[Ipopt-tickets] [Ipopt] #80: configuration failed using cygwin

[Ipopt]

#80: configuration failed using cygwin
--------------------------------+-------------------------------------------
  Reporter:  andreasw           |       Owner:  andreasw
      Type:  defect             |      Status:  assigned
  Priority:  normal             |   Component:  Ipopt   
   Version:  3.4 (C++ Version)  |    Severity:  normal  
Resolution:                     |    Keywords:          
--------------------------------+-------------------------------------------
Changes (by andreasw):

  * owner:  ipopt-team => andreasw
  * status:  new => assigned

Comment:

 Hi Ben,

 Looking at the config.log files it seems that you are using the gcc
 compiler version 3.4.4 and the gfortran compiler 4.4.0.  The error message
 happens because the 3.4.4 C compiler does not know about the Fortran
 runtime libraries from version 4.4.0.  And even if it would find it, it is
 quite likely that the different in major versions would be incompatible.

 I would suggest to use compatible compilers.  Where did you get the 4.4.0
 version?  If you installed that yourself you can probably also install all
 the other compilers in the GCC collection.  By the way, 4.4.0 is as far as
 I know the development version, and it probably not working so well.  I
 would suggest to use 4.3.1 or something that is an official release.

 The problem you observe is independent of the COIN stuff.  After you
 update your compilers, you can verify that mixed-language linking is
 working by putting the following lines into a file called {{{fprog.f}}}

 {{{
       subroutine fprog(a)
       double precision a
       write (*,*) a
       return
       end
 }}}

 and the following into a file called {{{bla.cpp}}}

 {{{
 extern "C" {
   void fprog_(double*);
 }

 int main() {
   double a = 10.;
   fprog_(&a);
 }
 }}}

 You can then compile, link and test it with
 {{{
 gfortran -c fprog.f
 g++ bla.cpp fprog.o -lgfortran
 ./a.out
 }}}

 Unless you get that to work, there is little chance that you can compile
 Ipopt.

 Also, you should not specify the {{{--with-mumps-dir}}} flag; Ipopt will
 find the code there if you put it into that directory following the
 instructions.  Using that flag, Ipopt assume that there is already a
 precompiled version, and instead you want to have Ipopt's Makefiles
 compile MUMPS for you.

 I hope this helps,

 Andreas

-- 
Ticket URL: <https://projects.coin-or.org/Ipopt/ticket/80#comment:1>
Ipopt <http://projects.coin-or.org/Ipopt>
Interior-point optimizer for nonlinear programs.