hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/ | GSoC2018: https://wp.me/p4pxJf-k1
hkaiser has quit [Quit: bye]
K-ballo has quit [Quit: K-ballo]
anushi has quit [Read error: Connection reset by peer]
anushi has joined #ste||ar
anushi has quit [Read error: Connection reset by peer]
anushi has joined #ste||ar
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
anushi has quit [Ping timeout: 245 seconds]
anushi has joined #ste||ar
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
twwright_ has joined #ste||ar
anushi has quit [Ping timeout: 245 seconds]
anushi has joined #ste||ar
anushi has quit [Ping timeout: 265 seconds]
anushi has joined #ste||ar
jaafar_ has joined #ste||ar
jaafar has quit [Ping timeout: 248 seconds]
anushi has quit [Remote host closed the connection]
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
nikunj has joined #ste||ar
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
<jakub_golinowski>
Hello I want to run opencv tests with hpx_main.hpp included to the test_main.cpp for the tests in opencv
<jakub_golinowski>
When I build opencv (with tests) in debug mode it is succesfull, however when I want to build them in the release mode I get the following error:
<jakub_golinowski>
/usr/bin/ld: /home/jakub/hpx_repo/build/hpx_11/release/lib/libhpx_init.a(hpx_main_argc_argv.cpp.o): undefined reference to symbol '_ZN3hpx6throwsE'
<jakub_golinowski>
/home/jakub/hpx_repo/build/hpx_11/release/lib/libhpx.so.1.2.0: error adding symbols: DSO missing from command line
<jakub_golinowski>
collect2: error: ld returned 1 exit status
<jakub_golinowski>
modules/core/CMakeFiles/opencv_perf_core.dir/build.make:900: recipe for target 'bin/opencv_perf_core' failed
<jakub_golinowski>
make[1]: *** Waiting for unfinished jobs....
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
diehlpk has joined #ste||ar
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
diehlpk has quit [Ping timeout: 255 seconds]
K-ballo has joined #ste||ar
diehlpk has joined #ste||ar
anushi has quit [Ping timeout: 256 seconds]
hkaiser has joined #ste||ar
diehlpk has quit [Ping timeout: 256 seconds]
anushi has joined #ste||ar
twwright_ has joined #ste||ar
<nikunj>
hkaiser, yt?
<hkaiser>
here
<nikunj>
I was thinking to integrate the code with hpx
<nikunj>
the implementation without the global object
<nikunj>
It is complete, but I need your help regarding cmake integrations
<nikunj>
should I add an option in add_hpx_executable to link with the static library?
<nikunj>
or should I do it in hpx_setup_target?
<nikunj>
or should it be implemented in a different way. I'm confused as to how I should integrate it.
<hkaiser>
nikunj: this is code that needs to be linked with executables, yes?
<nikunj>
yes, at link time I need to link it with the static library (that I'll create from my code) along with a compiler option -Wl,-wrap=__libc_start_main
<hkaiser>
either add it to the static hpx_init library, or create your own static library and make sure it gets linked with all executables, similarly to hpx_init
<hkaiser>
both cases will require changes to hpx_setup_target
<nikunj>
Since linking should only be done when using hpx_main.hpp
<hkaiser>
uhh, yes
<hkaiser>
this is more problematic
<hkaiser>
but I think a configuration option could do the trick
<nikunj>
yes, I was thinking the same. Something like HPX_FROM_MAIN
<hkaiser>
that could be used to control the linking and could generate a pp constant that triggers proper compilation if hpx_main is used
<hkaiser>
otoh, what happens if hpx_main.hpp is not used and you link with your code anyways?
<hkaiser>
can you make it so that in this case your code is not executed?
<nikunj>
The code will break as hpx_init will be called twice
<hkaiser>
nanashi55: also, we call all of our configuration options HPX_WITH_FOO and the corresponding pp constant HPX_HAVE_FOO
<hkaiser>
nikunj: ^^
<hkaiser>
gtg, sorry
<nikunj>
ok, I will try to think of a way.
<nikunj>
Through which the code will work even when linked with the file
<M-ms>
jakub_golinowski: I don't know what might be wrong, but you could check what your ${HPX_LIBRARIES} contains
<jakub_golinowski>
ok
<M-ms>
you're including hpx_main in the test_core.cpp file now (or another similar test file)?
nikunj97 has joined #ste||ar
<jakub_golinowski>
for now I included it only to perf_main.cpp and test_main.cpp from the core module
<M-ms>
I will have more time from tuesday onwards, so try to focus now on checking what other tests fail or you can start looking into qt or webcam capture already
<M-ms>
ok
<M-ms>
alternatively see if you can do hpx_setup_executable one of the test executable targets (not a good solution, but to see if anything is different)
<jakub_golinowski>
and libhpx.so.1.2.0 is not one of them
<jakub_golinowski>
that seems like a good direction
<jakub_golinowski>
but why is the debug mode properly compiling is still a mystery to me
<M-ms>
I don't know either, but I do know that the linker (or compiler?) may be removing unused symbols in release mode
<jakub_golinowski>
ok seems like a good direction as well
<M-ms>
libhpx is not, but hpx is, and when providing libraries for the linker the lib prefix is left out, so it does actually find libhpx
<jakub_golinowski>
and is not an issue that the suffix is missing (.so.1.2.0)
<jakub_golinowski>
?
<M-ms>
jakub_golinowski: it's not actually complaining about finding libhpx, it's not finding hpx::throws
<M-ms>
the extension is also automatically tried
<M-ms>
(it actually checks for libhpx.so, which is a symlink to libhpx.so.1.2.0)
<jakub_golinowski>
M-ms, well the error is a bit confusing for me. The first line seems to mean that within libhpx_init.a there was undefined reference to hpx::throws
<jakub_golinowski>
However the second line suggests that libhpx is missing from command line (I gather in this case from cmake build system)
<M-ms>
jakub_golinowski: yep, you might be right
<M-ms>
You could try playing around with the order of hpx and hpx_init in HPX_LIBRARIES, but to be honest I don't know yet what the problem is
<M-ms>
If someone else has ideas I'd love to hear
twwright_ has quit [Quit: twwright_]
anushi has quit [Ping timeout: 248 seconds]
<heller>
nikunj97: the answer could be target specific defines that get passed down to the dependency
<nikunj97>
hkaiser, could you please elaborate
<nikunj97>
hkaiser: do you mean I should pass a parameter to add_hpx_executable through which cmake can identify if it needs to link with the static library?
<M-ms>
jakub_golinowski: would you be free around 8 on tuesday for a call? I'm flexible after tuesday evening though, so we can do it on wednesday or thursday as well and you've kept me well up to date
<M-ms>
it's mostly if you'd like to discuss things in more detail
<jakub_golinowski>
You mean 8 am?
<M-ms>
8 in the evenint
<jakub_golinowski>
Then I prefere Wednesday as in the evening on Tuesday I am out
<M-ms>
Good, let's do Wednesday then, we can do it a bit earlier then
twwright_ has joined #ste||ar
parsa[[w]] has joined #ste||ar
parsa[w] has quit [Ping timeout: 265 seconds]
twwright_ has quit [Quit: twwright_]
quaz0r has quit [Ping timeout: 248 seconds]
jakub_golinowski has quit [Remote host closed the connection]