Thank you very much for your answer, again.
I tried what you did and it works. The problem is that the project I try to
interface is much more complicated than this (and I cannot modify it) but still
it helps.
I will try to do the same thing with the real class GRelation. But for this, I
need to use another C compiler than the standard one (I have to use the last
version of gcc) and to modify the passed options (not only add some).
I read documentation about defsystem but I didn't find how to do it ? Could you
please tell me where I could find that information ?
Best regards,
Sascha Van Cauwelaert
Le 24 nov. 2011 à 20:55, Martin Simmons a écrit :
> I've tried to replicate what you are doing, by adding the following hacks to
> your function in a file called gtest.cpp:
>
> #include <iostream>
> #include <sstream>
> #include <string>
>
> class GRelation {
> private:
> int _arity;
> public:
> int arity() { return _arity; };
> GRelation(int arity) { _arity = arity; };
> };
>
> std::ostream& operator<<(std::ostream& out, GRelation &g)
> {
> out << "<GRelation" << g.arity() << ">";
> return out;
> };
>
> I then passed it through SWIG 1.3.31:
>
> swig -cffi -c++ -module gtestswig gtest.cpp
> cat gtest.cpp gtest_wrap.cxx > gtest_code_and_wrap.cpp
>
> Beware that I've not read much C++ or SWIG documentation, so the above might
> be somewhat bogus!
>
> Following that, it ran without errors with CFFI 0.10.6 on Mac OS X 10.6.8 like
> this:
>
> (asdf:initialize-source-registry
> '(:source-registry
> (:directory "~/cffi_0.10.6")
> (:directory "~/alexandria")
> (:directory "~/babel_0.3.0")
> (:directory "~/trivial-features_0.1")
> :inherit-configuration))
> (asdf:load-system :cffi)
> (compile-file "gtestswig" :load t)
> (defsystem gtestswig (:c-default-options "-lstdc++")
> :members (("gtest_code_and_wrap.cpp" :type :c-file)))
> (compile-system 'gtestswig :force t :load t)
> (test)
>
> --
> Martin Simmons
> LispWorks Ltd
> http://www.lispworks.com/
>
>
>
>
>>>>>> On Wed, 23 Nov 2011 16:20:51 +0100, Sascha Van Cauwelaert said:
>>
>> Thank you for your answer.
>>
>> Actually, I have an additional clue, which make me think that it is not
>> related with garbage collection. If, in Lispworks, I use a wrapped function
>> on the pointer to the C object (e.g. to get the arity), it works fine !
>>
>> It seems that I just can't call a function which print some information
>> about the object in a C++ function, wrapped to be used in LispWorks ... If I
>> do it, I get the error :
>>
>> lispworks-personal-6-0-1-macos-universal(49641,0xb0314000) malloc: *** error
>> for object 0x17bfc0f8: pointer being freed was not allocated
>> *** set a breakpoint in malloc_error_break to debug
>>
>> I don't get how it can work in SBCL and not in LispWorks, and I have no clue
>> because I don't know the internal implementation of LispWorks.
>>
>> I would really appreciate a little bit of your help.
>>
>> Regards,
>>
>> Sascha Van Cauwelaert
>>
>>
>> Le 22 nov. 2011 à 21:16, Martin Simmons a écrit :
>>
>>>>>>>> On Mon, 21 Nov 2011 10:28:56 +0100, Sascha Van Cauwelaert said:
>>>>
>>>> Hello everybody,
>>>>
>>>> I am using SWIG to generate CFFI code to be used into LispWorks 6. The
>>>> interface is between C++ and Common Lisp. Then, I try to call that function
>>>> from LispWorks :
>>>>
>>>> std::string test() {
>>>> freopen ("/tmp/mystdout.txt","w",stdout);
>>>> freopen ("/tmp/mystderr.txt","w",stderr);
>>>> std::cout << "Redirected stdout." << std::endl;
>>>> std::cerr << "Redirected stderr." << std::endl;
>>>>
>>>> GRelation testgr(3);
>>>>
>>>> std::cout << "test ! " << std::endl;
>>>> std::cerr << "test cerr! " << std::endl;
>>>> std::cout << testgr.arity() << std::endl;
>>>> std::cerr << testgr.arity() << std::endl;
>>>> std::cout << "test after! " << std::endl;
>>>> std::cerr << "test cerr after! " << std::endl;
>>>> std::stringstream os;
>>>> os << testgr << std::endl ;
>>>> return os.str();
>>>> }
>>>>
>>>> As you can see, I just create one object and print lots of stuff, including
>>>> an object attribute. Everything I print before trying to print the
>>>> attribute
>>>> prints ok in the redirected outputs. The attribute is not printed and
>>>> nothing else afterwards. I also get the following error when I add testgr
>>>> to
>>>> the stringstream os :
>>>>
>>>> lispworks-personal-6-0-1-macos-universal(52881,0xb0314000) malloc: ***
>>>> error for object 0x17be20f8: pointer being freed was not allocated
>>>>
>>>> This is what happen in LispWorks. If I compile directly in C++, no
>>>> problem. If I use SBCL (other Common Lisp implementation) instead of
>>>> LispWorks, everything works correctly ! The problem is that I have to use
>>>> LispWorks ...
>>>>
>>>> Does anybody has had a similar problem ? I guess I am not the only one who
>>>> tried to call some C++ from LispWorks. Any clue is very welcome. I think
>>>> maybe the problem can be linked to the garbage collection from LispWorks
>>>> but
>>>> I don't know much about this.
>>>>
>>>> I thank you in advance for your help, I am really stuck here.
>>>
>>> I don't see how garbage collection can cause it, because malloc is not
>>> affected by the LispWorks GC (unless something is doing bad things with
>>> finalization routines).
>>>
>>> --
>>> Martin Simmons
>>> LispWorks Ltd
>>> http://www.lispworks.com/
>>>
>>> _______________________________________________
>>> cffi-devel mailing list
>>> cffi-devel@common-lisp.net
>>> http://lists.common-lisp.net/cgi-bin/mailman/listinfo/cffi-devel
>>
>
> _______________________________________________
> cffi-devel mailing list
> cffi-devel@common-lisp.net
> http://lists.common-lisp.net/cgi-bin/mailman/listinfo/cffi-devel
_______________________________________________
cffi-devel mailing list
cffi-devel@common-lisp.net
http://lists.common-lisp.net/cgi-bin/mailman/listinfo/cffi-devel
