1 2018-04-05 00:30:39	0|bitcoin-git|[13bitcoin] 15sipa pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/bfaed1ab2ec7...88430cbab4dc
  2 2018-04-05 00:30:40	0|bitcoin-git|13bitcoin/06master 141e747e3 15Mark Friedenbach: Make segwit failure due to CLEANSTACK violation return a SCRIPT_ERR_CLEANSTACK error code.
  3 2018-04-05 00:30:40	0|bitcoin-git|13bitcoin/06master 1488430cb 15Pieter Wuille: Merge #12167: Make segwit failure due to CLEANSTACK violation return a SCRIPT_ERR_CLEANSTACK error code...
  4 2018-04-05 00:31:15	0|bitcoin-git|[13bitcoin] 15sipa closed pull request #12167: Make segwit failure due to CLEANSTACK violation return a SCRIPT_ERR_CLEANSTACK error code (06master...06cleanstack-script-error) 02https://github.com/bitcoin/bitcoin/pull/12167
  5 2018-04-05 00:48:31	0|bitcoin-git|[13bitcoin] 15instagibbs opened pull request #12888: debug log number of unknown wallet records on load (06master...06unknownrec) 02https://github.com/bitcoin/bitcoin/pull/12888
  6 2018-04-05 00:48:39	0|bitcoin-git|[13bitcoin] 15MarcoFalke pushed 3 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/88430cbab4dc...9a2db3b3d511
  7 2018-04-05 00:48:40	0|bitcoin-git|13bitcoin/06master 149c92c8c 15John Newbery: [tests] Remove Comparison Test Framework
  8 2018-04-05 00:48:40	0|bitcoin-git|13bitcoin/06master 14e80c640 15John Newbery: [tests] Remove bip9-softforks.py...
  9 2018-04-05 00:48:41	0|bitcoin-git|13bitcoin/06master 149a2db3b 15MarcoFalke: Merge #11818: I accidentally [deliberately] killed it [the ComparisonTestFramework]...
 10 2018-04-05 00:48:59	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #11818: I accidentally [deliberately] killed it [the ComparisonTestFramework] (06master...06remove_comp_framework) 02https://github.com/bitcoin/bitcoin/pull/11818
 11 2018-04-05 01:32:32	0|bitcoin-git|[13bitcoin] 15Empact closed pull request #12812: Test ReadConfigFile (06master...06test-read-config-file) 02https://github.com/bitcoin/bitcoin/pull/12812
 12 2018-04-05 04:55:45	0|bitcoin-git|[13bitcoin] 15buddilla opened pull request #12889: doc: add qrencode to brew install instructions (#1) (06master...06master) 02https://github.com/bitcoin/bitcoin/pull/12889
 13 2018-04-05 04:57:35	0|bitcoin-git|[13bitcoin] 15buddilla closed pull request #12889: doc: add qrencode to brew install instructions (#1) (06master...06master) 02https://github.com/bitcoin/bitcoin/pull/12889
 14 2018-04-05 06:54:13	0|wumpus|<jnewbery> wumpus: mind if I take over #7729? <- no, please do
 15 2018-04-05 06:54:17	0|gribble|https://github.com/bitcoin/bitcoin/issues/7729 | rpc: introduce label API for wallet by laanwj · Pull Request #7729 · bitcoin/bitcoin · GitHub
 16 2018-04-05 06:54:32	0|wumpus|that PR is cursed :)
 17 2018-04-05 06:59:40	0|wumpus|aj: could have something to do with how boost test is optimized, from what I remember
 18 2018-04-05 07:00:08	0|wumpus|some of the tests use BOOST_EQUAL and such in inner loops, while they're somewhat heavyweight
 19 2018-04-05 07:00:29	0|wumpus|(IIRC, for example one of the verbosity options causes them to log everything)
 20 2018-04-05 07:03:21	0|wumpus|aj: you could use my methodology from #10026 (need to update it) to find if there are any tests that are especially show with gcc or clang
 21 2018-04-05 07:03:23	0|gribble|https://github.com/bitcoin/bitcoin/issues/10026 | Overview of slow unit tests · Issue #10026 · bitcoin/bitcoin · GitHub
 22 2018-04-05 07:04:27	0|aj|wumpus: well, i tried to debug it, but gcc started doing things as fast as clang so now i don't know
 23 2018-04-05 07:04:40	0|aj|wumpus: better than clang doing things as slow as gcc at least
 24 2018-04-05 07:04:42	0|wumpus|okay, might have been comparing a debug build against a release build
 25 2018-04-05 07:04:55	0|wumpus|perf differences are brutal between those
 26 2018-04-05 07:05:34	0|aj|wumpus: oh, could have been -- do you know which bit is the difference there?
 27 2018-04-05 07:06:10	0|wumpus|lower optimization level as well as the mutex checking stuff overhead
 28 2018-04-05 07:06:58	0|aj|wumpus: oh... if anything it seemed that --with-debug was going faster :-/
 29 2018-04-05 07:13:31	0|bitcoin-git|13bitcoin/06master 145b10ab0 15John Newbery: [trivial] Add newlines to end of log messages....
 30 2018-04-05 07:13:31	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/9a2db3b3d511...2fc94370f510
 31 2018-04-05 07:13:32	0|bitcoin-git|13bitcoin/06master 142fc9437 15Wladimir J. van der Laan: Merge #12887: [trivial] Add newlines to end of log messages....
 32 2018-04-05 07:14:21	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #12887: [trivial] Add newlines to end of log messages. (06master...06log_messages_newlines) 02https://github.com/bitcoin/bitcoin/pull/12887
 33 2018-04-05 07:33:13	0|bitcoin-git|13bitcoin/06master 14a5bca13 15Luke Dashjr: Bugfix: Include <memory> for std::unique_ptr
 34 2018-04-05 07:33:13	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/2fc94370f510...bd59c4395c50
 35 2018-04-05 07:33:14	0|bitcoin-git|13bitcoin/06master 14bd59c43 15Wladimir J. van der Laan: Merge #12859: Bugfix: Include <memory> for std::unique_ptr...
 36 2018-04-05 07:34:00	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #12859: Bugfix: Include <memory> for std::unique_ptr (06master...06incl_memory) 02https://github.com/bitcoin/bitcoin/pull/12859
 37 2018-04-05 07:54:28	0|bitcoin-git|13bitcoin/06master 14a5263fb 15Indospace.io: doc: Use bitcoind in Tor documentation
 38 2018-04-05 07:54:28	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/bd59c4395c50...2b54155a459c
 39 2018-04-05 07:54:29	0|bitcoin-git|13bitcoin/06master 142b54155 15Wladimir J. van der Laan: Merge #12877: doc: Use bitcoind in Tor documentation...
 40 2018-04-05 07:55:11	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #12877: doc: Use bitcoind in Tor documentation (06master...06patch-1) 02https://github.com/bitcoin/bitcoin/pull/12877
 41 2018-04-05 12:01:10	0|mryandao|is travis being flaky again?
 42 2018-04-05 12:16:29	0|wumpus|what's the problem?
 43 2018-04-05 12:21:40	0|mryandao|#12240 failed due to a timeout for 2
 44 2018-04-05 12:21:43	0|gribble|https://github.com/bitcoin/bitcoin/issues/12240 | [rpc] Introduced a new `fees` structure that aggregates all sub-field fee types denominated in BTC by mryandao · Pull Request #12240 · bitcoin/bitcoin · GitHub
 45 2018-04-05 12:22:00	0|mryandao|"The job exceeded the maximum time limit for jobs, and has been terminated."
 46 2018-04-05 12:22:13	0|wumpus|fantastic
 47 2018-04-05 12:22:36	0|mryandao|¯\_ (ツ) _/¯
 48 2018-04-05 12:23:10	0|wumpus|I've restarted job 2
 49 2018-04-05 12:23:30	0|aj|i've been trying out the 32bit windows travis checks with more logging; seems like when it succeeds, the coinselection tests take about 15m. i think i'm going to have one fail now, and it's taking ages to compile before even getting to the tests
 50 2018-04-05 12:23:39	0|mryandao|cant the bitcoin core project afford dedicated machines with gitlab ci?
 51 2018-04-05 12:23:44	0|aj|https://travis-ci.org/ajtowns/bitcoin/builds/362589978
 52 2018-04-05 12:25:42	0|aj|wallet/test/coinselector_tests.cpp(17): Leaving test suite "coinselector_tests"; testing time: 950408ms
 53 2018-04-05 12:27:24	0|wumpus|mryandao: it's more of a problem of people than resources, no one really wants to babysit servers. There used to be a "pulltester" running some CI script on a VM, switching to travis was great. But yeah lately...
 54 2018-04-05 12:29:54	0|wumpus|I think cfields tried pretty much everything except waving a wad of money in travis representative's face to get more capacity, but no luck
 55 2018-04-05 12:30:06	0|aj|make is taking 18s on one host, but 822s on the other
 56 2018-04-05 12:30:17	0|jonasschnelli|there are also the nightly gitian builds I run... though not a real CI (runs no tests)
 57 2018-04-05 12:30:36	0|jonasschnelli|ideally the gitian builder would test the binaries in some VMs
 58 2018-04-05 12:30:42	0|aj|(make all i mean, not make check)
 59 2018-04-05 12:45:14	0|wumpus|jonasschnelli: yes, that would be nice, esp for targets not covered by travis at the moment
 60 2018-04-05 12:45:58	0|wumpus|aj: yes, compile speed varies a lot between platforms - I suppose memory amount and speed plays a large role, apart from CPU speed
 61 2018-04-05 12:46:41	0|aj|wumpus: this is the same platform though, i686-w64-mingw32 with DPKG_ADD_ARCH i386
 62 2018-04-05 12:48:19	0|wumpus|wait, your'e building *on* the same platform or *for* the same platform
 63 2018-04-05 12:48:35	0|wumpus|or both?
 64 2018-04-05 12:49:03	0|aj|mostly difference in compile time rather than test time though -- 13m for make all, 20m for test, 11m for make depends, and that's almost the timeout
 65 2018-04-05 12:49:42	0|aj|wumpus: same job twice, i just rearranged the arg order so it didn't get deduped
 66 2018-04-05 12:49:56	0|wumpus|ok
 67 2018-04-05 12:50:24	0|aj|with make check replaced by test_bitcoin.exe -l test_suite or so
 68 2018-04-05 12:51:13	0|wumpus|your results confuse me a lot
 69 2018-04-05 12:51:22	0|aj|:)
 70 2018-04-05 12:51:40	0|aj|i guess it means some of their w64 hosts are overloaded/slow/broken?
 71 2018-04-05 12:52:26	0|aj|(and also, our coinselector_tests are a bit slow on w64 for some reason? 15 mins seems crazy)
 72 2018-04-05 12:54:40	0|sipa|that is crazy
 73 2018-04-05 12:54:48	0|sipa|how fast are they on i686?
 74 2018-04-05 12:55:25	0|wumpus|so I would guess that's due to boost::test, doing BOOST_EQUALS etc in an inner loop
 75 2018-04-05 12:56:59	0|aj|haven't tried on other arches; running now as https://travis-ci.org/ajtowns/bitcoin/builds/362613881
 76 2018-04-05 13:13:14	0|sipa|wumpus: ?
 77 2018-04-05 13:13:26	0|wumpus|sipa: hm?
 78 2018-04-05 13:14:01	0|wumpus|oh I can check how long the test takes, if you mean that
 79 2018-04-05 13:15:00	0|sipa|ah, can we move the inner loop checks out?
 80 2018-04-05 13:15:42	0|sipa|accumulate a set of failure strings, and thwn assert that it is empty (with the set as message?
 81 2018-04-05 13:16:45	0|wumpus|yes, that would be a good option
 82 2018-04-05 13:17:42	0|jnewbery|wumpus: great, I'll have a shot at uncursing it
 83 2018-04-05 13:20:15	0|wumpus|coinselector_tests takes 10.5 seconds here locally
 84 2018-04-05 13:21:10	0|wumpus|(linux, x86_64)
 85 2018-04-05 13:21:44	0|aj|sigh. "test_bitcoin.exe" doesn't work well on non windows funnily enough
 86 2018-04-05 13:22:41	0|wumpus|what's the problem?
 87 2018-04-05 13:22:55	0|aj|linux doesn't add ".exe" :)
 88 2018-04-05 13:24:24	0|wumpus|#8650 "unit tests can be really slow under wine because BOOST_CHECK logs something for all tests. This patch makes them faster by only logging tests which fail. PR'd an alternative to #8632."
 89 2018-04-05 13:24:26	0|gribble|https://github.com/bitcoin/bitcoin/issues/8650 | Make tests much faster by replacing BOOST_CHECK with FAST_CHECK by JeremyRubin · Pull Request #8650 · bitcoin/bitcoin · GitHub
 90 2018-04-05 13:24:27	0|gribble|https://github.com/bitcoin/bitcoin/issues/8632 | Speed up prevector tests by parallelization by JeremyRubin · Pull Request #8632 · bitcoin/bitcoin · GitHub
 91 2018-04-05 13:24:59	0|wumpus|same problem, but that solution was kind of over the top as it replaces BOOST_CHECK completely, everywhere
 92 2018-04-05 13:25:49	0|wumpus|so in the end we just sped up the prevector tests instead - coinselector tests likely have the same problem
 93 2018-04-05 13:41:29	0|wumpus|aj: should still work if wine is installed
 94 2018-04-05 14:09:15	0|sipa|wumpus: are you sure it's due to coin selection?
 95 2018-04-05 14:09:32	0|sipa|i'm looking at the total number of assertions in every test, and coin selection is tiny
 96 2018-04-05 14:09:48	0|sipa|it has 7022 assertions out of 4850320
 97 2018-04-05 14:09:49	0|wumpus|sipa: that's what aj said - okay in that case it's another thing that is slow in wine
 98 2018-04-05 14:10:31	0|sipa|coins_tests/updatecoins_simulation_test has 1223073 assertions
 99 2018-04-05 14:11:50	0|wumpus|whoa
100 2018-04-05 14:12:03	0|wumpus|aj ^^
101 2018-04-05 14:13:51	0|sipa|what is the command used to actually run the test for windows on travis?
102 2018-04-05 14:14:02	0|sipa|wine src/bitcoin/test_bitcoin.exe ?
103 2018-04-05 14:18:14	0|wumpus|just 'make check', which will do that indirectly, I expect
104 2018-04-05 14:18:44	0|bitcoin-git|[13bitcoin] 15sipa opened pull request #12890: [DEBUG TRAVIS] Detailed unit test report for win32 (06master...06201803_debugtravis) 02https://github.com/bitcoin/bitcoin/pull/12890
105 2018-04-05 14:19:08	0|sipa|yes but i want to modify the arguments it's being called with
106 2018-04-05 14:19:19	0|bitcoin-git|[13bitcoin] 15jnewbery opened pull request #12891:  [logging] add lint-logs.sh to check for newline termination. (06master...06log_lint) 02https://github.com/bitcoin/bitcoin/pull/12891
107 2018-04-05 14:20:29	0|aj|sipa: i've been running "travis_wait 50 src/test/test_bitcoin.exe -l test_suite"
108 2018-04-05 14:22:42	0|aj|wallet/test/coinselector_tests.cpp(17): Leaving test suite "coinselector_tests"; testing time: 21124667us   # us not ms, so just 21s on x86-64 with qt5
109 2018-04-05 14:28:16	0|aj|without the wallet compiled in, all the tests seem to take 33,566,015us so 33s, which seems fine
110 2018-04-05 14:28:39	0|aj|and the qt4 test times are reported with units of "mks" so i don't know what that means
111 2018-04-05 14:28:50	0|aj|https://travis-ci.org/ajtowns/bitcoin/builds/362625669
112 2018-04-05 14:30:33	0|wumpus|https://stackoverflow.com/questions/28970229/what-is-the-mks-unit-reported-by-the-boost-unit-test-framework
113 2018-04-05 14:30:45	0|wumpus|apparently, the Russian abbreviation for millisecond...
114 2018-04-05 14:32:04	0|ryanofsky|wumpus, had been wondering if you think #10244 is ready, pinged you here: https://github.com/bitcoin/bitcoin/pull/10244#issuecomment-378357905
115 2018-04-05 14:32:08	0|gribble|https://github.com/bitcoin/bitcoin/issues/10244 | Refactor: separate gui from wallet and node by ryanofsky · Pull Request #10244 · bitcoin/bitcoin · GitHub
116 2018-04-05 14:33:19	0|wumpus|... or microsecond (sorry, the answer is unclear too)
117 2018-04-05 14:34:05	0|wumpus|ryanofsky: looks like it!
118 2018-04-05 14:34:19	0|aj|sipa: might be faster to build with mingw32 locally, and just include test_bitcoin.exe in the debug commit
119 2018-04-05 14:35:00	0|sipa|?
120 2018-04-05 14:36:04	0|aj|sipa: as opposed to actually compiling test_bitcoin.exe on travis, just include it in the PR and run it on travis
121 2018-04-05 14:38:04	0|sipa|then i have to make sure the binary is compatible etc
122 2018-04-05 14:39:20	0|sipa|hmm, i guess that's not an issue for win binaries
123 2018-04-05 14:42:48	0|wumpus|it's indeed not an issue; the windows binaries are self-contained except for OS libs
124 2018-04-05 14:43:59	0|sipa|compilation is very slow on travis; don't we use ccache across runs or something?
125 2018-04-05 14:45:50	0|wumpus|it should ($HOME/.ccache is among the travis-cached directories)
126 2018-04-05 14:54:44	0|sipa|aj: any other explanations about what could make the coin selection test so slow on windows (relatively to other platforms)?
127 2018-04-05 14:55:20	0|aj|sipa: i wonder if it's hitting a slow interpretation path in wine somehow?
128 2018-04-05 14:59:41	0|wumpus|aj: the way to find out would be to try on real windows
129 2018-04-05 15:00:14	0|wumpus|if it's also slow there, gcc-mingw64 is the problem, if not, it's wine
130 2018-04-05 16:19:51	0|bitcoin-git|[13bitcoin] 15laanwj pushed 22 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/2b54155a459c...5f0c6a7b0e47
131 2018-04-05 16:19:52	0|bitcoin-git|13bitcoin/06master 1471e0d90 15Russell Yanofsky: Remove direct bitcoin calls from qt/bitcoin.cpp
132 2018-04-05 16:19:52	0|bitcoin-git|13bitcoin/06master 14ea73b84 15Russell Yanofsky: Add src/interface/README.md
133 2018-04-05 16:19:53	0|bitcoin-git|13bitcoin/06master 14c0f2756 15Russell Yanofsky: Remove direct bitcoin calls from qt/optionsmodel.cpp
134 2018-04-05 16:20:05	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10244: Refactor: separate gui from wallet and node (06master...06pr/ipc-local) 02https://github.com/bitcoin/bitcoin/pull/10244
135 2018-04-05 16:49:52	0|BlueMatt|so we now have 4 pending PRs which rewrite CValidationState, and all of them stomp on each others' toes significantly
136 2018-04-05 16:50:36	0|BlueMatt|should probably pick one and get it merged asap instead of waiting until someone *else* decides they also want to rewrite CValidationState
137 2018-04-05 16:51:22	0|instagibbs|BlueMatt, clearly we need a standard on how to rewrite CValidationState
138 2018-04-05 16:52:11	0|BlueMatt|nono, we need someone else to sit down and *write* a standard, and then a new PR for it
139 2018-04-05 16:53:31	0|wumpus|we should get all the people that want to rewrite CValidationState into an (IRC) room and get them to agree what their goal is
140 2018-04-05 16:53:43	0|sipa|can we please create a BCIP process first?
141 2018-04-05 16:53:57	0|sipa|"bitcoin core implementation proposal"
142 2018-04-05 16:54:19	0|wumpus|definitely
143 2018-04-05 16:54:59	0|bitcoin-git|[13bitcoin] 15jl2012 closed pull request #8654: [WIP] Reuse sighash computations across evaluation (06master...06sighashcache) 02https://github.com/bitcoin/bitcoin/pull/8654
144 2018-04-05 16:55:05	0|BlueMatt|wumpus: well it looks like Empact backed out the changes in #12463 that essentially made it into #11639, and #11523 needs rebase pretty significantly (and I'm not a huge fan of parts of it, hence #11639)
145 2018-04-05 16:55:07	0|gribble|https://github.com/bitcoin/bitcoin/issues/12463 | Drop the return and corruptionPossible arguments from CValidationState::DoS, and rename to ::Reject by Empact · Pull Request #12463 · bitcoin/bitcoin · GitHub
146 2018-04-05 16:55:09	0|gribble|https://github.com/bitcoin/bitcoin/issues/11639 | Rewrite the interface between validation and net_processing wrt DoS by TheBlueMatt · Pull Request #11639 · bitcoin/bitcoin · GitHub
147 2018-04-05 16:55:09	0|wumpus|write four BCIPs, then randomly copy/paste from them into one, which is used for the implementation guideline
148 2018-04-05 16:55:10	0|gribble|https://github.com/bitcoin/bitcoin/issues/11523 | [Refactor] CValidation State by JeremyRubin · Pull Request #11523 · bitcoin/bitcoin · GitHub
149 2018-04-05 16:55:12	0|gribble|https://github.com/bitcoin/bitcoin/issues/11639 | Rewrite the interface between validation and net_processing wrt DoS by TheBlueMatt · Pull Request #11639 · bitcoin/bitcoin · GitHub
150 2018-04-05 16:55:33	0|BlueMatt|but #11639 has been open for months and gotten no review except for suhas
151 2018-04-05 16:55:35	0|gribble|https://github.com/bitcoin/bitcoin/issues/11639 | Rewrite the interface between validation and net_processing wrt DoS by TheBlueMatt · Pull Request #11639 · bitcoin/bitcoin · GitHub
152 2018-04-05 16:55:54	0|sipa|i'll review 11639
153 2018-04-05 16:56:00	0|wumpus|would make sense to add it to high priority then, I guess
154 2018-04-05 16:56:10	0|sipa|would it help if i concept ack it quickly?
155 2018-04-05 16:56:21	0|wumpus|it can't hurt at least
156 2018-04-05 16:56:32	0|BlueMatt|it would, yes
157 2018-04-05 16:56:48	0|BlueMatt|I can rebase it, but the rebase-needed parts are pretty much in the scripted-diffs, so whatever
158 2018-04-05 16:56:50	0|wumpus|for an interface change concepts ack are important
159 2018-04-05 16:56:59	0|sipa|if it's clear that that is the approach we want to take, i think we should prioritize the larger-refactor over the patch-up small prs
160 2018-04-05 16:57:17	0|wumpus|right
161 2018-04-05 16:57:33	0|midnightmagic|'/w 2
162 2018-04-05 16:57:39	0|midnightmagic|woops, sorry.
163 2018-04-05 16:57:42	0|sipa|unless the larger refactor will take so long regardless that it's better to have small fixes actually in master
164 2018-04-05 17:28:38	0|bitcoin-git|[13bitcoin] 15jnewbery opened pull request #12892: [wallet] [rpc] introduce 'label' API for wallet (06master...067729_jnewbery) 02https://github.com/bitcoin/bitcoin/pull/12892
165 2018-04-05 17:46:51	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #7729: rpc: introduce 'label' API for wallet (06master...062016_03_wallet_label_api) 02https://github.com/bitcoin/bitcoin/pull/7729
166 2018-04-05 18:00:58	0|jonasschnelli|meeting?
167 2018-04-05 18:01:11	0|wumpus|I think in an hour
168 2018-04-05 18:01:28	0|jonasschnelli|Oh. I see.
169 2018-04-05 18:02:40	0|BlueMatt|Last chance to get reviews in for last week's high-priority list before you get shamed at meeting!
170 2018-04-05 18:10:36	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #12873: [ci] Run functional tests using bitcoin-qt in one Travis job (06master...062018-04-03-travis-func-qt) 02https://github.com/bitcoin/bitcoin/pull/12873
171 2018-04-05 18:36:08	0|MarcoFalke|> jonasschnelli ideally the gitian builder would test the binaries in some VMs
172 2018-04-05 18:36:20	0|MarcoFalke|I do, if you fix the osx build ;)
173 2018-04-05 18:36:42	0|MarcoFalke|windows is WIP
174 2018-04-05 18:37:30	0|jonasschnelli|MarcoFalke: is it broken in general or only on my builder (haven't checked)?
175 2018-04-05 18:37:42	0|MarcoFalke|jonasschnelli: On your box
176 2018-04-05 18:37:49	0|MarcoFalke|There is some network timeout to aws
177 2018-04-05 18:38:03	0|MarcoFalke|fetchin a dependency that was recently bumped
178 2018-04-05 18:39:09	0|jonasschnelli|I'll have a look.
179 2018-04-05 18:39:22	0|MarcoFalke|Awesome, thx
180 2018-04-05 18:39:45	0|jonasschnelli|Currently traveling back to CH and back in my office next week
181 2018-04-05 18:40:26	0|MarcoFalke|Cool. Wir haben uns schon gefragt wo du so lange bleibst, heh
182 2018-04-05 18:42:28	0|MarcoFalke|suggested topics for meeting: Slow unit test/Run unit tests in parallel
183 2018-04-05 18:45:38	0|wumpus|good idea
184 2018-04-05 18:45:52	0|sipa|yep
185 2018-04-05 18:46:04	0|sipa|jonasschnelli: where were you?
186 2018-04-05 18:52:46	0|jonasschnelli|sipa: I was 2 month in Indonesia... meeting was 3am... Internet speed was mostly 56kbit (shared with a couple of other people)
187 2018-04-05 18:55:18	0|sipa|haha
188 2018-04-05 18:55:38	0|sipa|BlueMatt: by "air" you mean "err" ?
189 2018-04-05 18:56:02	0|BlueMatt|no? I breathe air
190 2018-04-05 18:56:47	0|BlueMatt|oh, you mean on github, no, I mean air, as in "Matt doesn't know AIR-glish"
191 2018-04-05 18:58:38	0|sipa|i preferr to err on the side of breathing air over earth
192 2018-04-05 18:59:58	0|sipa|*BOOM*
193 2018-04-05 19:00:11	0|lightningbot|Meeting started Thu Apr  5 19:00:20 2018 UTC.  The chair is wumpus. Information about MeetBot at http://wiki.debian.org/MeetBot.
194 2018-04-05 19:00:11	0|lightningbot|Useful Commands: #action #agreed #help #info #idea #link #topic.
195 2018-04-05 19:00:11	0|wumpus|#startmeeting
196 2018-04-05 19:00:15	0|wumpus|#bitcoin-core-dev Meeting: wumpus sipa gmaxwell jonasschnelli morcos luke-jr btcdrak sdaftuar jtimon cfields petertodd kanzure bluematt instagibbs phantomcircuit codeshark michagogo marcofalke paveljanik NicolasDorier jl2012 achow101 meshcollider jnewbery maaku fanquake promag provoostenator
197 2018-04-05 19:00:17	0|jonasschnelli|hi
198 2018-04-05 19:00:20	0|sipa|hi
199 2018-04-05 19:00:22	0|Murch|hi
200 2018-04-05 19:00:26	0|BlueMatt|This week's high-priority-for-review stats: 11857 got a few rounds of review (me, ryanofsky and sjors), 12560 went horribly under-reviewed (with only two comments from me and one from jimpo this week, no real reviews!), 11775 got one round of review from jimpo (which I missed until today, sorry about that!). MVP: jimpo. Overall rating: Needs Significant Improvement (for everyone except jimpo, for jimpo: Good Job!).
201 2018-04-05 19:00:26	0|jamesob|yo
202 2018-04-05 19:00:27	0|phantomcircuit|wat
203 2018-04-05 19:00:30	0|wumpus|hi
204 2018-04-05 19:00:54	0|meshcollider|Hi
205 2018-04-05 19:00:59	0|wumpus|#topic high priority for review
206 2018-04-05 19:01:09	0|cfields|hi
207 2018-04-05 19:02:21	0|BlueMatt|I mean no point nominating new things if its not gonna get any additional review. Might as well just directly ping jimpo and ask him to take a look.
208 2018-04-05 19:02:27	0|wumpus|well, I guess we need to keep it at the current list then, if the current ones don't get enough review we certainly shouldn't add more :)
209 2018-04-05 19:02:30	0|jnewbery|hi
210 2018-04-05 19:02:48	0|jamesob|This PR fixes a null pointer deref that's currently in master: https://github.com/bitcoin/bitcoin/pull/12836
211 2018-04-05 19:03:07	0|MarcoFalke|^ Needs rebase
212 2018-04-05 19:03:21	0|achow101|hi
213 2018-04-05 19:04:08	0|wumpus|if something fixes an important issue such as a null pointer dereference (an existing one, not a potential one), please mention that in the PR title!
214 2018-04-05 19:04:21	0|instagibbs|achow101, if you rebase psbt I'd nominate it for high prio, not sure you have the time to carry it right now
215 2018-04-05 19:04:32	0|wumpus|"Make WalletInitInterface and DummyWalletInit private" really doesn't communicate that
216 2018-04-05 19:04:42	0|MarcoFalke|Also, those fixes should go in without having them to put on high-prio
217 2018-04-05 19:04:44	0|achow101|instagibbs: I'll try to do that later today or tomorrow
218 2018-04-05 19:04:57	0|wumpus|yes, apart from needing rebase it seems to have enough review to go in
219 2018-04-05 19:05:09	0|jonasschnelli|indeed
220 2018-04-05 19:05:20	0|meshcollider|wumpus: maybe he wasn't aware it fixed that
221 2018-04-05 19:05:32	0|wumpus|but please, don't hide fixes in refactor PRs
222 2018-04-05 19:05:34	0|kanzure|hi.
223 2018-04-05 19:05:36	0|wumpus|meshcollider: right , okay
224 2018-04-05 19:06:09	0|jnewbery|I think he wasn't aware of the bug that he fixed when he opened the PR
225 2018-04-05 19:06:48	0|wumpus|I see MarcoFalke already improved the title
226 2018-04-05 19:07:07	0|wumpus|#topic Slow unit test/Run unit tests in parallel
227 2018-04-05 19:07:08	0|cfields|jimpo: thanks for the reviews
228 2018-04-05 19:07:40	0|MarcoFalke|I thought that running the unit tests in parallel (similar to how the functional tests are run in parallel) is a free win
229 2018-04-05 19:07:54	0|jonasschnelli|MarcoFalke: is that possible with boost?
230 2018-04-05 19:07:58	0|MarcoFalke|Seems like they can be parallelized even on a single core
231 2018-04-05 19:08:08	0|jonasschnelli|https://www.boost.org/doc/libs/1_57_0/libs/test/doc/html/open-issues.html
232 2018-04-05 19:08:09	0|BlueMatt|yea, most of them use our globals in them
233 2018-04-05 19:08:10	0|cfields|MarcoFalke: i must admit, I kinda grumbled looking at your PR. Seems like it's really just a huge failure of the boost framework
234 2018-04-05 19:08:17	0|BlueMatt|we're a *long way* off from being able to do that, no?
235 2018-04-05 19:08:17	0|MarcoFalke|jonasschnelli: I adapted the google parallel tests wrapper
236 2018-04-05 19:08:28	0|wumpus|I hope it won't cause any ugly race conditions and such
237 2018-04-05 19:08:35	0|wumpus|we have so many intermittent travis failures as is :/
238 2018-04-05 19:08:38	0|MarcoFalke|BlueMatt: It works for me
239 2018-04-05 19:08:42	0|wumpus|at this point I'd prefer more stable tests to faster ones
240 2018-04-05 19:08:43	0|jonasschnelli|AFAIK boost test can't be run in parallel...
241 2018-04-05 19:08:44	0|MarcoFalke|at least locally
242 2018-04-05 19:09:08	0|MarcoFalke|You spin up different processes of course
243 2018-04-05 19:09:11	0|BlueMatt|oh, sorry, i didnt realize they were separate processes, was thinking no way in hell separate threads works
244 2018-04-05 19:09:16	0|wumpus|ohh smart
245 2018-04-05 19:09:18	0|cfields|jonasschnelli: iirc MarcoFalke's PR creates a wrapper that runs them individually, in parallel
246 2018-04-05 19:09:25	0|jnewbery|> I'd prefer more stable tests to faster ones
247 2018-04-05 19:09:31	0|jonasschnelli|PR #?
248 2018-04-05 19:09:34	0|MarcoFalke|like test_bitcoin -t wallet/t1 & test_bitcoin -t wallet/t2
249 2018-04-05 19:09:39	0|jnewbery|We need faster too! Travis PR builds are timing out all over the place
250 2018-04-05 19:10:06	0|MarcoFalke|jnewbery: that is a wine issue. Not sure if we can do much about it
251 2018-04-05 19:10:12	0|jonasschnelli|Yes. The amount of tests we added during the last year made SAS CI pretty hard
252 2018-04-05 19:10:20	0|MarcoFalke|I looked to realize I know not enough of wine to be of any use
253 2018-04-05 19:10:22	0|jamesob|not to mention the Travis backlog has been pretty deep lately
254 2018-04-05 19:10:25	0|wumpus|jnewbery: I was afraid of some race condition fest, but he spawns multiple processes, so that concern is gone
255 2018-04-05 19:10:46	0|achow101|what pr number?
256 2018-04-05 19:10:58	0|cfields|#12831
257 2018-04-05 19:10:58	0|wumpus|#12831
258 2018-04-05 19:11:00	0|gribble|https://github.com/bitcoin/bitcoin/issues/12831 | [WIP] Run unit tests in parallel by MarcoFalke · Pull Request #12831 · bitcoin/bitcoin · GitHub
259 2018-04-05 19:11:02	0|gribble|https://github.com/bitcoin/bitcoin/issues/12831 | [WIP] Run unit tests in parallel by MarcoFalke · Pull Request #12831 · bitcoin/bitcoin · GitHub
260 2018-04-05 19:11:09	0|MarcoFalke|Oh Chaincode Labs is willing to sponsor us additional 10 jobs for travis
261 2018-04-05 19:11:23	0|MarcoFalke|I hope that goes through until next week
262 2018-04-05 19:11:28	0|jonasschnelli|MarcoFalke: nice!
263 2018-04-05 19:11:29	0|sipa|pulling in parallel seems like huge overkill though
264 2018-04-05 19:11:48	0|jonasschnelli|That's what I just thought
265 2018-04-05 19:12:03	0|wumpus|MarcoFalke: nice, but does offering travis more money help? afaik what cfields said, it doens't
266 2018-04-05 19:12:20	0|cfields|didn't jeremy start on a replacement for boost_test at one point?
267 2018-04-05 19:12:27	0|cfields|yes, I've tried before, but by all means try again!
268 2018-04-05 19:12:27	0|wumpus|yes...
269 2018-04-05 19:12:29	0|MarcoFalke|wumpus: No, it will only increase the number of jobs
270 2018-04-05 19:12:36	0|MarcoFalke|So it clears the backlog faster
271 2018-04-05 19:12:49	0|jonasschnelli|wumpus: I guess money means going away from free OS travis to private support which seems to be slower even if you pay a lot? I may be wrong though.
272 2018-04-05 19:12:49	0|MarcoFalke|It doesn't increase the quality or anything
273 2018-04-05 19:13:08	0|MarcoFalke|jonasschnelli: No it is a out-of-band update
274 2018-04-05 19:13:21	0|wumpus|#8650
275 2018-04-05 19:13:23	0|gribble|https://github.com/bitcoin/bitcoin/issues/8650 | Make tests much faster by replacing BOOST_CHECK with FAST_CHECK by JeremyRubin · Pull Request #8650 · bitcoin/bitcoin · GitHub
276 2018-04-05 19:13:27	0|jonasschnelli|MarcoFalke: Okay. Good to know.
277 2018-04-05 19:13:33	0|jamesob|$$$ = more parallelism at the travis job level
278 2018-04-05 19:13:38	0|cfields|MarcoFalke: the issue that we had before is that they had no option for extra-paid open-source projects. Just paid and free. Maybe that's changed recently?
279 2018-04-05 19:13:55	0|MarcoFalke|cfields: I contacted the support
280 2018-04-05 19:14:05	0|MarcoFalke|They don't have anything listed on the public website/plans
281 2018-04-05 19:14:25	0|cfields|MarcoFalke: huh. I guess it's new then. Great :)
282 2018-04-05 19:14:41	0|MarcoFalke|apache or someone did it a few years ago, I am just trying the same
283 2018-04-05 19:14:50	0|meshcollider|Cool :)
284 2018-04-05 19:14:54	0|wumpus|great
285 2018-04-05 19:14:57	0|phantomcircuit|jamesob, i seem to remember the threshold for payed support being better than the free support for oss being pretty high
286 2018-04-05 19:14:59	0|sipa|hell yes, go for it
287 2018-04-05 19:15:17	0|jonasschnelli|8650 looks after a huge win.
288 2018-04-05 19:15:21	0|MarcoFalke|Doing a wholesale replacement of the test framework seems not a short term solution and perpendicualr to running the tests in parallel
289 2018-04-05 19:15:41	0|jtimon|thanks chaincode for the travis jobs!
290 2018-04-05 19:15:49	0|wumpus|jtimon: +1
291 2018-04-05 19:16:15	0|MarcoFalke|8650 seems like WIP
292 2018-04-05 19:16:17	0|wumpus|MarcoFalke: agree, would be a longer-term concern, if it can be done with boost test that's preferable
293 2018-04-05 19:16:39	0|cfields|MarcoFalke: I only mentioned it because it'll probably be done at some point anyway. And if so, we'd want to write it with parallelism in mind.
294 2018-04-05 19:16:45	0|wumpus|for now at least
295 2018-04-05 19:16:50	0|wumpus|8650 loses boost test features
296 2018-04-05 19:16:57	0|sipa|MarcoFalke: i can't believe that what we need from parallel can't be done with 20 lines of bash
297 2018-04-05 19:17:02	0|wumpus|e.g. logging what values mismatch
298 2018-04-05 19:17:25	0|wumpus|sipa: yes - just list the test suites, then distribute them over processes
299 2018-04-05 19:17:31	0|jonasschnelli|agree
300 2018-04-05 19:17:49	0|MarcoFalke|I can't write bash, so someone else has to volunteer
301 2018-04-05 19:17:51	0|wumpus|sounds fairly doable in bash, or at least python
302 2018-04-05 19:17:58	0|sipa|MarcoFalke: i hereby volunteer
303 2018-04-05 19:18:04	0|MarcoFalke|the current thing is python
304 2018-04-05 19:18:09	0|aj|20 lines of python sounds preferable...
305 2018-04-05 19:18:14	0|wumpus|python is preferable to me
306 2018-04-05 19:18:18	0|wumpus|at least I can help review and maintain it
307 2018-04-05 19:18:19	0|BlueMatt|ugh, y'all bash-haters
308 2018-04-05 19:18:24	0|MarcoFalke|aj: It has nice features such as a cache for the run times
309 2018-04-05 19:18:24	0|sipa|aj hereby volunteered :p
310 2018-04-05 19:18:43	0|MarcoFalke|sot the sorting would be done automatically and based on your specs
311 2018-04-05 19:19:18	0|sipa|MarcoFalke: ok, 22 lines of bash :)
312 2018-04-05 19:19:25	0|jonasschnelli|IMO the whole testing system is already pretty complex. I wouldn't set the burden higher
313 2018-04-05 19:19:33	0|MarcoFalke|sipa: Pull requests very welcome :)
314 2018-04-05 19:19:41	0|sipa|anyway, i'll see what i can do
315 2018-04-05 19:19:56	0|wumpus|ok, so we should look at 12831
316 2018-04-05 19:20:09	0|MarcoFalke|And the one sipa proposes
317 2018-04-05 19:20:09	0|sipa|#12831
318 2018-04-05 19:20:12	0|gribble|https://github.com/bitcoin/bitcoin/issues/12831 | [WIP] Run unit tests in parallel by MarcoFalke · Pull Request #12831 · bitcoin/bitcoin · GitHub
319 2018-04-05 19:20:19	0|MarcoFalke|#?????
320 2018-04-05 19:20:25	0|MarcoFalke|tba
321 2018-04-05 19:20:32	0|jamesob|at what grain does 12831 do parallelism? per file? boost test case?
322 2018-04-05 19:20:43	0|MarcoFalke|jamesob: Whatever you like
323 2018-04-05 19:20:43	0|sipa|jamesob: one test case per process
324 2018-04-05 19:20:49	0|MarcoFalke|Currently ^
325 2018-04-05 19:20:49	0|wumpus|per test suite, which is the only parallelism that makes sense
326 2018-04-05 19:21:28	0|jonasschnelli|I guess finer (case) would result in concurrency issue
327 2018-04-05 19:21:34	0|sipa|no...
328 2018-04-05 19:21:38	0|MarcoFalke|^
329 2018-04-05 19:21:47	0|sipa|they're all in separate processes
330 2018-04-05 19:21:55	0|jonasschnelli|Have we made sure there are no dependencies between cases?
331 2018-04-05 19:21:57	0|sipa|concurrency doesn't even come into the pictire
332 2018-04-05 19:22:00	0|wumpus|that sounds like a ton of overhead
333 2018-04-05 19:22:10	0|wumpus|launcing a process for every test case
334 2018-04-05 19:22:14	0|sipa|wumpus: 250 process creations.
335 2018-04-05 19:22:15	0|achow101|cases should be independent of each other
336 2018-04-05 19:22:16	0|sipa|?
337 2018-04-05 19:22:18	0|wumpus|yes
338 2018-04-05 19:22:33	0|aj|test suite is the file, test case is the function (and each case has many checks)
339 2018-04-05 19:22:36	0|jonasschnelli|achow101: Yes. But are they (ex. wallet test)?
340 2018-04-05 19:22:57	0|jonasschnelli|But however, suite is what we want not cases
341 2018-04-05 19:23:02	0|jonasschnelli|*suites
342 2018-04-05 19:23:12	0|cfields|so, the tests can be built as a library...
343 2018-04-05 19:23:14	0|MarcoFalke|The savings from --jobs=2 eat all the overhead from running in 250 processes
344 2018-04-05 19:23:21	0|wumpus|sounds like a better granularity to me too
345 2018-04-05 19:23:33	0|wumpus|in any case we need to get rid of the txt file with all the test cases
346 2018-04-05 19:23:36	0|wumpus|and generate that automatically
347 2018-04-05 19:23:37	0|meshcollider|Agree
348 2018-04-05 19:23:40	0|jonasschnelli|Yes.
349 2018-04-05 19:23:53	0|sipa|that seems easy
350 2018-04-05 19:24:00	0|jonasschnelli|(same should be done for the functional test IMO, *OT* though9
351 2018-04-05 19:24:08	0|wumpus|too easy to forget a test now
352 2018-04-05 19:24:10	0|sipa|we can grep for test cases/suites
353 2018-04-05 19:24:37	0|MarcoFalke|wumpus: If we keep the list it would be linted on travis of course. *ducks*
354 2018-04-05 19:24:47	0|cfields|not sure how it works, but if boost provides a reasonable api that let us fork() into each suite, we could write our own test_main.cpp to do so, no?
355 2018-04-05 19:24:53	0|wumpus|jonasschnelli: yes, there were plans for that too, embedding some metadata in a header at the top of the py files, and automatically generating the lists, but orthogonal :)
356 2018-04-05 19:25:01	0|wumpus|MarcoFalke: :-(
357 2018-04-05 19:25:10	0|jonasschnelli|+1 :(
358 2018-04-05 19:25:27	0|MarcoFalke|Fine
359 2018-04-05 19:25:54	0|jtimon|wumpus: if we have one dir with all the cases and only that, it should be simple, perhaps we want to maintain the list of extra ones to skip the slow suites by default
360 2018-04-05 19:26:16	0|wumpus|jtimon: there are some other parameters: sort order, and which list it goes into
361 2018-04-05 19:26:31	0|wumpus|jtimon: but anyhow off topic now
362 2018-04-05 19:26:33	0|MarcoFalke|Other topics
363 2018-04-05 19:26:35	0|jonasschnelli|topic proposal: multiwallet merge (luke-jr brought this up last time while I was not here)
364 2018-04-05 19:26:35	0|MarcoFalke|?
365 2018-04-05 19:26:45	0|wumpus|#topic multiwallet GUI
366 2018-04-05 19:26:54	0|jonasschnelli|But I guess luke-jr is not here now
367 2018-04-05 19:27:10	0|jtimon|right, mhmm, I guess you can rename the tests starting with numbers to keep the order, but that's kind of ugly
368 2018-04-05 19:27:24	0|wumpus|cfields: I don't think doing it that way would change the challenges
369 2018-04-05 19:27:28	0|jonasschnelli|I heard that the merge of #12610 was done while it was still controversial...
370 2018-04-05 19:27:30	0|gribble|https://github.com/bitcoin/bitcoin/issues/12610 | Multiwallet for the GUI by jonasschnelli · Pull Request #12610 · bitcoin/bitcoin · GitHub
371 2018-04-05 19:27:38	0|wumpus|jonasschnelli: I'm happy that you merged it
372 2018-04-05 19:27:46	0|jonasschnelli|If I did so, my appologies. I only wanted to make progress.
373 2018-04-05 19:27:49	0|cfields|ok
374 2018-04-05 19:27:50	0|wumpus|if there's anything to be fixed, file a new PR
375 2018-04-05 19:28:03	0|jonasschnelli|PRs to fix design mistakes are welcome
376 2018-04-05 19:28:08	0|jonasschnelli|yes
377 2018-04-05 19:28:19	0|wumpus|luke-jr overblows that part imo
378 2018-04-05 19:28:43	0|meshcollider|Yeah it's good that something has been put in, it's been weeks of small disagreement holding it up
379 2018-04-05 19:28:44	0|jnewbery|jonasschnelli: better to get it in now so it has time to soak. Rough edges can be fixed in future PRs
380 2018-04-05 19:29:05	0|jonasschnelli|I overhauled luke-jr's version mainly because of things like this: https://github.com/bitcoin/bitcoin/pull/11383/files#diff-2c15c5b52f35ea388ebab757eaab0f1cR506
381 2018-04-05 19:29:19	0|wumpus|but in any case, the idea of open source software is collaborative development, we can't make progress with something like this if it stays a PR forever, and it had a quite lot of review IIRC
382 2018-04-05 19:29:23	0|jonasschnelli|Erm this: https://github.com/bitcoin/bitcoin/pull/11383/files#diff-2c15c5b52f35ea388ebab757eaab0f1cR903
383 2018-04-05 19:29:46	0|wumpus|yes
384 2018-04-05 19:29:49	0|jonasschnelli|Yes. I took also care to keep luke-jr authorship in commits during my overhaul.
385 2018-04-05 19:30:29	0|jonasschnelli|Okay. Done with that topic then. Thanks
386 2018-04-05 19:30:33	0|wumpus|so it's ok, any other topics?
387 2018-04-05 19:30:59	0|sipa|let me think
388 2018-04-05 19:31:17	0|jnewbery|I had a topic: merge #10244
389 2018-04-05 19:31:23	0|gribble|https://github.com/bitcoin/bitcoin/issues/10244 | Refactor: separate gui from wallet and node by ryanofsky · Pull Request #10244 · bitcoin/bitcoin · GitHub
390 2018-04-05 19:31:23	0|jnewbery|but wumpus already did
391 2018-04-05 19:31:30	0|jamesob|woo!
392 2018-04-05 19:31:31	0|sipa|i don't have anything
393 2018-04-05 19:31:39	0|wumpus|yes, congrats :)
394 2018-04-05 19:31:44	0|jnewbery|\o/
395 2018-04-05 19:31:46	0|sipa|yay
396 2018-04-05 19:31:49	0|jonasschnelli|nice! Yes.
397 2018-04-05 19:31:52	0|MarcoFalke|Good to see that in
398 2018-04-05 19:31:56	0|BlueMatt|dont forget to review High-Priority PRs this week!
399 2018-04-05 19:32:05	0|BlueMatt|</meeting>?
400 2018-04-05 19:32:12	0|MarcoFalke|#action  dont forget to review High-Priority PRs this week!
401 2018-04-05 19:32:35	0|lightningbot|Log:            http://www.erisian.com.au/meetbot/bitcoin-core-dev/2018/bitcoin-core-dev.2018-04-05-19.00.log.html
402 2018-04-05 19:32:35	0|lightningbot|Meeting ended Thu Apr  5 19:32:44 2018 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
403 2018-04-05 19:32:35	0|lightningbot|Minutes:        http://www.erisian.com.au/meetbot/bitcoin-core-dev/2018/bitcoin-core-dev.2018-04-05-19.00.html
404 2018-04-05 19:32:35	0|lightningbot|Minutes (text): http://www.erisian.com.au/meetbot/bitcoin-core-dev/2018/bitcoin-core-dev.2018-04-05-19.00.txt
405 2018-04-05 19:32:35	0|wumpus|#endmeeting
406 2018-04-05 19:33:30	0|aj|any chance of getting #12878 added to high-pri list? it keeps needing rebasing, and at least it's easy to review...
407 2018-04-05 19:33:32	0|gribble|https://github.com/bitcoin/bitcoin/issues/12878 | [refactor] Config handling refactoring in preparation for network-specific sections by ajtowns · Pull Request #12878 · bitcoin/bitcoin · GitHub
408 2018-04-05 19:36:20	0|cfields|aj: will review asap.
409 2018-04-05 19:36:22	0|wumpus|aj: added
410 2018-04-05 19:36:56	0|aj|wumpus: thanks
411 2018-04-05 19:44:12	0|bitcoin-git|[13bitcoin] 15chivambo opened pull request #12893: junior (06master...06patch-1) 02https://github.com/bitcoin/bitcoin/pull/12893
412 2018-04-05 19:44:56	0|bitcoin-git|[13bitcoin] 15sipa closed pull request #12893: junior (06master...06patch-1) 02https://github.com/bitcoin/bitcoin/pull/12893
413 2018-04-05 20:02:46	0|bitcoin-git|[13bitcoin] 15MarcoFalke opened pull request #12894: tests: Avoid test suite name collision in wallet crypto_tests (06master...06Mf1804-testRenameWalletCryptoTests) 02https://github.com/bitcoin/bitcoin/pull/12894
414 2018-04-05 20:09:51	0|sipa|cat src/{,wallet/,qt/}test/*.cpp | fgrep BOOST_FIXTURE_TEST_SUITE | cut -d '(' -f 2 | cut -d ',' -f 1 | shuf | xargs -n 1 -P 4 -I "{}" -- ./src/test/test_bitcoin -t "{}/*"
415 2018-04-05 20:10:25	0|sipa|MarcoFalke: ^
416 2018-04-05 20:12:31	0|achow101|one line of bash?
417 2018-04-05 20:13:14	0|MarcoFalke|sipa: Needs white space stripped as well?
418 2018-04-05 20:13:21	0|sipa|hmm?
419 2018-04-05 20:13:33	0|MarcoFalke|TEST_SUITE(  a   , ...)
420 2018-04-05 20:13:45	0|MarcoFalke|The one before and after "a"
421 2018-04-05 20:13:55	0|MarcoFalke|Otherwise I get "Test setup error: no test cases matching filter or all test cases were disabled"
422 2018-04-05 20:14:04	0|sipa|heh
423 2018-04-05 20:14:22	0|MarcoFalke|That xargs -P is cool
424 2018-04-05 20:14:55	0|sipa|add | tr -d '[:space:]' | somewhere
425 2018-04-05 20:15:28	0|jamesob|I think `shuf -z` + `xargs -0` is recommended
426 2018-04-05 20:15:41	0|MarcoFalke|Also, nice to use one iteration of bogosort to sort by run time
427 2018-04-05 20:16:16	0|sipa|jamesob: i don't think we're particularly worried about test names with newlines in them :)
428 2018-04-05 20:16:45	0|sipa|but yes
429 2018-04-05 20:16:48	0|jamesob|sipa: I'll make it a point to include one in my next testcase ;)
430 2018-04-05 20:16:58	0|MarcoFalke|Someone should see if it works with emojis in test names
431 2018-04-05 20:17:09	0|jamesob|I mean heck, if we're gonna do this as a one-liner we can at least overengineer it a little
432 2018-04-05 20:24:33	0|webuser323|sipa, MarcoFalke, I was just reading meeting log and though maybe 'man parallel' since you mentioned .sh for running tests in parallel. It's a great little tool.
433 2018-04-05 20:28:12	0|sipa|webuser323: it is, and MarcoFalke wanted to use that initially
434 2018-04-05 20:28:24	0|sipa|unfortunately it's a bit of a hassle.to include it in our repo
435 2018-04-05 20:32:46	0|webuser323|no problems, just thought I'll chime in with this in case it slipped somehow
436 2018-04-05 20:33:15	0|sipa|thanks!
437 2018-04-05 20:34:11	0|bitcoin-git|[13bitcoin] 15practicalswift opened pull request #12895: tests: Add note about test suite name uniqueness requirement to developer notes (06master...06check-uniqueness-of-test-suite-names) 02https://github.com/bitcoin/bitcoin/pull/12895
438 2018-04-05 20:35:26	0|wumpus|you could always abuse make to run things parallel *ducks*
439 2018-04-05 20:37:29	0|aj|wumpus: nice
440 2018-04-05 20:41:36	0|sipa|wumpus: oh!
441 2018-04-05 20:42:05	0|sipa|do you know how make does thread counting across processes?
442 2018-04-05 20:42:44	0|sipa|it's so cute
443 2018-04-05 20:44:28	0|achow101|instagibbs: rebased psbt
444 2018-04-05 20:46:56	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #12894: tests: Avoid test suite name collision in wallet crypto_tests (06master...06Mf1804-testRenameWalletCryptoTests) 02https://github.com/bitcoin/bitcoin/pull/12894
445 2018-04-05 20:46:56	0|wumpus|sipa: no, I don't know
446 2018-04-05 20:47:42	0|cfields|sipa: the special fd thing?
447 2018-04-05 20:47:53	0|sipa|wumpus: it has a shared fifo; every time a thread starts, it reads a char from that fifo, and when that is done, it writes another character
448 2018-04-05 20:48:12	0|sipa|it doesn't know or care where that char goes :p
449 2018-04-05 20:48:53	0|wumpus|ahh that makes sense, nice, didn't know you could share a fifo that way
450 2018-04-05 20:49:07	0|cfields|sipa: feel free to hack that into test_runner so that we could hook it up to make and -jX would just work as intended :)
451 2018-04-05 20:49:20	0|cfields|that's been on my personal todo for ages
452 2018-04-05 20:52:07	0|instagibbs|achow101, great
453 2018-04-05 21:04:38	0|bitcoin-git|[13bitcoin] 15practicalswift opened pull request #12896: docs: Fix conflicting statements about initialization in developer notes (06master...06remove-conflicing-statements) 02https://github.com/bitcoin/bitcoin/pull/12896
454 2018-04-05 22:38:06	0|bitcoin-git|[13bitcoin] 15practicalswift closed pull request #12782: Explicitly state our assumptions about LookupBlockIndex(...) return values (06master...06LookupBlockIndex-assumptions) 02https://github.com/bitcoin/bitcoin/pull/12782
455 2018-04-05 22:40:45	0|bitcoin-git|[13bitcoin] 15practicalswift opened pull request #12897: Add GetBlockIndex(const uint256& hash) for when the caller assumes that the block index exists for the given block hash (06master...06GetBlockIndex) 02https://github.com/bitcoin/bitcoin/pull/12897