Re: [ESPResSo-devel] ESPResSo not linking on freshest commit

[Tristan Bereau]

Dear all,

I've found the source of my compilation error (last email). Espresso
currently does not compile (git: 31c4458e95dba10f8...) if
ELECTROSTATICS is commented out. I suppose the methods qBegin() and
qEnd() in EspressoSystemInterface.cpp would need to be surrounded by
#ifdef ELECTROSTATICS. But I haven't checked whether that works.

Best,
Tristan

On Tue, Jan 21, 2014 at 6:19 PM, Tristan Bereau <address@hidden> wrote:
> Dear Stefan and Florian,
>
> Silly me--thanks for the pointer. I do get another error now, though :-)
>
> EspressoSystemInterface.cpp: At global scope:
> EspressoSystemInterface.cpp:107:71: error: no
> 'SystemInterface::const_real_iterator&
> EspressoSystemInterface::qBegin()' member function declared in class
> 'EspressoSystemInterface'
>  SystemInterface::const_real_iterator &EspressoSystemInterface::qBegin() {
>                                                                        ^
> EspressoSystemInterface.cpp:112:75: error: no 'const
> const_real_iterator& EspressoSystemInterface::qEnd()' member function
> declared in class 'EspressoSystemInterface'
>  const SystemInterface::const_real_iterator &EspressoSystemInterface::qEnd() {
>                                                                            ^
>
> and also some warnings (I'm only showing them here in case they're
> related to the error):
>
> In file included from EspressoSystemInterface.cpp:1:0:
> EspressoSystemInterface.hpp: In constructor
> 'EspressoSystemInterface::EspressoSystemInterface()':
> EspressoSystemInterface.hpp:107:10: warning:
> 'EspressoSystemInterface::m_q_gpu_end' will be initialized after
> [-Wreorder]
>    float *m_q_gpu_end;
>           ^
> EspressoSystemInterface.hpp:98:8: warning:   'bool
> EspressoSystemInterface::m_gpu' [-Wreorder]
>    bool m_gpu;
>         ^
> In file included from EspressoSystemInterface.cpp:1:0:
> EspressoSystemInterface.hpp:10:3: warning:   when initialized here [-Wreorder]
>    EspressoSystemInterface() : m_gpu_npart(0), m_r_gpu_begin(0),
> m_r_gpu_end(0), m_v_gpu_begin(0), m_v_gpu_end(0), m_q_gpu_begin(0),
> m_q_gpu_end(0), m_gpu(false), m_needsParticleStructGpu(false),
> m_splitParticleStructGpu(false)  {};
>    ^
>
>
> I don't have cuda set up on my machine, and that was indeed shown at
> the end of ./configure. Can you reproduce the above-mentioned error?
>
> Best,
> Tristan
>
>
> On Tue, Jan 21, 2014 at 6:09 PM, Stefan Kesselheim
> <address@hidden> wrote:
>> Dear Tristan,
>> it seems the corresponding cpp file was not added to the makefile.
>> This could be related to not rebuilding the makefiles (./bootstrap.sh) or 
>> the fact that it is missing in Makefile.am (in src).
>> Please check both. If it does not vanish by bootstrapping, please ask 
>> Florian Weik, he surely has the right patch for Makefile.am.
>> Cheers
>> Stefan
>>
>> On Jan 21, 2014, at 6:02 PM, Tristan Bereau <address@hidden> wrote:
>>
>>> Dear all,
>>>
>>> I'm posting this on the developer's list, since I'm working with the
>>> latest version (31c4458e95dba10f8...) of ESPResSo. Would anyone know
>>> why I'm getting the following error while linking the code?
>>>
>>> forces.cpp:(.text+0x8c7): undefined reference to `espressoSystemInterface'
>>> forces.cpp:(.text+0x8cc): undefined reference to
>>> `EspressoSystemInterface::update()'
>>> collect2: error: ld returned 1 exit status
>>>
>>> I can't find any obvious problem--the class gets instantiated in
>>> EspressoSystemInterface.hpp and forces.cpp does load that file.
>>>
>>> Any help would be much appreciated.
>>>
>>> Best,
>>> Tristan
>>
>>