00:29:26 <yanmaani> Hi, is this a good place to ask questions about elliptic-curve cryptography?
00:32:48 <UkoeHB_> yanmaani: try #monero-research-lab
00:33:46 <yanmaani> thanks
08:04:25 <selsta> fluffypony: please update the DNS hashes to v0.17 whenever you have time
14:10:37 <hyc> so ... extracting glibc-2.19 in a usable form is a bit tricky
14:10:43 <hyc> but I got a successful build
14:11:07 <hyc> for x86-64 so far.
14:11:51 <hyc> unfortunately there are no cross-compiler packages from 2.19, so it's laid out as if running natively on a machine
14:12:11 <hyc> and stuff needs to be moved around to be where the cross-compiler expects to find it
14:13:58 <hyc> but I verified that there is no reference to signgam in the resulting binary
14:38:11 <hyc> hm, that may be overkill. all we need to do is grab the older /usr/include/x86_64-linux-gnu/bits/math-finite.h
14:38:24 <hyc> which is where lgamma is defined, and where signgam gets used
14:39:00 <hyc> alternately, try to tweak the __USE_MISC or __USE_XOPEN flags to avoid it getting used in 2.23
14:46:04 <hyc> another alternative - just declare an int "signgam" in one of our own files and let the linker resolve against it
14:46:13 <hyc> there's nothing special about that var
14:47:20 <hyc> but I can't find anywhere that actually references lgamma in our source tree
14:48:38 <hyc> so how is this getting referenced in the object files? it's in cryptonote_core/tx_pool.o and cryptonote_protocol/levin_notify.o
14:48:38 <moneromooo> This might have come up before. IIRC some gamma fnuction is used by a boost function we use.
14:48:52 <moneromooo> Some RNG related function.
14:48:55 <hyc> ah
14:50:05 <moneromooo> Ah. poisson_distribution.
14:50:12 <hyc> I see 67ade8005
14:50:43 <moneromooo> I've still got this chat from may:
14:50:45 <moneromooo> < TheCharlatan> I'm still not sure where the symbol, "signgam", actually is being introduced. As far as I can tell we are not using any of the gamma functions tha>
14:50:48 <moneromooo> <+moneromooo> gamma as in obscure libm stuff ?
14:50:51 <moneromooo> < TheCharlatan> yeah
14:50:53 <moneromooo> <+moneromooo> I suppose nm -a on object files will show you what's requiring it.
14:50:57 <moneromooo> <+moneromooo> Shows as... D IIRC ?
14:50:59 <moneromooo> <+moneromooo> No, should be U.
14:51:02 <moneromooo> <+moneromooo> I have hits in tx_pool and levin_notify.
14:51:04 <moneromooo> <+moneromooo> Can't see why atm though.
14:51:07 <moneromooo> <+moneromooo> objdump shows it used suspiciously close to poisson_distribution
14:51:10 <moneromooo> <+moneromooo> In both.
14:51:12 <moneromooo> <+moneromooo> I kinda remember something about this, like gamma is the CDF of hte poisson distribution of something along those lines, so it makes sense at first >
14:51:15 <moneromooo> <+moneromooo> The man page says something which might help:
14:51:18 <moneromooo> <+moneromooo>        signgam:
14:51:20 <moneromooo> <+moneromooo>            _XOPEN_SOURCE
14:51:23 <moneromooo> <+moneromooo>                || /* Since glibc 2.19: */ _DEFAULT_SOURCE
14:51:25 <moneromooo> <+moneromooo>                || /* Glibc versions <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE
14:51:28 <moneromooo> <+moneromooo> Not sure how much earlier than 2.19 though.
14:52:35 <hyc> crypto::random_poisson_xxx
14:54:15 <hyc> std::poisson_distribution
14:54:21 <hyc> so it's not in boost now
14:59:49 <hyc> and std::poisson_distribution uses std::gamma
14:59:57 <hyc> std::lgamma
15:00:26 <hyc> none of which references signgam, nothing in the C++ library cares about it
15:00:33 <hyc> what a crock
15:04:05 <hyc> maybe we can just wrap lgamma_r
15:52:51 <TheCharlatan> hyc I attempted that, but gave up after a while.
15:53:46 <hyc> we can stash a local copy of include/xx/bits/math-finite.h
15:53:53 <hyc> and tweak the definition
15:53:57 <hyc> I just tried that, it works
15:54:47 <TheCharlatan> ah, brilliant.
17:31:09 -xmr-pr- xiphon opened pull request #6863: cmake: CheckTrezor - use ${CMAKE_CURRENT_LIST_DIR} to locate tests
17:31:09 -xmr-pr- > https://github.com/monero-project/monero/pull/6863
17:31:09 -xmr-pr- hyc opened pull request #6862: Fix lgamma/signgam dependency
17:31:09 -xmr-pr- > https://github.com/monero-project/monero/pull/6862
17:31:44 <hyc> TheCharlatan: not entirely sure the patch was done in the best place, but couldn't think of anything better
17:48:19 <selsta> .merges
17:48:19 -xmr-pr- 6808 6822 6828 6833
19:00:08 <needmoney90> https://usercontent.irccloud-cdn.com/file/soCLTubt/zf19wzbcjer51.png
21:40:12 <hyc> hmmm this patch doesn't affect any mining code, so why is CI test failing? https://github.com/monero-project/monero/runs/1216724747
21:42:22 <moneromooo> Possibly because it'd also fail with master.
21:44:49 <moneromooo> https://github.com/monero-project/monero/pull/6836 was recently merged. The original problem was just the github machines being slow and timeouts triggering.
21:45:05 <moneromooo> Or selsta's machine being slow ? Not sure.
21:45:09 <hyc> ah ok
21:50:44 <selsta> 6836 was just to make it more clear what exactly is failing, "Failed to mine successor to block 13 (initial block = 10)"
21:51:04 <selsta> so slow github machine? or why would it fail?
21:52:10 <selsta> only happens rarely so most likely timeout related
21:52:17 <moneromooo> It could be a Microsoft plot, to start introducing more and more spurious failures by poking into RAM, to slowly take more and more time from free software maintainers and drive their output into the ground...
21:53:13 <hyc> lol
21:53:27 <hyc> CI=continuous interference
21:53:31 <moneromooo> :D
21:54:27 <hyc> have a similar problem with the OpenLDAP test suite, many of the scripts have a short sleep to let stuff happen before checking for a result
21:54:44 <hyc> and those sleeps have to be lengthened when running on our CI system
21:55:12 <hyc> the alternative is continually polling the server to check for completion, but the polling itself can be a major slowdown on a bogged-down VM
21:55:43 <hyc> running tests on windows is the worst...
22:07:02 <hecate> hey, I have an issue "undefined reference to a symbol" and "error adding symbols: DSO missing from command line". I am trying to use the gmp library within miner.cpp. Do you know how I can tell the Makefile to link this library when compiling miner.cpp? (I think this is what I need)
22:09:33 <moneromooo> Add -lgmp to the link line. If using cmake, add "gmp" to the cmake target.
22:13:05 <hecate> thank you, I'm just a novice to working on big project like this. Where is the link line for miner.cpp? I have included -lgmp when building a local file, so I know that is what I need to add, just don't know exactly where
22:19:35 <moneromooo> CMakeLists.txt, same directory.
22:26:45 <hecate> set(CMAKE_CXX_FLAGS "-lgmp")
22:26:45 <hecate> like this^ just in the CMakeLists.txt file?
22:27:14 <moneromooo> I'd add it to the existing lib list.
22:45:18 <hecate> hmm, there are a lot of lib variables I see, but this line looks promising:
22:45:18 <hecate> set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -D_GNU_SOURCE ${MINGW_FLAG} ${STATIC_ASSERT_CPP_FLAG} ${WARNINGS} ${CXX_WARNINGS} ${COVERAGE_FLAGS} ${PIC_FLAG} ${CXX_SECURITY_FLAGS}")
22:46:03 <hecate> I tried placing it in EXTRA_LIBRARIES, but that didnt seem to work
22:55:18 <selsta> hecate: so just `list(APPEND EXTRA_LIBRARIES "-lgmp")` did not work?
22:56:27 <hecate> my syntax was not that, and so probably wrong. Seems to be compiling how I have it now, will try that afterwards
23:01:44 <hecate> thanks to both of you for the help. I greatly appreciate it. I was messing with this codebase for almost a week before reaching out here. You guys are really helpful