Can you build this in OSX? I can not

[blockcorporation]

I can build bitcoin, peerunity, and others but for this project I can not build.

I get the following error

ld: symbol(s) not found for architecture x86_64

I would think it is something to do with the code? Maybe we can fix this to make it easier to compile on OSX?

[blockcorporation]

Perhaps related to this?

Peerunity/Peerunity#148

[bearsylla]

I don't have a Mac machine to test with, however, I managed to have a community member to compile blackcoind using macports on his mac.

sudo port install boost db48 openssl miniupnpc
git clone https://github.com/rat4/blackcoin/
cd blackcoin/src
make -f makefile.osx

[bearsylla]

For compiling the QT wallet, appearently there is some bugs that affect bitcoin as well, https://trac.macports.org/ticket/46951

[blockcorporation]

Yes, I was able to compile the daemon but still unable to compile the -qt. I should have been more specific.

[bearsylla]

According to the link I gave in privous comment, It seems that compiling the QT require patching of all the boost headers.

[bearsylla]

Have you tried using homebrew instead of macports?

[blockcorporation]

Yeah I have tried with both. Fresh installs. I believe it is a misconfigured -qt makefile file for mavericks.

[jprichardson]

Anymore luck with this? Tried on Yosemite at all? I may give it a shot on Yosemite/homebrew.

[nanocoinorg]

I had these issues and I made modifications to the *.pro file that has it compiling. Unfortunately now I see a runtime error:

Assertion failed: (!pthread_mutex_lock(&m)), function lock, file /opt/local/include/boost/thread/pthread/recursive_mutex.hpp, line 110.
Abort trap: 6

It seems other projects have ran into this issue, once I solve it I can do a pull request.

[ghost]

By default AssertLockHeld() is a noop and therefore shouldn't matter. A backtrace of the assertion fail might shed light on the problem.

[nanocoinorg]

I typically use linux so I'm still trying to find the best way to insert a backtrace to get information regarding the error.

[nanocoinorg]

I went through and commented out each AssertLockHeld and it didn't change anything so it is unrelated. I must say I find it funny that Litecoin and Reddcoin both have simply removed the asserts considering the usefulness in debugging.

[nanocoinorg]

I have this fixed, it requires two minor changes to the .pro file. If you need me to build your newest OSX binary I can do this for you too.

[ghost]

That'd be great if you share the fix.

[mammix2]

Im having the same issues today, If you could share that fix for this error would be great :)

'Assertion failed: (!pthread_mutex_lock(&m)), function lock, file /opt/local/include/boost/thread/pthread/recursive_mutex.hpp, line 110.'

[mammix2]

probably not ideal but adding NDEBUG to the .pro DEFINES allowed it to compile and run.
More of a workaround than a fix.

[ghost]

NDEBUG is too dangerous workaround, as it let wallet run despite the unknown bug.

[mammix2]

Can you explain in more detail why it's too dangerous?
Just for my knowledge i would like to understand. Thanks

[xranby]

mammix2: the assert is a runtime check to make sure that two simultaneously running threads of the program that manipulate a shared piece of information take turns by acquiring a lock before they alter the shared information. The assert has failed this means that data-corruption will occur if the two threads alters the shared piece of information at the same time. By using NDEBUG you basically say its OK to make the program continue to run despite that it has detected an error which is dangerous to your data.

If this kind of assert only triggers on OSX then the root cause of the issue may be in the Boost library’s or in the OSX posix thread handling.

[mammix2]

@xranby, thankyou very much. Very helpful info :)

[xranby]

FYI: blackcoin internally uses pthread_cond_wait and this function has been broken on Mac OS X
it is fixed in boost 1.57
http://www.boost.org/users/history/version_1_57_0.html

I would recommend you to try building blackcoin on OSX using at least boost 1.57

The boost 1.58 beta also fixes some more thread issues:
http://www.boost.org/users/history/version_1_58_0.html

[mammix2]

@xranby, perfect, i'll give that a go

[mammix2]

unfortunately still getting the same error after updating boost. Updated ports and confirmed i'm now running version boost @1.57.0_1

Any ideas?
do i need to change any boost suffix in the .pro file to specify 1.57 ?

[xranby]

can you attach a debugger like gdb or lldb and print a backtrace
(bt using gdb)
https://sourceware.org/gdb/onlinedocs/gdb/Backtrace.html
(thread backtrace when using lldb)
http://lldb.llvm.org/tutorial.html
when you hit the assert?

[mammix2]

Using LLDB within my Qt IDE, i think the first assert causing the error is in sync.h line 66
I'm far from being an expert at this stuff, could you confirm if this is correct?
Here is the full backtrace.
https://gist.github.com/mammix2/a8096aa3001bca71f42a

[ghost]

That looks like the static initialization order fiasco, and should be fixed by commit 7d766db

[mammix2]

@rat4, well done buddy. that has sorted the problem :)
@xranby thanks for your help too.

[mammix2]

If you guys want an BC Mac wallet compiled let me know. I'll be glad to help :)

[bearsylla]

From IRC: < nscrypto> hey, when compiling on osx I get this error http://pastebin.com/j2uKgT6s

It was boost 1.58

Edit: Update

< nscrypto> ok, i just tried compiling bitcoin and I get the same error
< nscrypto> must be boost/macports

[xranby]

The boost/static_assert.hpp is broken when boost has defined BOOST_NO_CXX11_STATIC_ASSERT

#ifndef BOOST_NO_CXX11_STATIC_ASSERT
#else
            // code here is in clearly wrong order when using boost 1.58
#endif

I will explain:

/opt/local/include/boost/static_assert.hpp:87:26: note: template is declared here
template <bool x> struct STATIC_ASSERTION_FAILURE;
                         ^

this line at 87 with the template definition has to be above the line 36 where the template is used:

/opt/local/include/boost/static_assert.hpp:36:48: note: expanded from macro 'BOOST_STATIC_ASSERT_MSG'
#     define BOOST_STATIC_ASSERT_MSG( B, Msg ) BOOST_STATIC_ASSERT( B )
                                               ^
/opt/local/include/boost/static_assert.hpp:169:13: note: expanded from macro 'BOOST_STATIC_ASSERT'
            sizeof(::boost::STATIC_ASSERTION_FAILURE< BOOST_STATIC_ASSERT_BOOL_CAST( __VA_ARGS__ ) >)>\
            ^

Reorder the two blocks and you have fixed boost.

[barnumbirr]

@xranby Could you be more specific please? I didn't understand what I needed to modify in /opt/local/include/boost/static_assert.hpp so that this would work with boost 1.58.

Thank you.

[ghost]

@Mrsmn current git master should work with boost 1.58

[MystPhysX]

Thank you @rat4

Other coins are experiencing this issue as well. I just wish this was higher up on the Google search results.

[barnumbirr]

I actually found an easier fix that worked for me on OSX 10.9.5:

brew install boost --c++11

Cheers.

[mammix2]

@MystPhysX i forgot about this thread fixing that assert issue. Hope you get it sorted :)

[MystPhysX]

@Mrsmn Thank you. I was not aware that Boost built without c++11 support by default, I'll have to keep that in mind in the future.

@mammix2 Cheers.