On Wednesday 10 December 2014 15:12:08 Milian Wolff wrote:
> On Tuesday 09 December 2014 23:51:04 Albert Astals Cid wrote:
> > El Dimarts, 9 de desembre de 2014, a les 02:15:58, Milian Wolff va
escriure:
> > > On Wednesday 03 December 2014 01:51:57 Aleix Pol wrote:
> > > > On Tue, Dec 2, 2014 at 6:59 PM, Milian Wolff <m...@milianw.de> wrote:
> > > > > Hey all,
> > > > >
> > > > > I have just finished writing a lengthy introduction to heaptrack, an
> > > > > alternative to Massif, see:
> > > > >
> > > > > http://milianw.de/blog/heaptrack-a-heap-memory-profiler-for-linux
> > > > >
> > > > > I'd like to see more people starting to use it. Any feedback, or
> > > > > even
> > > > > patches,
> > > > > is welcome!
> > > > >
> > > > > Especially, I'd love to see more people use it on their pet project
> > > > > in
> > > > > KDE.
> > > > > Quite often, you'll find useless temporary allocations, overly large
> > > > > memory
> > > > > consumption or even significant memory leaks. C++/Qt/KDE code can be
> > > > > extremely
> > > > > efficient, but you have to code accordingly. If you have any
> > > > > questions
> > > > > to
> > > > > the
> > > > > results you obtain from heaptrack, or how to improve the situation,
> > > > > don't
> > > > > hesitate to ask me. Please ask on a public mailing list, such that
> > > > > others
> > > > > can
> > > > > benefit from the discussion as well.
> > > > >
> > > > > Furthermore, I hereby request an official code review. Heaptrack
> > > > > currently
> > > > > lives in a scratch repository:
> > > > >
> > > > > http://quickgit.kde.org/?p=scratch%2Fmwolff%2Fheaptrack.git
> > > > >
> > > > > I want to move this to extragear, skipping playground altogether, if
> > > > > possible.
> > > >
> > > > Cool stuff! I'll definitely give it a try!
> > > >
> > > > Regarding the move to extragear, you can start requesting the
> > > > repository
> > > > to
> > > > get it in playground and then request the move to kdereview, if you're
> > > > confident it's pristine. ;)
> > >
> > > Heaptrack is now in kdereview, I'd like to invite everyone to review my
> > > code at the new location:
> > >
> > > git clone kde:heaptrack
> >
> > Doesn't link here.
> >
> > Linking CXX executable heaptrack_print
> > CMakeFiles/heaptrack_print.dir/heaptrack_print.cpp.o: In function `main':
> > heaptrack_print.cpp:(.text+0x3f4f): undefined reference to
> > `boost::program_options::options_description::m_default_line_length'
>
> <snip>
>
> Can you update and try again? I think I was missing the REQUIRED in the
> find_package statements, which should have triggered an early-return. You
> don't have Boost installed, do you?
>
> Hm, but thinking about it - you seem to have the headers around, otherwise
> it wouldn't even get that far. So something is fishy with your system,
> compared to mine. Can you debug this a bit please?
On Ubuntu it's the same, the libboost-program-options1.55-dev pkg just
contains the .a and .so. Headers are part of the (generic) libboost1.55-dev
pkg.
With the REQUIRED in place it errors out early now. => All fine.
> make heaptrack_print VERBOSE=1 |& tee make_log.txt
> -> esp. the linker invocation would be interesting, it should be something
> like
>
> /usr/lib/icecream/libexec/icecc/bin/g++ -std=c++11 -O2 -g -DNDEBUG
> CMakeFiles/heaptrack_print.dir/heaptrack_print.cpp.o -o heaptrack_print -
> rdynamic -lboost_iostreams -lboost_program_options
>
> If this does not include the -lboost_* stuff, can you try to add
>
> message(FATAL_ERROR "Boost libs: ${Boost_LIBRARIES}")
>
> below the find_package in the CMakeLists.txt? What CMake do you use, btw.?
> And maybe play with the following options, and toggle them on/off:
>
> set(Boost_USE_STATIC_LIBS OFF)
> set(Boost_USE_MULTITHREADED ON)
> set(Boost_USE_STATIC_RUNTIME OFF)
>
> (see
> http://stackoverflow.com/questions/6646405/how-do-you-add-boost-libraries-i
> n-cmakelists-txt)
>
> Bye
>
> >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe
> >> <<
--
Kevin Funk | kf...@kde.org | http://kfunk.org
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
