On a clean ubuntu 14.04.
(where can Servus be found? It does not seem to be in https://github.com/Eyescale)

git clone https://github.com/Eyescale/Buildyard
cd Buildyard
cmake .
make
make Lunchbox
Can't build Lunchbox: missing Servus
make Servus
make[1]: *** No rule to make target `Servus'. Stop.

byteswap.h is not found on Mac OS X, apparently the GNU byteswap.h says:
//#warning "byteswap.h is an unportable GNU extension! Don't use!"

Various similar problems exist with other libraries,
and one solution suggests using
the OSX header <libkern/OSByteOrder.h> and
use OSSwapInt32, OSSwapInt64 functions.

In "lunchbox.bitOperation.h"
...

ifdef _MSC_VER

pragma warning (push)

pragma warning (disable: 4985) // inconsistent decl of ceil

include <intrin.h>

pragma warning (pop)

elif defined APPLE

include <libkern/OSByteOrder.h>

define bswap_32 OSSwapInt32

define bswap_64 OSSwapInt64

elif ...

I think unregisterRequest should be deprecated and be replaced by a new function that behaves as the new registerRequest, including also the relinquish() call within.
Since the unregisterRequest method is meant to be used only when the request is no longer needed, it shouldn't wait anymore, so a relinquish() call should be implicit. Otherwise it would fall into a double wait() when the destructor gets called.

124s] /home/abuild/rpmbuild/BUILD/Lunchbox-1.12.0/CMake/common/cpp/test.h:100:51: error: too many arguments to function 'void lunchbox::abort()'
[ 124s] lunchbox::abort( true /dumpThreads/ );
[ 124s] ^
[ 124s] In file included from /home/abuild/rpmbuild/BUILD/Lunchbox-1.12.0/lunchbox/thread.h:24:0,
[ 124s] from /home/abuild/rpmbuild/BUILD/Lunchbox-1.12.0/CMake/common/cpp/test.h:34,
[ 124s] from /home/abuild/rpmbuild/BUILD/Lunchbox-1.12.0/tests/any.cpp:18:
[ 124s] /home/abuild/rpmbuild/BUILD/Lunchbox-1.12.0/lunchbox/debug.h:36:19: note: declared here
[ 124s] LUNCHBOX_API void abort();

See more: https://pmbs.links2linux.de/package/live_build_log/home:awissu:equalizer-qt5/Lunchbox/openSUSE_Tumbleweed/i586

It seems the CMake/common/ittnotify.h is licensed by GPLv2, not LGPLv2, thus probably the result cannot be LGPLv2, but only GPLv2 or 'GPLv2 and LGPLv2', or drop the problematic file.

The full license text should be included in the project, please add it to the repo, e.g.:
https://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt
https://www.gnu.org/licenses/old-licenses/gpl-2.0.txt

Reason seems to be an overflow in tv_nsec.

Discovered with new unit test in Equalizer: Apparently gpu-sd/avahi uses signals, which causes the test watchdog thread to wake up and terminate early. Make lunchbox::sleep() signal safe.

I am trying to package Lunchbox for Gentoo Linux. Therefore, I do not follow the official build instructions using Buildyard, but call CMake through the package manager with Buildyard disabled.

While this procedure used to work with Lunchbox 1.9.1, the CMake build fails with 1.10.0. The relevant error should be the following:

CMake Error at CMake/common/clangcheckTargets.cmake:42 (get_target_property):
  INTERFACE_LIBRARY targets may only have whitelisted properties.  The
  property "SOURCES" is not allowed.
Call Stack (most recent call first):
  CMake/common/TargetHooks.cmake:36 (add_clangcheck)
  /usr/share/cmake/Modules/FindThreads.cmake:207 (add_library)
  pthreads/CMakeLists.txt:4 (find_package)
  CMakeLists.txt:64 (include)

The complete build log including the CMake call is available here:
https://gist.github.com/seifertm/932e6ff8205e9b4917e7

libLunchbox.so.1.17.0
libLunchbox.so.10

It's due to:

project(Lunchbox VERSION 1.17.0)
set(Lunchbox_VERSION_ABI 10)

IMHO if the SONAME is libLunchbox.so.10, I think the library version should be libLunchbox.so.10.0.0.

compressorTurboJPEG.cpp:35
static bool _init = tjInitCompress();

The returned handle is never free'd, thus creating a memleak.

I have looked at a few source files for your current software. I have noticed that some checks for return codes are missing.

Would you like to add more error handling for return values from functions like the following?

• pthread_mutex_lock ⇒ Lock::set
• pthread_cond_signal ⇒ Condition::signal

Noticed when compiling Equalizer, details in
Eyescale/Equalizer#498

Ubuntu 14.04.3.
adding the following option (link against libleveldb.so) works. Please fix the CMakeLists.txt. Thanks!
-lleveldb
to the offending line, so that it ends as
/usr/bin/c++ -Wall -Wextra -Winvalid-pch -Winit-self -Wno-unknown-pragmas -Werror -Wshadow -fmax-errors=5 -Wnon-virtual-dtor -Wsign-promo -Wvla -fno-strict-aliasing -std=c++11 -fopenmp -g CMakeFiles/perf-persistentMap.dir/persistentMap.cpp.o -o perf-persistentMap -rdynamic ../../lib/libLunchbox.so.1.12.0 -lboost_regex -lboost_serialization -lboost_filesystem -lboost_system -lboost_thread -lboost_unit_test_framework -lpthread /usr/lib/openmpi/lib/libmpi_cxx.so /usr/lib/openmpi/lib/libmpi.so -ldl -lhwloc ../../lib/libServus.so.1.1.0 -Wl,-rpath,/home/rgarcia/lrz/equalizer/Equalizer/build/lib:/usr/lib/openmpi/lib -lleveldb

On Fedora Rawhide (f35):

19/29 Test #19: perf-memory ......................***Exception: SegFault  0.19 sec

./perf-memory
  PID.Thread  | Filename:line                    |   ms | Message
   35.1        ox-1.17.0/lunchbox/thread.cpp:155       0 Thread #1 type (anonymous namespace)::Watchdog successfully initialized
   35.Main     hbox-1.17.0/lunchbox/init.cpp:70        0 Log level DEBUG topics 4 date Wed Jun 16 00:40:29 2021
Segmentation fault (core dumped)

When trying to compile the git version of Lunchbox (commit e3645e5), the build exits with the following error:

[ 52%] Building CXX object lunchbox/CMakeFiles/Lunchbox.dir/servus.cpp.obj
cd /usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999_build/lunchbox && /usr/bin/x86_64-w64-mingw32-g++  -DBOOST_ALL_NO_LIB -DBOOST_TEST_DYN_LINK -DBoost_NO_BOOST_CMAKE -DCXX_ARRAY_SUPPORTED -DCXX_AUTO_SUPPORTED -DCXX_FINAL_OVERRIDE_SUPPORTED -DCXX_NULLPTR_SUPPORTED -DCXX_SHAREDPTR_SUPPORTED -DCXX_TEMPLATE_ALIAS_SUPPORTED -DCXX_TUPLE_SUPPORTED -DCXX_UNORDERED_MAP_SUPPORTED -DEQ_PLUGIN_BUILD -DHAVE_BYTESWAP_H -DLUNCHBOX_BUILD_DIR=\"/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999_build/\" -DLUNCHBOX_DSO_NAME=\"libLunchbox.dll.3\" -DLUNCHBOX_LITTLEENDIAN -DLUNCHBOX_SHARED -DLUNCHBOX_USE_BOOST=1 -DLUNCHBOX_USE_V1_API=1 -DLUNCHBOX_VERSION=1.9.1 -DLunchbox_EXPORTS -DWARN_DEPRECATED -DWin32=1  -DNDEBUG -O2 -pipe -fomit-frame-pointer  -Wall -Wextra -Winvalid-pch -Winit-self -Wno-unknown-pragmas -Wshadow -fmax-errors=5 -Wnon-virtual-dtor -Wsign-promo -Wvla -fno-strict-aliasing -std=c++11 @CMakeFiles/Lunchbox.dir/includes_CXX.rsp   -o CMakeFiles/Lunchbox.dir/servus.cpp.obj -c /usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999/lunchbox/servus.cpp
/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999/lunchbox/servus.cpp: In member function 'lunchbox::Servus::Result lunchbox::detail::Servus::browse(int32_t)':
/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999/lunchbox/servus.cpp:226:31: error: '_in' was not declared in this scope
/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999/lunchbox/servus.cpp:226:44: error: '_handleEvents' was not declared in this scope
/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999/lunchbox/servus.cpp:227:5: warning: control reaches end of non-void function [-Wreturn-type]
make[2]: *** [lunchbox/CMakeFiles/Lunchbox.dir/servus.cpp.obj] Error 1
make[2]: Leaving directory `/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999_build'
make[1]: *** [lunchbox/CMakeFiles/Lunchbox.dir/all] Error 2
make[1]: Leaving directory `/usr/x86_64-w64-mingw32/tmp/portage/dev-libs/lunchbox-9999/work/lunchbox-9999_build'
make: *** [all] Error 2

If I am not mistaken, _in and _handleEvents are only declared, when LUNCHBOX_USE_DNSSD is set, but they are still referenced outside an #ifdef in line 226.

The changes were introduced with commit a90a2da. Reverting this commit makes Lunchbox compile without errors.

....
Make Error at CMakeLists.txt:11 (message):
  CMake/common missing, run: git submodule update --init

I tried both the zip and tar.gz files from the release (lunchbox-1.17.0) and the submodule is not there. I think the release should be complete including the submodules, otherwise it complicates the downstream distro packaging (I am trying to get the package into Fedora).

When I used cmake to config in windows , the erorrs are below:
Could NOT find PkgConfig (missing: PKG_CONFIG_EXECUTABLE)
Can't find dns_sd.h header file.
Could NOT find DNSSD (missing: _dnssd_LIBRARY _dnssd_INCLUDE_DIR)

So,

1. How to get PkgConfig?
2. Where to get Dnssd lib ?

thanks a lot.

It seems that ThreadID::ZERO causes SIGSEGV when used in initialization list in LB_TS_VAR macro.

Currently I`m trying to run Lunchbox on Android, and compiling it with arm-gcc. The LFQueue test fails with SIGSEGV because of two structs defined with LB_TS_VAR in lfQueue.h

Lunchbox lunchbox/string.h introduce a function named prepend.

This function is declared and defined in a header but is not tagged as "inline". Triggering a multiple definition error when used inside equalizer.

CMakeFiles/triply.dir/vertexBufferBase.cpp.o: In function lunchbox::string::prepend(std::string const&, std::string const&)': vertexBufferBase.cpp:(.text+0x40): multiple definition of lunchbox::string::prepend(std::string const&, std::string const&)'

@hernando @eile

Reported externally, gcc 4.7 on Ubuntu 12.10 fails on an unit test:

[ 73%] Building CXX object tests/CMakeFiles/lfVector.dir/lfVector.cpp.o
In file included from /home/stewart/equalizer/Lunchbox/lunchbox/lfVector.h:236:0,
                from /home/stewart/equalizer/Lunchbox/tests/lfVector.cpp:21:
/home/stewart/equalizer/Lunchbox/lunchbox/lfVector.ipp: In function ‘void _runSerialTest() [with V = lunchbox::LFVector<unsigned int>; T = unsigned int]’:
/home/stewart/equalizer/Lunchbox/lunchbox/lfVector.ipp:153:34: error: array subscript is below array bounds [-Werror=array-bounds]
/home/stewart/equalizer/Lunchbox/lunchbox/lfVector.ipp: In function ‘int testMain(int, char**)’:
/home/stewart/equalizer/Lunchbox/lunchbox/lfVector.ipp:153:34: error: array subscript is below array bounds [-Werror=array-bounds]
cc1plus: all warnings being treated as errors
make[3]: *** [tests/CMakeFiles/lfVector.dir/lfVector.cpp.o] Error 1
make[3]: Leaving directory `/home/stewart/equalizer/Lunchbox/Release'
make[2]: *** [tests/CMakeFiles/lfVector.dir/all] Error 2
make[2]: Leaving directory `/home/stewart/equalizer/Lunchbox/Release'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/home/stewart/equalizer/Lunchbox/Release'
make: *** [install] Error 2

Tried to build for fedora-rawhide (f35):

/usr/bin/ctest --force-new-ctest-process 
Test project /builddir/build/BUILD/Lunchbox-1.17.0/redhat-linux-build
      Start  1: any
 1/28 Test  #1: any ..............................   Passed    0.02 sec
      Start  2: anySerialization
 2/28 Test  #2: anySerialization .................   Passed    0.02 sec
      Start  3: bitOperation
 3/28 Test  #3: bitOperation .....................Subprocess aborted***Exception:   0.23 sec
      Start  4: buffer
 4/28 Test  #4: buffer ...........................   Passed    0.22 sec
      Start  5: clock
 5/28 Test  #5: clock ............................   Passed    1.03 sec
      Start  6: debug
 6/28 Test  #6: debug ............................   Passed    0.02 sec
      Start  7: dso
 7/28 Test  #7: dso ..............................   Passed    0.02 sec
      Start  8: file
 8/28 Test  #8: file .............................   Passed    0.02 sec
      Start  9: future
 9/28 Test  #9: future ...........................   Passed    0.41 sec
      Start 10: init
10/28 Test #10: init .............................   Passed    0.02 sec
      Start 11: intervalSet
11/28 Test #11: intervalSet ......................Subprocess aborted***Exception:   0.21 sec
      Start 12: issue1
12/28 Test #12: issue1 ...........................   Passed    0.36 sec
      Start 13: lfQueue
13/28 Test #13: lfQueue ..........................   Passed    1.02 sec
      Start 14: memoryMap
14/28 Test #14: memoryMap ........................   Passed    0.07 sec
      Start 15: monitor
15/28 Test #15: monitor ..........................   Passed    7.12 sec
      Start 16: mtQueue
16/28 Test #16: mtQueue ..........................   Passed    0.47 sec
      Start 17: perThread
17/28 Test #17: perThread ........................   Passed    0.10 sec
      Start 18: perf-lfVector
18/28 Test #18: perf-lfVector ....................   Passed    2.37 sec
      Start 19: perf-mutex
19/28 Test #19: perf-mutex .......................   Passed    7.72 sec
      Start 20: perf-rwLock
20/28 Test #20: perf-rwLock ......................   Passed   10.46 sec
      Start 21: pluginFactory
21/28 Test #21: pluginFactory ....................   Passed    0.01 sec
      Start 22: refPtr
22/28 Test #22: refPtr ...........................   Passed    5.19 sec
      Start 23: requestHandler
23/28 Test #23: requestHandler ...................   Passed    0.02 sec
      Start 24: result
24/28 Test #24: result ...........................Subprocess aborted***Exception:   0.21 sec
      Start 25: rng
25/28 Test #25: rng ..............................   Passed    0.71 sec
      Start 26: string
26/28 Test #26: string ...........................Subprocess aborted***Exception:   0.20 sec
      Start 27: thread
27/28 Test #27: thread ...........................   Passed    2.87 sec
      Start 28: threadPool
28/28 Test #28: threadPool .......................   Passed    1.11 sec
86% tests passed, 4 tests failed out of 28

The instance of XXXX::detail::PageThread::PageThread is static and inherits from lunchbox::Thread. It
calls setName( "Main" ); in its ctor, before lunchbox::_logInstance is ctor'd:

#0  0x00007ffff5413835 in lunchbox::TLS::get (this=0x7ffff582f748 <lunchbox::_logInstance+8>) at /YYYY/Lunchbox/lunchbox/tls.cpp:66
#1  0x00007ffff540ecd8 in lunchbox::PerThread<lunchbox::Log, &lunchbox::perThreadDelete<lunchbox::Log> >::get (this=0x7ffff582f740 <lunchbox::_logInstance>) at /YYYY/Lunchbox/lunchbox/perThread.ipp:43
#2  0x00007ffff540d6d5 in lunchbox::Log::instance () at /YYYY/Lunchbox/lunchbox/log.cpp:327
#3  0x00007ffff5411325 in lunchbox::Thread::setName (name="Main") at /YYYY/Lunchbox/lunchbox/thread.cpp:304
#4  0x00007ffff4ff5b6d in XXXX::detail::PageThread::PageThread (this=0x7ffff582b320 <XXXX::(anonymous namespace)::pageThread_>) at /YYYY/XXXX/XXXXIndex.cpp:648
#5  0x00007ffff4ff3ad6 in __static_initialization_and_destruction_0 (__initialize_p=1, __priority=65535) at /YYYY/XXXX/XXXXIndex.cpp:756
#6  0x00007ffff4ff3b11 in _GLOBAL__sub_I_XXXXIndex.cpp(void) () at /YYYY/XXXX/XXXXIndex.cpp:821
....
#34 0x00007ffff7811830 in __libc_start_main (main=0x49dab0 <main>, argc=1, argv=0x7fffffffe2d8, init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffe2c8)
at ../csu/libc-start.c:291
#35 0x000000000049d9d9 in _start ()

Because static init is arbitrary, tls_'s impl_.get() fails:

(gdb) p impl_
$1 = (lunchbox::detail::TLS * const) 0x0

I can work around it by forcing XXXX to be ctor'd later, but this may hit other people too.

In file included from E:/BackUp/source/Equalizer/1.6.0/Buildyard/src/Lunchbox/lunchbox/compressor/snappy/snappy.h:45:0,
from E:/BackUp/source/Equalizer/1.6.0/Buildyard/src/Lunchbox/lunchbox/compressor/compressorSnappy.cpp:20:
E:/BackUp/source/Equalizer/1.6.0/Buildyard/src/Lunchbox/lunchbox/compressor/snappy/snappy-stubs-public.h:48:21: fatal error: sys/uio.h: No such file or directory
#include <sys/uio.h>
^
compilation terminated.
lunchbox/CMakeFiles/Lunchbox.dir/build.make:418: recipe for target 'lunchbox/CMakeFiles/Lunchbox.dir/compressor/compressorSnappy.cpp.obj' failed

• PluginRegistry
• Plugin
• CPU compressor implementations
• CPU compressor convenience functions
• GPU compressor convenience functions

• create pthread tls API to hide pthread calls
• use in perThread(Ref).h
• remove Crude test if pthread.h was included

... and likely on Windows too.

Unit test already covers this.

3>D:\B7\Buildyard-master\src\Lunchbox\tests\requestHandler.cpp(72): error C2220: warning treated as error - no 'object' file generated
3>D:\B7\Buildyard-master\src\Lunchbox\tests\requestHandler.cpp(72): warning C4305: 'argument' : truncation from 'int' to 'const bool'

/WX compiler flag

this warning/error break build in buildyard

I solve this -> //TEST( future == 0xC0FFEE );
or TEST( future.get() == 0xC0FFEE ); // i don't know if it is right :D

Hi all,

Our bug scanner has reported a data race issue at thread.cpp#L131 and thread.cpp#L202.

Followings are code snippets.

void Thread::_runChild()
{
    setName(boost::lexical_cast<std::string>(_impl->index));
    _impl->id._impl->pthread = pthread_self();

and

bool Thread::start()
{
    ...
    // avoid memleak, we don't use pthread_join
    pthread_detach(_impl->id._impl->pthread);

The race occurs when _impl->id._impl->pthread = pthread_self(); and pthread_detach(_impl->id._impl->pthread); are executed simultaneously.

SourceBrella Inc.,
Yu

The unit test for issue #5 fails:

time > timeout - 1 failed (l.41): 0.095057 < 250

Some parameters (like "${CMAKE_SOURCE_DIR}/CMake" and "${DOC_DIR}") are passed to CMake commands in your build scripts without enclosing them by quotation marks. I see that these places will result in build difficulties if the contents of the used variables will contain special characters like spaces.

I would recommend to apply advices from a Wiki article.

Need for Eyescale/gpusd#9:

• a function to turn a daemon into a process
• Better Log functionality to redirect to a file

Does skv::PersistentMap not implement getValues() and takeValues ? Or, may I not use skv?

The errers are:

/home/lxq/eq/Equalizer/Lunchbox/lunchbox/persistentMap.cpp: In function ‘lunchbox::detail::PersistentMap* {anonymous}::_newImpl(const servus::URI&)’:
/home/lxq/eq/Equalizer/Lunchbox/lunchbox/persistentMap.cpp:73:54: error: invalid new-expression of abstract class type ‘lunchbox::skv::PersistentMap’
return new lunchbox::skv::PersistentMap( uri );
^
In file included from /home/lxq/eq/Equalizer/Lunchbox/lunchbox/persistentMap.cpp:56:0:
/home/lxq/eq/Equalizer/Lunchbox/lunchbox/skv/persistentMap.h:43:7: note: because the following virtual functions are pure within ‘lunchbox::skv::PersistentMap’:
class PersistentMap : public detail::PersistentMap
^
/home/lxq/eq/Equalizer/Lunchbox/lunchbox/persistentMap.cpp:38:18: note: virtual void lunchbox::detail::PersistentMap::getValues(const Strings&, const ConstValueFunc&) const
virtual void getValues( const Strings& keys,
^
/home/lxq/eq/Equalizer/Lunchbox/lunchbox/persistentMap.cpp:40:18: note: virtual void lunchbox::detail::PersistentMap::takeValues(const Strings&, const ValueFunc&) const
virtual void takeValues( const Strings& keys,
^
lunchbox/CMakeFiles/Lunchbox.dir/build.make:446: recipe for target 'lunchbox/CMakeFiles/Lunchbox.dir/persistentMap.cpp.o' failed
make[2]: *** [lunchbox/CMakeFiles/Lunchbox.dir/persistentMap.cpp.o] Error 1
CMakeFiles/Makefile2:3472: recipe for target 'lunchbox/CMakeFiles/Lunchbox.dir/all' failed
make[1]: *** [lunchbox/CMakeFiles/Lunchbox.dir/all] Error 2
Makefile:160: recipe for target 'all' failed
make: *** [all] Error 2

We're at revision 60, there have been a few performance improvements: http://code.google.com/p/snappy/source/list

The alignment causes problems on Win32 (VS2008 & 2010) if an atomic is stored in a std::vector like in eq/client/detail/channel.ipp:74 where you get this error:

1>Compiling...
1>channel.cpp
1>D:\Program Files (x86)\Microsoft Visual Studio 9.0\VC\include\vector(717) : error C2719: '_Val': formal parameter with __declspec(align('8')) won't be aligned
1> e:\dev\buildyard\src\equalizer\libs\eq\client\detail/channel.ipp(78) : see reference to class template instantiation 'std::vector<_Ty>' being compiled
1> with
1> [
1> _Ty=eq::detail::Channel::FrameStatistics
1> ]

Learn why: http://thetweaker.wordpress.com/2010/05/05/stdvector-of-aligned-elements/

When using the monitor with non-atomic variables, e.g., uint128_t, the waitFoo() functions are not thread safe since they compare the value outside the lock first.

Solve by always comparing inside mutex and providing specializations for save 'T's

As I mentioned in a previous issue, I am currently trying to package Lunchbox for Gentoo Linux. While the standard build with Buildyard works nicely, I fear that a Lunchbox package cannot make it into the official package list of Gentoo (the Portage Tree), because the CMake build loads external dependencies during build time.

Loading the CMake modules through git causes two problems from my point of view:

• If the loaded CMake modules are not verified after the download, there is a risk that the build uses compromised CMake files, posing a potential security threat.
• There is no possibility to install the Lunchbox package without a running Internet connection

I don't know much about other distributions' packaging policies, but I imagine that they had similar concerns.

Is there a possibility that the CMake modules will be included in future releases?

When I use cmake to generate Visual Studio 2013 project, it shows following cmake entries and values:
Name Value
_LIBRARY_DEBUG Boost_THREAD
_LIBRARY_RELEASE Boost_THREAD

But, I have correctly set Boost_THREAD_LIBRARY_DEBUG and Boost_THREAD_LIBRARY_RELEASE

Can anyone tell me what happen, and how to solve ?

Some BBP viz components including BBPSDK uses the header lunchbox/string.h

However, this one is never installed because not part of the install() phase of Lunchbox. Making impossible to compile BBPSDK if it does not embed Lunchbox.