking at the pubkey from scriptsig?
 11 2014-12-22 01:49:02 <Luke-Jr> right
 12 2014-12-22 01:49:07 <Luke-Jr> the problem is really key reuse, not address reuse
 13 2014-12-22 01:49:16 <proserpine-> yup
 14 2014-12-22 01:49:33 <Luke-Jr> 
ACTION ponders if this ought to be explained on the wiki

 15 2014-12-22 01:49:51 <Luke-Jr> jgarzik: happen to know how Linux kernel modules are implemented?
 16 2014-12-22 01:51:00 <sipa> Luke-Jr: that sounds like a very vague question
 17 2014-12-22 01:51:06 <sipa> vague/broad
 18 2014-12-22 01:52:03 <proserpine-> http://www.tldp.org/LDP/lkmpg/2.6/html/ ?
 19 2014-12-22 01:53:06 <Luke-Jr> sipa: yes, I have no clue what specifically I need to know there :/
 20 2014-12-22 01:53:31 <Luke-Jr> basically I'm interested in how kernel modules get loaded, are they patched or whatever
 21 2014-12-22 01:53:42 <Luke-Jr> trying to RE bitmain's antminer blobs ..
 22 2014-12-22 01:54:20 <BlueMatt> Luke-Jr: no, they're loaded very similarly to .so's
 23 2014-12-22 01:54:28 <BlueMatt> you have an init and exit function, if you want
 24 2014-12-22 01:54:40 <BlueMatt> otherwise you have just new functions
 25 2014-12-22 01:54:59 <BlueMatt> (and ofc there are apis to register handling a device)
 26 2014-12-22 01:55:00 <Luke-Jr> I can't say I really know how .so's are loaded either.
 27 2014-12-22 01:55:15 <Luke-Jr> this blob seems to have a function at address 0, that does nothing but return -25
 28 2014-12-22 01:55:16 <BlueMatt> anyway, broad question, not particularly ot :p
 29 2014-12-22 01:55:29 <BlueMatt> then probably ignore that function :p
 30 2014-12-22 01:55:29 <Luke-Jr> but it's in a place I would expect some kind of hardware communication
 31 2014-12-22 01:55:48 <sipa> well the big question is whether it uses relocation, or position-independent code, i guess
 32 2014-12-22 01:55:57 <BlueMatt> true
 33 2014-12-22 01:56:09 <sipa> (and depending on the platform, a .so can use both afaik)
 34 2014-12-22 01:56:54 <Luke-Jr> I found it strange to have addresses with code at 0
 35 2014-12-22 01:57:02 <Luke-Jr> so it must have something like that
 36 2014-12-22 01:57:45 <Luke-Jr> really tempting to just drop antminer S1 from the "future supported" list and apologise to everyone who was waiting for it
 37 2014-12-22 01:57:54 <Luke-Jr> until the day when/if bitmain cares to comply with GPL
 38 2014-12-22 02:30:10 <jgarzik> Luke-Jr, yes
 39 2014-12-22 02:31:28 <Luke-Jr> jgarzik: I've hand-decompiled one of the functions to http://codepad.org/nTJQgDfW - what I find confusing is the code at address 0, which is a mere return -25 in objdump
 40 2014-12-22 02:44:39 <jgarzik> Luke-Jr, sounds like code that is runtime patched to not do that
 41 2014-12-22 02:44:58 <jgarzik> Luke-Jr, with the code being compiled into default-safe mode
 42 2014-12-22 02:45:08 <jgarzik> which fails, because (presumably) runtime fixups have not been applied
 43 2014-12-22 02:45:26 <jgarzik> part of the kmod loading protocol has not been executed somewhere
 44 2014-12-22 02:45:50 <jgarzik> the kernel rewrites its own code in various places
 45 2014-12-22 02:45:56 <Luke-Jr> jgarzik: how do I figure out what the end result of that is?
 46 2014-12-22 02:46:11 <jgarzik> Luke-Jr, load a module and look a /dev/mem?
 47 2014-12-22 02:46:16 <Luke-Jr> hmm
 48 2014-12-22 02:46:26 <jgarzik> Luke-Jr, an on-disk representation != in-memory representation
 49 2014-12-22 02:46:32 <jgarzik> Luke-Jr, just like userspace programs
 50 2014-12-22 02:46:42 <Luke-Jr> hmm
 51 2014-12-22 02:46:56 <Luke-Jr> jgarzik: usually the userspace programs are possible to follow from objdump though! :P
 52 2014-12-22 02:49:17 <Luke-Jr> 
ACTION ponders how he'll find it in /dev/mem

 53 2014-12-22 03:01:15 <Luke-Jr> jgarzik: any way to get it as-in-/dev/mem with debug symbols?
 54 2014-12-22 03:09:22 <jgarzik> Luke-Jr, "yes" but my memory gets rusty at this point...  :/   the kernel itself needs to dump debug info, so it has that internally.
 55 2014-12-22 03:09:34 <jgarzik> Luke-Jr, my preferred debugger for that reason was always a custom kernel w/ added printk's
 56 2014-12-22 03:10:21 <Luke-Jr> which sadly doesn't work for blobs really
 57 2014-12-22 03:37:39 <Luke-Jr> 
ACTION ponders why objdump is giving him different disassembly for the same bytes :x

 58 2014-12-22 03:56:51 <Luke-Jr> hmm
 59 2014-12-22 03:56:54 <Luke-Jr> in /dev/mem 11303d8:       0c000000        jal     0x0
 60 2014-12-22 06:02:56 <midnightmagic> that's pretty cool; memory profile is much less as of HEAD of about Dec 12. RSS is only at 350MB for mainnet and 100MB for testnet right now.
 61 2014-12-22 06:06:03 <gmaxwell> midnightmagic: right, you can decrease it further by lowering the database cache, we increased it at the same time. (As we'd saved so much elsewhere...)
 62 2014-12-22 06:07:15 <midnightmagic> very nice, it means i can run it better on small hardware. hooray for extending the useful lifetime of machines I've slaved over for a decade :)
 63 2014-12-22 06:07:18 <gmaxwell> a minimum memory usage run would be something like -dbcache=4 -par=1 -rpcthreads=1
 64 2014-12-22 06:08:13 <gmaxwell> midnightmagic: more interesting is keeping it usable in VPSes (which tend to be memory starved), and giving a little more breathing room in DOS attacks that cause higher memory usage.
 65 2014-12-22 06:08:18 <brand0> midnightmagic, you were having problems running regtestnet?
 66 2014-12-22 06:16:48 <midnightmagic> brand0: I was not. The machine in question has an unusual amount of RAM onboard (2GB) so I'm a ways off from obsolescence for it.
 67 2014-12-22 06:17:47 <midnightmagic> gmaxwell: was it libsec you were curious about on big-endian?
 68 2014-12-22 06:27:39 <gmaxwell> midnightmagic: all tested on BE now, too slow.  You can, however, start testing bitcoind on BE if your BE host has enough horsepower and space!
 69 2014-12-22 06:41:07 <midnightmagic> :-P
 70 2014-12-22 06:41:17 <midnightmagic> you didn't tell me it was more than curiosity :)
 71 2014-12-22 06:42:13 <gmaxwell> midnightmagic: it wasn't, but I cleaned and the PPC hosts were in the way so I hooked them up.
 72 2014-12-22 06:42:30 <midnightmagic> what's the PPC platform you're on?
 73 2014-12-22 06:42:49 <midnightmagic> got some mac or something or do you have something more interesting
 74 2014-12-22 06:43:08 <gmaxwell> xserv G5, old dual processor rackmount macs. (I have a pair, one with debian, one with a really old copy of OSX)
 75 2014-12-22 06:43:18 <midnightmagic> ah neat.
 76 2014-12-22 06:43:38 <gmaxwell> 8gb ram too. Actually not bad systems for the $50/ea or whatever I paid for them. Still slow, but they're old.
 77 2014-12-22 06:43:45 <midnightmagic> too bad cell went the way of the dodo, i would've liked to have one of those coprocessor boards at a decent price
 78 2014-12-22 06:44:02 <midnightmagic> I got my quad-core G5 ppc mac for $20 :0D
 79 2014-12-22 06:44:40 <gmaxwell> Yea, I used to have a PS3 with linux on it, but then I played portal 2 and it upgraded the system and made linux available. (I'd even bought the darn thing for porting Opus to CELL, which I'd done)
 80 2014-12-22 06:46:07 <midnightmagic> there were machines based around cell but when I looked into the prices they were exhorbitant!
 81 2014-12-22 06:46:18 <midnightmagic> completely absurd pricing..
 82 2014-12-22 07:30:38 <maaku> I still have a japanese PS3 in storage with the firmware that was root exploitable
 83 2014-12-22 07:31:06 <maaku> intended to do some cell + nvidia gpu hacking, but of course never had the time :(
 84 2014-12-22 07:33:58 <proserpine-> wait, is there BE support for bitcoin??
 85 2014-12-22 07:34:42 <proserpine-> bitcoind*
 86 2014-12-22 07:36:30 <gmaxwell> proserpine-: not merged yet, PR 5510, seems to work so far.
 87 2014-12-22 07:37:58 <proserpine-> heh coolio, i will test
 88 2014-12-22 07:38:53 <gmaxwell> (I have a PPC host synced up to tip with master)
 89 2014-12-22 08:21:47 <wumpus> well it'd be nice if it was possible to run bitcoind on say, off-the-shelf NAS boxes, or even on openwrt routers, although that's still a long way away with regard to both verification performance and memory usage
 90 2014-12-22 08:23:45 <gmaxwell> wumpus: well, you can now more or less reasonably run it on a $35 board (odroid c1).
 91 2014-12-22 08:24:08 <wumpus> but big endian support is a necessary step for that as a lot use mips
 92 2014-12-22 08:24:14 <gmaxwell> which does get us into the realm of reasonably having a full node inside miner hardware, since thats the price point of their linux running microcontroolers.
 93 2014-12-22 08:24:17 <wumpus> gmaxwell: yep
 94 2014-12-22 08:30:20 <wumpus> that's indeed a very nice board for the price
 95 2014-12-22 08:30:50 <Belxjander> gmaxwell / wumpus  : Raspberry Pi with an External USB connected HDD enclosure (with HDD populated for storage) works as a base cheap device for at least blockchain recording...
 96 2014-12-22 08:31:07 <Belxjander> as I have that setup here using a very cheap RPi and a spare SATA HDD doing little else
 97 2014-12-22 08:32:01 <gmaxwell> Belxjander: rpi is really pretty underpowered. I don't consider bitcoind as working on an rpi. though it does kinda limp along sorta.
 98 2014-12-22 08:32:17 <wumpus> rpi is *weak* though, though I'm happy you've got it to work
 99 2014-12-22 08:32:21 <Belxjander> I'm also looking at what I can do for running bitcoin-core on a BE PPC setup as well
100 2014-12-22 08:32:40 <Belxjander> wumpus: I only wanted to set it up for blockchain recording at this point
101 2014-12-22 08:32:54 <Belxjander> get a version of the blockchain "on disk" and then work from there for the AOS build
102 2014-12-22 08:32:58 <wumpus> it isn't that cheap compared to say the board gmaxwell mentions, and just a lot of slower, it's just that tons of them have been produced
103 2014-12-22 08:33:09 <wumpus> Belxjander: sure
104 2014-12-22 08:33:31 <gmaxwell> rpi is the same price as odroid c1, $35, IIRC.
105 2014-12-22 08:34:11 <midnightmagic> Belxjander: you've been around for a dog's age; does this mean you're also recording fork events? how far back does your blockchain database go?
106 2014-12-22 08:34:41 <midnightmagic> Belxjander: also, are you recording testnet and how long have you been doing that?
107 2014-12-22 08:35:50 <Belxjander> midnightmagic: only recently got the RPi to do the build as I rebuilt it as an ArchLinux installation with the SATA HDD paired with it as a /home directory
108 2014-12-22 08:36:12 <Belxjander> I started with the jgarzik torrent data and then had it record additional blocks from there
109 2014-12-22 08:42:33 <Luke-Jr> midnightmagic: fwiw, I did save a lot of old (2011) stale blocks: http://luke.dashjr.org/tmp/code/staleblocks.txz (16 MB)
110 2014-12-22 08:44:32 <midnightmagic> Luke-Jr: thank you
111 2014-12-22 08:44:51 <midnightmagic> Luke-Jr: would you mind posting a sha2 of that file in here?
112 2014-12-22 08:44:56 <midnightmagic> .. by any chance
113 2014-12-22 08:45:02 <gmaxwell> Luke-Jr: someone should probably commit to a hash of that file ASAP to prove that they existed as of now at least.
114 2014-12-22 08:45:13 <midnightmagic> :)
115 2014-12-22 08:45:16 <Luke-Jr> 4b2fe99115e8d3ad0956a9dd76edca8a365551d6a7c5bb0b2adbed0364e53ed0  staleblocks.txz
116 2014-12-22 08:45:18 <gmaxwell> I probably have an old disk with a bunch of upto 2012 orphans.
117 2014-12-22 08:45:35 <Luke-Jr> note that I did that on my webserver, so if it turns out to be compromised for any reason, the hash is worthless
118 2014-12-22 08:46:06 <Luke-Jr> gmaxwell: what does it matter when they existed?
119 2014-12-22 08:46:25 <gmaxwell> Luke-Jr: because right now I can create a kazillion old forks pretending to do all kinds of psycho stuff.
120 2014-12-22 08:46:35 <midnightmagic> should be showing up in namecoin by the time the next block shows up
121 2014-12-22 08:46:42 <gmaxwell> E.g. making it look like people knew about recently discovered vulnerabilties long ago, et.c
122 2014-12-22 08:47:06 <Luke-Jr> gmaxwell: timestamping these doesn't say anything about the validity of others
123 2014-12-22 08:48:18 <gmaxwell> no but what it does say is that if you made them up, it was before now. had we done this back when diff 100k was no so totally trivial it would have been a bit of evidence that they weren't fake.
124 2014-12-22 08:49:14 <Luke-Jr> sure, I just don't see the practical value in proving that :p
125 2014-12-22 08:49:17 <midnightmagic> holy moly fast torified download batman
126 2014-12-22 08:50:12 <gmaxwell> Luke-Jr: well someone would like to look at historic blocks and, you know, know that they're actually historic.
127 2014-12-22 08:51:04 <midnightmagic> gmaxwell: hey does this mean we're going to build a bunch of crazy blocks to freak people out in 50 years when the crazy people rediscover it and don't think to check the logs of this channel?
128 2014-12-22 08:51:08 <midnightmagic> :-D
129 2014-12-22 08:51:33 <midnightmagic> let's fuel some distant future conspiracy theories. surely they'll still be around then.
130 2014-12-22 08:51:59 <midnightmagic> 4b2fe99115e8d3ad0956a9dd76edca8a365551d6a7c5bb0b2adbed0364e53ed0  staleblocks.txz
131 2014-12-22 08:52:34 <Luke-Jr> when/if someone starts a proper timestamping service, someone should timestamp the daily log here :P
132 2014-12-22 08:53:03 <midnightmagic> and if someone does, don't ever tell anyone you're doing it unless you want to (potentially) answer subpoenas until the end of time
133 2014-12-22 08:53:57 <midnightmagic> aw man I was gyp'd, I don't think this is a quad-core g5 at all. Bah. Worst $20 ever spent.
134 2014-12-22 08:54:09 <wumpus> midnightmagic: ?
135 2014-12-22 08:54:30 <midnightmagic> wumpus: the machine I promised I'd be installing netbsd on so I could help test some of the libsec and big-endian diffs.
136 2014-12-22 08:54:37 <wumpus> midnightmagic: they send you a fake device?
137 2014-12-22 08:55:03 <gmaxwell> midnightmagic: sometimes people on ebay (for example) mismark things because they identify them by finding a device that _looks_ just like them.
138 2014-12-22 08:55:04 <midnightmagic> wumpus: no, my (former) coworker clearly got a crappier machine than he said he did.
139 2014-12-22 08:55:24 <midnightmagic> no it was old stock from (former) workplace that we fought over and drew straws over for spare parts.
140 2014-12-22 08:55:38 <wumpus> I did know about chinese counterfeit USB sticks and harddisks, but ARM boards
141 2014-12-22 08:55:40 <wumpus> oh :)
142 2014-12-22 09:00:29 <jonasschnelli> was there already a discussion about test coverage history? Tools like coveralls.io?
143 2014-12-22 09:02:11 <Luke-Jr> we used to have something in pulltester for it, but I think it got disabled :/
144 2014-12-22 09:03:06 <gmaxwell> jonasschnelli: we had a thing in pull tester that would whine when your commits decreased coverage, but it was too sensitive (in particular the randomized tests caused false positives) and it got disabled.
145 2014-12-22 09:03:58 <gmaxwell> If it doesn't do branch coverage testing like LCOV then I think it's not really that interesting.
146 2014-12-22 09:04:16 <jonasschnelli> hmm... maybe it makes sense to use coveralls.io? Sample here: https://coveralls.io/r/cainus/node-coveralls?branch=master
147 2014-12-22 09:04:31 <Luke-Jr> I tried to figure coveralls out for BFGMiner a while ago, and gave up for some reason
148 2014-12-22 09:04:41 <Luke-Jr> not that BFGMiner has lots of tests..
149 2014-12-22 09:05:02 <jonasschnelli> i assume it's pretty seamless with traviss
150 2014-12-22 09:05:25 <Luke-Jr> it wasn't simply a matter of turning it on, at least
151 2014-12-22 09:05:49 <gmaxwell> jonasschnelli: seems to only do line coverage, which is not hugely informative. At least where we actually care we do want to see that e.g. both success and failure cases have been tested.
152 2014-12-22 09:08:00 <jonasschnelli> tracking the coverage on codeline-basis looks also informative for me... but yes. it's not everything.
153 2014-12-22 09:08:59 <jonasschnelli> i think including coveralls.io in travis.ml would be pretty easy...
154 2014-12-22 09:09:18 <Luke-Jr> IMO anything is better than nothing.
155 2014-12-22 09:09:36 <gmaxwell> 
ACTION sigh well just one more excuse for a half assed thing that we'll not bother making right.

156 2014-12-22 09:09:50 <Luke-Jr> I mean, ideally, we'd have only free software for our infrastructure, but this is still obviously better than waiting for a free solution to pop up
157 2014-12-22 09:09:51 <gmaxwell> Like our builder that can no longer give windows binaries for people to try.
158 2014-12-22 09:09:52 <Belxjander> would it be plausible to iterate all the blocks out as JSON text into individual files by numeric index from the initial block?
159 2014-12-22 09:10:20 <gmaxwell> Luke-Jr: we _had_ a free solution, it's just that our tests sometimes honestly had lower coverage and the thing we setup whined.
160 2014-12-22 09:10:37 <gmaxwell> Belxjander: its certantly plausable, would waste a ton of space.
161 2014-12-22 09:11:01 <gmaxwell> Luke-Jr: which I assume would still be the case now, though presumably we've tightened up the coverage a little bit.
162 2014-12-22 09:11:03 <wumpus> uploading binaries from the test system would still be possible, it's just that it creates a potential security issue
163 2014-12-22 09:11:09 <Luke-Jr> gmaxwell: a free solution that interferes with Bitcoin development is sadly inferior to a non-free solution that works reliably
164 2014-12-22 09:12:03 <wumpus> which serving binaries from random pull requests does in the first place, of course, but they'd be effectively unauthenticated and be able to overwrite binaries of other pulls as well
165 2014-12-22 09:12:04 <gmaxwell> Luke-Jr: the issue was not in the test tool, the issue was in our tests. (well not really 'issue' that some of the tests are randomized is intentional, since it slightly increases coverage for a given amount of cpu.. since we only need one failure to know something is wrong)
166 2014-12-22 09:12:28 <Luke-Jr> gmaxwell: the issue IMO was that pulltester often just fell apart entirely and complained about everything for no reason
167 2014-12-22 09:12:33 <Luke-Jr> bigger issue*
168 2014-12-22 09:12:36 <jonasschnelli> test coverage per code line would probably help, drive people to increase amount of unit tests?
169 2014-12-22 09:12:39 <lewellyn> speaking of free solutions, i have a list of list providers to finish going through :P
170 2014-12-22 09:12:40 <wumpus> that's a serious issue and the reason I've been holding that back, I don't want our domain to start serving malware
171 2014-12-22 09:13:19 <Luke-Jr> wumpus: domains are cheap. setup bitcoin-malware.org?\
172 2014-12-22 09:13:23 <gmaxwell> hah
173 2014-12-22 09:13:32 <lewellyn> make it auto-download bitcoin core
174 2014-12-22 09:13:56 <gmaxwell> "Sure, someone just posted a fix for that, you can test it at https://bitcoin-malware.com/builds/20150121/..."
175 2014-12-22 09:14:04 <wumpus> :-)
176 2014-12-22 09:14:10 <Luke-Jr> gmaxwell: that makes me like it even better
177 2014-12-22 09:14:30 <wumpus> I'm really afraid of that and it's one more thing that needs to be monitored
178 2014-12-22 09:14:48 <lewellyn> i prefer a page which explains in detail that bitcoin isn't malware and then does a redirect to the appropriate installer for the user agent, after like 7 seconds. :D
179 2014-12-22 09:14:51 <Luke-Jr> bitcoin-builds-that-will-steal-your-money.net
180 2014-12-22 09:14:53 <wumpus> I have enough things to babysit already
181 2014-12-22 09:15:15 <gmaxwell> jonasschnelli: tests that only increase line coverage and don't actually exaust the interesting cases are bad because they hide bugs (people think code is tested when its not).
182 2014-12-22 09:15:23 <wumpus> so if you want to provide a build for a pull, provide it yourself and use your own reputation to convince people not ours
183 2014-12-22 09:15:31 <lewellyn> "so how do i download bitcoin?" "go to bitcoin-malware.com and wait. you WILL get the wallet."
184 2014-12-22 09:15:46 <Luke-Jr> wumpus: yeah, we'll need to make builds much easier if we want that IMO
185 2014-12-22 09:15:57 <wumpus> Luke-Jr: it's dead easy with the depends system
186 2014-12-22 09:16:04 <gmaxwell> wumpus: buildings things like OSX images require things like access to a mac to extract files and such.
187 2014-12-22 09:16:05 <Luke-Jr> wumpus: it's dead slow too
188 2014-12-22 09:16:18 <wumpus> Luke-Jr: only the first time
189 2014-12-22 09:16:33 <wumpus> Luke-Jr: it's very smart and recompiles only what changed
190 2014-12-22 09:16:35 <gmaxwell> I should try the depends stuff and see if it actually works on things that aren't ubuntu.
191 2014-12-22 09:16:35 <Luke-Jr> wumpus: is it dead easy for Windows developers?
192 2014-12-22 09:16:40 <wumpus> Luke-Jr: and it doesn't need a VM or such
193 2014-12-22 09:16:50 <wumpus> Luke-Jr: which windows developers?
194 2014-12-22 09:17:03 <Luke-Jr> wumpus: *crickets*
195 2014-12-22 09:17:11 <gmaxwell> :P
196 2014-12-22 09:17:11 <gmaxwell> wumpus: well if they had a way to get builds!
197 2014-12-22 09:17:28 <wumpus> gmaxwell: well the osx problem is not trivially solvable, that's a legal issue not a technical one
198 2014-12-22 09:17:31 <jonasschnelli> gmaxwell? i think you don't need a mac to bulild a OSX image with gitian
199 2014-12-22 09:18:01 <gmaxwell> wumpus: in any case, it's annoying that travis has no facility to just fetch the builds it just did. It was a step down that we'll likely never fix because the travis stuff is good enough.
200 2014-12-22 09:18:09 <wumpus> gmaxwell: I don't want them to abuse our test system for that tough.
201 2014-12-22 09:18:14 <gmaxwell> (well good enough if you're someone who can build your own binaries)
202 2014-12-22 09:18:33 <wumpus> gmaxwell: it's up to them to fix that, not up to us
203 2014-12-22 09:19:09 <wumpus> if you can't build your own binaries you shouldn't be doing development
204 2014-12-22 09:19:34 <gmaxwell> wumpus: if someone was sending random non-serious PRs to get builds we'd zot them, we never had issues with that.  It means that when someone shows up in #bitcoin with some issue we just fixed (or an issue I just fixed in response to their request), and they're on windows or OSX I've got not good way to make them a binary to try out.
205 2014-12-22 09:19:36 <wumpus> it's dangerous enough for developers with 20+ years of experience
206 2014-12-22 09:20:20 <gmaxwell> Which is something I've done a dozen times without issue until we just lost the capability.
207 2014-12-22 09:20:32 <wumpus> gmaxwell: you don't have gitian set up?
208 2014-12-22 09:20:43 <wumpus> it's really not rocket science man :)
209 2014-12-22 09:21:18 <gmaxwell> never could get it working right on fedora. Not even with devrandom helping.  Guess I'm due to try again now that my laptop is gentoo.
210 2014-12-22 09:21:43 <wumpus> yes, macosx requires some extra files to be extracted on a mac system, but I'm sure someone will send them to you if you ask in pm
211 2014-12-22 09:21:44 <Luke-Jr> FWIW, I don't have gitian set up post-depends changes
212 2014-12-22 09:22:06 <Luke-Jr> wumpus: that's not quite secure, is it? :|
213 2014-12-22 09:22:28 <Luke-Jr> or I guess the hash is still verified
214 2014-12-22 09:22:30 <gmaxwell> warren apparently got it working on fedora but it required a bunch of stuff.  Generally, I'm really hesitant about stuff that requires me taking a bunch of binary code. :-/  I really should setup another host for it. Just more housekeeping which I'm massively behind on.
215 2014-12-22 09:22:33 <wumpus> Luke-Jr: well if it gets the same output.
216 2014-12-22 09:22:55 <Luke-Jr> wumpus: my point is if A, B-with-input-from-A, and C-with-input-from-A match, it's not as meaningful
217 2014-12-22 09:23:21 <Luke-Jr> gmaxwell: taking binary code for Mac builds is unavoidable
218 2014-12-22 09:23:28 <Luke-Jr> just a matter of from where
219 2014-12-22 09:23:36 <wumpus> gmaxwell: a debian VM works fine, see eg https://github.com/bitcoin/bitcoin/blob/master/doc/gitian-building.md
220 2014-12-22 09:23:45 <Luke-Jr> gmaxwell: FWIW, nested VT-x *should* work fine now - and I think gitian works without VT-x anyway
221 2014-12-22 09:24:01 <Luke-Jr> I did finally get gitian running natively on Gentoo with KVM though
222 2014-12-22 09:24:04 <gmaxwell> Luke-Jr: the binary stuff it needs, does it need to actually _run_ it?
223 2014-12-22 09:24:16 <Luke-Jr> gmaxwell: no, I think it just links with it
224 2014-12-22 09:24:38 <gmaxwell> in any case the whole host distro setup ends up being binary and irreproducable, and that all gets run.
225 2014-12-22 09:24:50 <Luke-Jr> I'm not aware of it doing any magic to execute Mac software on Linux, at least :D
226 2014-12-22 09:25:22 <Luke-Jr> gmaxwell: yes, I can't really advise people run Bitcoin software on Mac/Windows.
227 2014-12-22 09:25:27 <wumpus> Luke-Jr: it works without
228 2014-12-22 09:25:35 <wumpus> Luke-Jr: use a VM + LXC inside that
229 2014-12-22 09:25:38 <gmaxwell> oh does the depends stuff. no longer shackel the build enviroment?
230 2014-12-22 09:25:44 <Luke-Jr> wumpus: ?
231 2014-12-22 09:25:56 <gmaxwell> ah yea okay it does.
232 2014-12-22 09:26:05 <wumpus> gmaxwell: well it doesn't go as far as creating a toolchain yet
233 2014-12-22 09:26:08 <wumpus> gmaxwell: pull requests welcome.
234 2014-12-22 09:26:26 <wumpus> it's a great step forward, but it's not up to replacing gitian yet
235 2014-12-22 09:26:52 <gmaxwell> wumpus: yea, ... the whole gitian "now upgrade the OS before building" really tosts the independance of the builds.
236 2014-12-22 09:27:24 <jonasschnelli> you don't need a mac to produce the MacOSX10.7.sdk.tar.gz
237 2014-12-22 09:27:25 <wumpus> gmaxwell: it does, but at least someone wrote it and it works, and it produced actually deterministic results most of the time... it's not an ideal solution but that just takes time
238 2014-12-22 09:27:44 <gmaxwell> sorry, I certantly wasn't trying to complain.
239 2014-12-22 09:27:46 <Luke-Jr> jonasschnelli: all the non-Mac extractors failed last I tried
240 2014-12-22 09:28:00 <wumpus> jonasschnelli: you don't?
241 2014-12-22 09:28:09 <wumpus> jonasschnelli: please makea  script please make a script
242 2014-12-22 09:28:25 <gmaxwell> wumpus: I'm happy we use it. I had just wasted too many hours uselessly trying to get it to work in the past. I should try again.
243 2014-12-22 09:28:42 <wumpus> gmaxwell: sorry you got burnt at some time, it's become slightly better
244 2014-12-22 09:28:51 <jonasschnelli> i'll provide a script
245 2014-12-22 09:29:04 <wumpus> looking at the number of gitian signatures that we get these days, people don't have as much trouble getting it to work anymore
246 2014-12-22 09:29:44 <gmaxwell> well there was some initial incompatiblity, then I got asked to use lxc but it turned out that wasn't actually working at that time. Then I think I mostly got it working but I built my outer VM with not enough space for it to actually work, and the host I built it on was low on space and yadda yadda.
247 2014-12-22 09:30:14 <gmaxwell> not really anyone's fault, but since other people had it working I didn't feel it justified more time. I guess I'll try again!
248 2014-12-22 09:31:50 <Luke-Jr> amid the Gentoo trolling stuff, I've had a few requests for Windows builds of my patched version, so I should update gitian too I guess XD
249 2014-12-22 09:33:13 <gmaxwell> wumpus: WRT cortex a8, I generally find testing on a8 is most useful: almost everything that makes a8 faster makes a9 faster, but the reverse is not true.... though I doubt we'll ever have many people running bitcoin core on a8. (I guess a5 is also in order, and thats now being found in quite a few faster chips like that new $35 odroid)
250 2014-12-22 09:33:48 <Belxjander> A( == ARM?
251 2014-12-22 09:33:52 <Belxjander> A8 even
252 2014-12-22 09:34:03 <Luke-Jr> all Cortex are ARM AFAIK?
253 2014-12-22 09:34:32 <wumpus> gmaxwell: those odroid boards sounds interesting, I'm probably going to get one
254 2014-12-22 09:34:49 <gmaxwell> Belxjander: yea, these are arm.
255 2014-12-22 09:35:13 <Belxjander> maybe a dumb question... but is it possible to get a block by the height value?
256 2014-12-22 09:35:21 <gmaxwell> wumpus: I would have ordered one already, except my novena just showed up and I try really hard to not buy more hardware when all the hardware I already have (of that general class) isn't in use.
257 2014-12-22 09:35:22 <Belxjander> and have it exported in JSON format ?
258 2014-12-22 09:36:40 <wumpus> gmaxwell: same kind of issue here
259 2014-12-22 09:37:18 <Luke-Jr> I wonder how the odroid compares with the rpi on power use
260 2014-12-22 09:37:26 <Luke-Jr> Belxjander: #bitcoin
261 2014-12-22 09:40:11 <Belxjander> Luke-Jr: okay...thanks not a development topic then?
262 2014-12-22 09:40:52 <Luke-Jr> Belxjander: no, it's a user topic; the answer is getblock(getblockhash(height)) keeping in mind multiple blocks may have the same height at different times
263 2014-12-22 09:42:31 <Belxjander> Luke-Jr: well I want to get anything up to before the last hours worth
264 2014-12-22 09:43:56 <wumpus> I wonder how Marvell  PXA2128/PJ4 compares, that's in my OLPC XO-4
265 2014-12-22 09:44:22 <Luke-Jr> wumpus: IIRC Marvell 'forked' an old old ARM core for their stuff
266 2014-12-22 09:46:39 <jonasschnelli> gitian: any idea why i get "./bin/gbuild:21:in `system!': failed to run on-target setarch x86_64 bash -x < var/build-script > var/build.log 2>&1 (RuntimeError)"?
267 2014-12-22 09:47:00 <Luke-Jr> jonasschnelli: that's the most vague error you could give ;)
268 2014-12-22 09:47:02 <Luke-Jr> check build.log
269 2014-12-22 09:47:32 <wumpus> Luke-Jr: a lot of talk about how power efficient the odroid-c1 is, but wasn't able to find actual measurements
270 2014-12-22 09:48:20 <wumpus> jonasschnelli: that error message just says something went wrong in the build and you should check build.log, it should probably say that
271 2014-12-22 10:02:07 <jonasschnelli> "wget: unable to resolve host address `samba.org'" is there any special need for a gitian-build internet setup? my debian vm can connect to the inet.
272 2014-12-22 10:03:52 <paveljanik> jonasschnelli, you have to get deps before running it
273 2014-12-22 10:04:28 <wumpus> jonasschnelli: https://github.com/bitcoin/bitcoin/issues/5428
274 2014-12-22 10:04:43 <wumpus> although I wonder why internet access in the builder isn't working, it should
275 2014-12-22 10:05:19 <wumpus> but cfields added a method to download them externally
276 2014-12-22 11:12:04 <btcdrak> jonasschnelli wumpus I can confirm gitian builds do work using the instructions at 5428
277 2014-12-22 11:12:56 <btcdrak> occasionally sf .net borks, but once just rerun make download again and it usually works second time. once you have the deps it's plain sailing.
278 2014-12-22 11:13:05 <wumpus> btcdrak: yep, but what I'm wondering is why the LXC setup instructions broke network connectivity for the builder, that used to just work, probably something that changed in debian
279 2014-12-22 11:19:30 <wumpus> I really need to replicate those steps again
280 2014-12-22 11:19:36 <wumpus> @cfields you had any luck?
281 2014-12-22 11:28:10 <wumpus> "DDoS attacks on in-flight data structures" (#5507) has an exciting title
282 2014-12-22 11:32:20 <wumpus> thinking about our earlier discussion, crosstool-ng + depends could be enough to replace gitian
283 2014-12-22 11:32:33 <sipa> yeah
284 2014-12-22 11:32:50 <sipa> i wonder how cfields thinks about that
285 2014-12-22 11:33:07 <wumpus> crosstool-ng gets you lots of toolchains easily, depends can build bitcoind's for those easily
286 2014-12-22 11:34:01 <wumpus> IIRC it can make mingw's toolchain too, though I have to still try that in practice
287 2014-12-22 11:34:09 <wumpus> macosx will likely be the only exception
288 2014-12-22 11:35:01 <sipa> ah, i hadn't heard about crosstool-ng; was just thinking of using an in-depends built compiler for each
289 2014-12-22 11:35:03 <wumpus> qemu-user can be used to test the generated bitcoind's tests
290 2014-12-22 11:35:19 <wumpus> it all exists, just need a lame script to hook them up :)
291 2014-12-22 11:35:45 <wumpus> crosstool-ng does exactly what you want
292 2014-12-22 11:36:23 <sipa> wumpus: also, asm optimizations for secp256k1 would be very welcome, and likely much more relevant the next few years than x86... but you probably have better things to do
293 2014-12-22 11:36:34 <wumpus> in a handy menuconfig you can configure the target architecture and it will do the work of bootstrapping the compiler and c library and even a gdb if you want that
294 2014-12-22 11:36:49 <sipa> what is rc1 still waiting for, btw?
295 2014-12-22 11:36:59 <wumpus> sipa: https://github.com/bitcoin/bitcoin/pulls?q=is%3Aopen+is%3Apr+milestone%3A0.10.0
296 2014-12-22 11:37:40 <wumpus> sipa: well I actually like doing that, but yes no saying when I get around to it
297 2014-12-22 11:38:38 <wumpus> sipa: we could tag -rc1 without those, of course, as it's a given there will be a -r2
298 2014-12-22 11:39:52 <sipa> i'll do some review and testing of those today
299 2014-12-22 11:40:28 <wumpus> #5253 is sufficiently tested/acked and I'm going to merge that
300 2014-12-22 11:42:07 <Luke-Jr> wumpus: crosstool-ng does not appear to be available on Gentoo FWIW
301 2014-12-22 11:42:22 <Luke-Jr> it would be nice if crossdev worked
302 2014-12-22 11:42:39 <Luke-Jr> (for deterministic matching, I mean)
303 2014-12-22 11:43:24 <wumpus> Luke-Jr: well the toolchain doesn't have to be built deterministicically itself
304 2014-12-22 11:44:06 <wumpus> Luke-Jr: of course you need to build it from the same versions and with the same settings to get the same results out of it
305 2014-12-22 11:44:58 <wumpus> Luke-Jr: crosstool-ng is very easy to build from source, the script could do that too
306 2014-12-22 11:45:23 <wumpus> Luke-Jr: I'm not using a debian package for that either but just running it from its build dir
307 2014-12-22 11:46:11 <wumpus> Luke-Jr: whoo, x86_64-unknown-linux-gnux32 toolchain is built, let's see if it works
308 2014-12-22 11:49:19 <wumpus> ok that didn't work, it simply makes x86_64 executables
309 2014-12-22 11:49:47 <Luke-Jr> wumpus: what does file say on them?
310 2014-12-22 11:50:11 <Luke-Jr> amd64/bin/bash: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.16, stripped
311 2014-12-22 11:50:13 <Luke-Jr> x32/bin/bash:   ELF 32-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 3.4.0, stripped
312 2014-12-22 11:50:37 <wumpus> ./test2: ELF 64-bit LSB  executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 3.12.33, not stripped
313 2014-12-22 11:51:02 <Luke-Jr> hm
314 2014-12-22 11:51:10 <Luke-Jr> what GCC version?
315 2014-12-22 11:51:29 <wumpus> I suppose you need to pass some magic to gcc's and binutil's build too, just the host tuple is not enough?
316 2014-12-22 11:51:46 <wumpus> 4.9.2
317 2014-12-22 11:52:26 <Luke-Jr> ouch, too new for me :P
318 2014-12-22 11:53:17 <Luke-Jr>         if tc_version_is_at_least 4.7 && has x32 $(get_all_abis TARGET) ; then            confgcc+=( --with-abi=$(gcc-abi-map ${TARGET_DEFAULT_ABI}) )        fi
319 2014-12-22 11:55:16 <Luke-Jr> --with-abi=mx32
320 2014-12-22 11:55:17 <Luke-Jr> I think
321 2014-12-22 12:07:14 <wumpus> I've used --with-abi=x32, let's see how far it goes
322 2014-12-22 12:13:36 <wumpus> sipa: #5459 doesn't look to difficult to test, just making validation fail randomly would do, going to give it a try
323 2014-12-22 12:22:29 <wumpus> bitcoind --simulate-overheating-cpu
324 2014-12-22 12:26:17 <Luke-Jr> wumpus: my CPU is regularly overheating, doesn't seem to hurt
325 2014-12-22 12:26:49 <Krellan> Mine too, sigh, too many cats on top of fans
326 2014-12-22 12:27:14 <Krellan> The cats love vanitygen
327 2014-12-22 12:30:37 <wumpus> Luke-Jr: the with-abi=x32 seems to have done it, though it now stops somwhere during build of glibc
328 2014-12-22 12:31:31 <Luke-Jr> Krellan: the other day when I was doing maintenance, I found the CPU heatsink was caked solid with dust :x
329 2014-12-22 12:32:22 <wumpus> especially laptops have that problem in my experience, and it's often hardest to get at the fans
330 2014-12-22 12:33:28 <Luke-Jr> so I basically was running without a fan all that time
331 2014-12-22 12:33:29 <Luke-Jr> lol
332 2014-12-22 12:38:55 <Krellan> Wow... I notice my fan speed slowing, and think I've lost a case fan entirely, need to open and clean it
333 2014-12-22 12:40:20 <wumpus> could be; another hint is if it starts to make lots of noise
334 2014-12-22 12:46:50 <jonasschnelli> gitian: my install.log shows troubles while loading http://10.0.3.2 precise Release.gpg (and others).... any idea?
335 2014-12-22 12:52:27 <average> apparently the news lately have only been of the form
336 2014-12-22 12:52:39 <average> "researchers discover[..]vulnerability[..]"
337 2014-12-22 12:52:55 <average> "researchers discover [..] anyone could [..] your X"
338 2014-12-22 12:52:59 <average> 14:52 <average> Date: Sun, 21 Dec 2014 10:23:10 +0200
339 2014-12-22 12:52:59 <average> 14:52 <average> Title: German Researchers Discover a Flaw That Could Let Anyone Listen to Your Cell Calls
340 2014-12-22 12:55:45 <wumpus> anyone can listen? wow, researchers turn your cell into a digital radio broadcasting device?
341 2014-12-22 12:56:01 <wumpus> oh #bitcoin
342 2014-12-22 13:00:19 <jonasschnelli> :)
343 2014-12-22 13:00:33 <Gnaf> My cell is only used for my dinner and the weather, weater they like it or not, so fuck those researchers
344 2014-12-22 13:00:42 <jonasschnelli> any idea what "W: Cannot check Release signature; keyring file not available /usr/share/keyrings/ubuntu-archive-keyring.gpg" means? building on VirtualBox Debian after wumpus manual
345 2014-12-22 13:02:15 <wumpus> jonasschnelli: no clue :/
346 2014-12-22 13:02:52 <jonasschnelli> i got to love gitian....
347 2014-12-22 13:06:17 <wumpus> jonasschnelli: the 10.0.3.2 error means LXC doesn't even have access to your local host
348 2014-12-22 13:06:44 <wumpus> which is the absolute minimum required for gitian otherwise it can't fetch the inner linux distro's packages
349 2014-12-22 13:07:02 <jonasschnelli> wumpus hmm... okay. let me debug that.
350 2014-12-22 13:07:11 <wumpus> it doesn't necessarily need to be able to connect to the internet (there's a workaround for that) but it does need access to the host
351 2014-12-22 13:07:56 <jonasschnelli> wumpus but it looks like that gitian is designed to run at ubuntu https://github.com/devrandom/gitian-builder/blob/27bd5b3869c1fbbcc9a545a2b983f32d9755e2c8/bin/make-base-vm
352 2014-12-22 13:08:16 <wumpus> jonasschnelli: yes, feel free to switch over the guide to ubuntu
353 2014-12-22 13:08:24 <wumpus> jonasschnelli: I probably made a mistake using debian
354 2014-12-22 13:08:58 <wumpus> not that that, in principle, should have any influence on lxc or networking issues
355 2014-12-22 13:09:28 <jonasschnelli> wumpus: yes. truly... let me install a ubuntu...
356 2014-12-22 13:10:05 <jonasschnelli> i think the LXC network is fine, but the base image was missing
357 2014-12-22 13:10:14 <jonasschnelli> but the base image won't build out-of-the box on debian...
358 2014-12-22 13:11:01 <wumpus> the reason for using debian is that it has less overhead, which is nice when running a vm-ism in a vm, although ubuntu has a server edition too
359 2014-12-22 13:12:38 <jonasschnelli> yes. i also prefer debian. I try to fix gitian instead of creating a ubuntu vm (not another vm please...)
360 2014-12-22 13:15:09 <wumpus> note that the guide tells you to change gitian's IP range, to avoid it overlapping with virtualbox's, maybe something went wrong there?
361 2014-12-22 13:16:02 <wumpus> ie maybe gitian broke the setting
362 2014-12-22 13:16:28 <Gnaf> jonasschnelli: tnx for the gitian guide link, i lost it on my way to nowhere,...
363 2014-12-22 13:17:00 <jonasschnelli> Gnaf i didn't post a link to a guide?!
364 2014-12-22 13:18:52 <Gnaf> jonasschnelli: <jonasschnelli> wumpus but it looks like that gitian is designed to run at ubuntu https://github.com/devrandom/gitian-builder/blob/27bd5b3869c1fbbcc9a545a2b983f32d9755e2c8/bin/make-base-vm
365 2014-12-22 13:19:07 <Gnaf> ;)
366 2014-12-22 13:19:27 <jonasschnelli> thats not a guide. that the source code of bin/make-base-vm
367 2014-12-22 13:19:29 <sipa> Gnaf: there is a guide on how to do gitian builds on debian
368 2014-12-22 13:19:39 <jonasschnelli> Gnaf: the guide is herehttps://github.com/bitcoin/bitcoin/blob/master/doc/gitian-building.md
369 2014-12-22 13:36:05 <jonasschnelli> was there already a discussion about functional testing on Qt level with mouse/keyboard input simulation?
370 2014-12-22 13:36:53 <sipa> hardly
371 2014-12-22 13:37:00 <sipa> how would you do that?
372 2014-12-22 13:38:16 <jonasschnelli> sipa: just brainstorming: build a tiny framework (maybe on cucumber base) where tests could written. The simulate user inputs (keyboard/mouse) and analyze the state and/or create screenshots for a human check.
373 2014-12-22 13:38:25 <jonasschnelli> i once did this of a OSX and iOS app
374 2014-12-22 13:38:48 <jonasschnelli> The = then
375 2014-12-22 13:40:18 <wumpus> http://qualityhour.wordpress.com/2012/02/01/automated-testing-of-qt-apps-with-testability-driver/
376 2014-12-22 13:45:21 <jonasschnelli> wumpus: cool stuff... the Testability Driver could be useful... but...
377 2014-12-22 13:45:48 <jonasschnelli> depends on how important the Qt level still is for bitcoin-core
378 2014-12-22 13:46:19 <wumpus> well, if you want to test it then apparently it is important to you?
379 2014-12-22 13:47:20 <wumpus> though it wouldn't be high on my list of priorities, the UI is simple enough at this point to test manually
380 2014-12-22 13:49:32 <wumpus> also having to update a test framework for every silly change, say, moving a button, will make it more involved to contribute
381 2014-12-22 13:50:36 <wumpus> then again, if you think it's useful, be my guest. I've tried to keep it possible to launch the various widgets without a model attacked, to make UI testing possible without affecting anything else, though that probably has bitrotted over time.
382 2014-12-22 13:50:43 <jonasschnelli> "importance" in general is a big field (but lets don't get philosophical). I think more test would not harm... will reduce the risk of upcoming pulls.
383 2014-12-22 13:50:49 <wumpus> s/attacked/attached/
384 2014-12-22 13:52:56 <sipa> wumpus: when compiling with -flto, my compiler warns that txmempool does a memset for 0 bytes
385 2014-12-22 13:53:04 <sipa> pretty impressive that it can detect such things
386 2014-12-22 13:53:38 <sipa> especially that it doesn't discover this without flto, so it must be indirectly through some instantiation of a function from another object
387 2014-12-22 13:54:42 <wumpus> sipa: indeed
388 2014-12-22 13:57:47 <sipa> hmm, testnet block sync just doesn't get going
389 2014-12-22 13:58:01 <sipa> headers synced fine, and then nothing
390 2014-12-22 13:59:15 <wumpus> strange
391 2014-12-22 13:59:18 <sipa> sorry, probably my local patch
392 2014-12-22 14:02:18 <wumpus> jonasschnelli: but if everything else equal, I'd prefer more testing on the core side
393 2014-12-22 14:04:11 <jonasschnelli> wumpus: Yes. Makes more sense there. More rpc-tests would be good
394 2014-12-22 14:21:36 <jonasschnelli> wumpus i just try – without any particular business reason – to contribute to the bitcoin project. Because i don't have any business reason but i'm willing to spend time, feel free to point me into the direction of important points to contribute to. There is no TODO.txt :)
395 2014-12-22 14:22:44 <wumpus> jonasschnelli: the TODO list is the open issue list :)
396 2014-12-22 14:23:03 <wumpus> jonasschnelli: also you can help out *a lot* by reviewing other's changes
397 2014-12-22 14:23:19 <jonasschnelli> which misses a priority system...
398 2014-12-22 14:23:23 <wumpus> that's really where the bottleneck is, testing and reviewing
399 2014-12-22 14:23:36 <wumpus> jonasschnelli: it's not, there is a medium and high priority label
400 2014-12-22 14:24:13 <jonasschnelli> testing and reviewing requires knowledge of the internal code which one might gain by producing litte code changes... a.k.a. little-projects
401 2014-12-22 14:24:16 <wumpus> jonasschnelli: if you have any doubts about the priority you can ask me, but on the other hand you should also pick things that you find interesting, if it isn't important there wouldn't be an issue open
402 2014-12-22 14:24:31 <jonasschnelli> https://github.com/bitcoin/bitcoin/issues?q=is%3Aopen+is%3Aissue+label%3A%22Priority+High%22
403 2014-12-22 14:24:47 <wumpus> I tend to immediately close issues that fall into 'don't bother' territory
404 2014-12-22 14:25:49 <jonasschnelli> what about the dns/blocking issue? important? theres no issue for that.
405 2014-12-22 14:26:29 <wumpus> I don't know - I've never had any concrete complaints about it at least
406 2014-12-22 14:27:03 <jonasschnelli> but right. the issue list should be "worked down" as goos as possible....
407 2014-12-22 14:27:05 <wumpus> if dns seeding isn't working there's always the built in seeds
408 2014-12-22 14:27:56 <wumpus> which, incidentally, still need to be updated for 0.10, but may be better if someone with a seed node does that as they have the information about reliable peers
409 2014-12-22 14:30:26 <wumpus> jonasschnelli: any update on the berkeleydb leak issue, btw?
410 2014-12-22 14:32:50 <jonasschnelli> wumpus: was trying hard but could not locate the leak. And somehow bdb does strange memory handling. But i will have a final look and will report to the open issue
411 2014-12-22 14:33:30 <jonasschnelli> valgrind on ubuntu gives me a different picture then my "Instruments" on OSX
412 2014-12-22 14:33:32 <wumpus> jonasschnelli: as said, if it is just some initial allocated objects it's not really a problem, only if it increases over time or by use
413 2014-12-22 14:34:02 <jonasschnelli> Indeed it's not a huge problem....
414 2014-12-22 14:34:06 <wumpus> berkeleydb is generally known to be questionable anyway
415 2014-12-22 14:34:19 <jonasschnelli> But it brought me to the point to find a way to get rid of bdb
416 2014-12-22 14:34:19 <wumpus> also we're using an ancient version
417 2014-12-22 14:35:32 <jonasschnelli> I really see a need to slowly shift away from the current wallet format to a more lightweight option
418 2014-12-22 14:36:31 <wumpus> just be aware that anything you come up with has to be supported for virtually forever
419 2014-12-22 14:36:34 <wumpus> no pressure or so
420 2014-12-22 14:36:35 <jonasschnelli> i once thought about factor more-clear out the wallet related code to build the possibility of another wallet-backend which could be switched by a bitcoind -arg
421 2014-12-22 14:36:43 <kanzure> hi jonasschnelli
422 2014-12-22 14:37:36 <jonasschnelli> Yes. support for bdb must be there forever. But if people could actively switch to another system (with manual wallet migration on script basis) it would be with less risks
423 2014-12-22 14:37:39 <wumpus> nah, a conversion utility that automatically gets called for old wallets would be fine
424 2014-12-22 14:37:41 <jonasschnelli> hi kanzure
425 2014-12-22 14:37:56 <midnightmagic> gmaxwell: how long did your ppc machine take to complete one libsec test run (and two for that matter?) I'm getting: real    0m21.336s for ./tests 2, and 0m14.133s for ./tests 1
426 2014-12-22 14:38:04 <midnightmagic> Is there another benchmark I should be attempting?
427 2014-12-22 14:40:28 <wumpus> the thing is, the current wallet *works* and there is very little interest in reviewing larger changes to it, see e.g. the large number of pulls by cozz
428 2014-12-22 14:41:12 <sipa> yeah :(
429 2014-12-22 14:41:47 <jonasschnelli> That's true. But who want's to stick forever with bdb and the wallet.dat?
430 2014-12-22 14:42:11 <sipa> this is also about myself... i just have too little interesting in the current wallet
431 2014-12-22 14:42:47 <wumpus> forever is a long time, but it could limp along for a long time in this way
432 2014-12-22 14:43:02 <wumpus> legacy software, basically
433 2014-12-22 14:43:11 <wumpus> people also still use COBOL and mainframes
434 2014-12-22 14:43:45 <jonasschnelli> COBOL is still everywhere when you walk in a Banks IT center... :)
435 2014-12-22 14:45:01 <jonasschnelli> Conclusion: leak is harmless, bdb removal is low-prio.
436 2014-12-22 14:45:05 <wumpus> sipa: so start a new one from scratch with jonasschnelli ;)
437 2014-12-22 14:45:17 <sipa> wumpus: if i ever have a few years time :)
438 2014-12-22 14:45:21 <wumpus> hehe
439 2014-12-22 14:45:26 <sipa> i probably will, but i don't like to promise
440 2014-12-22 14:45:41 <sipa> there are many other interesting things to do as well...
441 2014-12-22 14:46:25 <jonasschnelli> kanzure: you probably need to rebase 5524
442 2014-12-22 14:46:26 <sipa> jonasschnelli: github says you're in zurich?
443 2014-12-22 14:46:46 <jonasschnelli> Sipa: yes. Zurich an Basel.
444 2014-12-22 14:47:04 <sipa> have you attended the bitcoin meetups there?
445 2014-12-22 14:47:26 <jonasschnelli> No. Are you there sometimes?
446 2014-12-22 14:47:26 <sipa> i live in zurich too, though it'll probably be februari before i'm back there
447 2014-12-22 14:47:46 <sipa> i have, yes
448 2014-12-22 14:47:49 <jonasschnelli> sipa thought you live in be?
449 2014-12-22 14:48:10 <sipa> i'm belgian, but i moved to zurich for google, and now i'l still lingering there :)
450 2014-12-22 14:49:31 <jonasschnelli> Sipa: Zurich is a nice place to live! I try to once check out the meetup in zurich...
451 2014-12-22 14:49:35 <wumpus> jonasschnelli: well not so much low-prio, just that it requires a large investment of time to make any change to the current wallet, and there is so many that really needs to be changed and only few developers interested on working on it
452 2014-12-22 14:52:13 <wumpus> jonasschnelli: just because it doesn't rank high on my priority list doesn't mean that's the end of it, everyone determines their own priorities, if there's still enough interest we could assign someone as 'wallet maintainer' who coordinates the changes
453 2014-12-22 14:54:15 <wumpus> jonasschnelli: but my feeling has always been that that that's not the case, but I've been wrong before, and thing may change as they discover their favourite web or SPV walllet ate their coins
454 2014-12-22 14:54:21 <kanzure> jonasschnelli: yep that's incoming in a few
455 2014-12-22 14:55:51 <jonasschnelli> wumpus: Yes. Could be that after the next SPV or web wallet fiasco people looking for a save port...
456 2014-12-22 14:56:39 <jonasschnelli> wumpus: maybe i try to slowly factor out the wallet backend with reasonable and reviewable prices of code.
457 2014-12-22 14:56:43 <jonasschnelli> prices = pieces
458 2014-12-22 14:57:25 <sipa> jonasschnelli: http://www.meetup.com/Bitcoin-Meetup-Switzerland/
459 2014-12-22 14:59:34 <jonasschnelli> sipa: will join once... thanks
460 2014-12-22 15:03:30 <jonasschnelli> sipa: i've thought you work for blockstream? Why play around in the google building?
461 2014-12-22 15:04:29 <sipa> jonasschnelli: i left google for blockstream, but haven't moved elsewhere yet
462 2014-12-22 15:05:40 <sipa> i've lived in zurich for 2 years
463 2014-12-22 15:09:15 <jonasschnelli> sipa: so you can now talk swiss german and know how to swim in the limmat.
464 2014-12-22 15:15:10 <sipa> ha
465 2014-12-22 15:15:12 <sipa> gruezi
466 2014-12-22 15:28:38 <btcdrak> cfields I have found something minor with gitian. I dont know if it's the expected behaviour, but it's not taking into account Bitcoin minor version number in filename output. So 0.10.0.1 or 0.10.0.0 will generate the same files name as bitcoin-foo-0.10.0.tar.gz for both versions
467 2014-12-22 15:37:50 <wumpus> btcdrak: I think that's always been the case, filenames like foo-0.10.0.0.tar.gz are ugly, although it'd of course be possible to leave out the last .0 if it's 0 only
468 2014-12-22 15:39:34 <wumpus> we only released a x.y.z.1 version once
469 2014-12-22 16:29:57 <wumpus> jonasschnelli: a potential approach for the wallet would be to freeze the wallet code in master, then create a newwallet branch where you (and others interested) can work on refactoring and reworking the thing, independent from most work on the master branch
470 2014-12-22 16:57:36 <michagogo> 15:15:20 <wumpus> note that the guide tells you to change gitian's IP range, to avoid it overlapping with virtualbox's, maybe something went wrong there?
471 2014-12-22 16:57:38 <michagogo> 15:16:13 <wumpus> ie maybe gitian broke the setting
472 2014-12-22 16:58:06 <michagogo> Stuff like that is why I (a while back) found the one-liner to change VBox's range instead :-)
473 2014-12-22 16:58:45 <wumpus> michagogo: yes, you can change vbox's range too, shouldn't matter which one you change
474 2014-12-22 17:00:06 <michagogo> I figured it's better to just change vbox because of possibilities like this -- I've seen how fickle gitian is, so I figured it's probably safer to adapt the environment to it as much as possible, rather than the other way around
475 2014-12-22 17:00:46 <wumpus> feel free to update the guide to do that instead
476 2014-12-22 17:01:44 <wumpus> although it's a pity that you can't change the ip range from virtualbox' gui, so you need platform-specific descriptions of how to do that, ie no clue where vboxmanage is on windows
477 2014-12-22 17:02:34 <michagogo> "C:\Program Files\Oracle\VirtualBox\VBoxManage.exe"
478 2014-12-22 17:02:41 <wumpus> ok, add that to the guide then
479 2014-12-22 17:02:57 <michagogo> 
ACTION doesn't remember how he changed it

480 2014-12-22 17:03:00 <michagogo> 
ACTION looks

481 2014-12-22 17:16:13 <slick2> any valuable guide regarding data types used in bitcoin c++ code
482 2014-12-22 17:16:46 <helo> header files <3
483 2014-12-22 17:16:55 <michagogo> wumpus: actually
484 2014-12-22 17:17:05 <michagogo> Why not just set it to bridged networking?
485 2014-12-22 17:17:22 <michagogo> Oh, wait, you want to SSH into it.
486 2014-12-22 17:17:43 <michagogo> How about having one interface bridged and one interface host-only?
487 2014-12-22 17:17:55 <wumpus> michagogo: mehh, bridged sucks
488 2014-12-22 17:18:26 <wumpus> michagogo: requires setting the network interface to promiscous and such, and introduces a virtual new machine into the network you're on
489 2014-12-22 17:18:43 <wumpus> michagogo: that won't work in the general case, ie if you have a stricter network admin :)
490 2014-12-22 17:19:16 <wumpus> michagogo: also it requires more privileges on the local machine (ie to create a bridged interface)
491 2014-12-22 17:19:38 <wumpus> michagogo: inner-process NAT networking is nice
492 2014-12-22 17:20:29 <michagogo> hmm
493 2014-12-22 17:20:31 <wumpus> you can effectively have the whole VM run as a userspace process
494 2014-12-22 17:20:57 <michagogo> I just tried booting my Ubuntu VM for the first time since copying it over from my image
495 2014-12-22 17:21:15 <michagogo> error: invalid arch independent ELF magic.