Problem with duplicate symbols when using gfortran 4.6 that are not present when using gfortran 4.5
Arjen Markus
arjen.markus895@gmail.com
Mon Nov 14 10:28:00 GMT 2011
Hi Tobias,
I will do my best to assemble some more information for you. The whole issue is
rather confusing and as I do have a number of versions of GCC installed, there
might be some interference.
2011/11/14 Tobias Burnus <burnus@net-b.de>:
> Hello,
>
>
> In principle, there was no change between 4.5 and 4.6 with regards to those
> symbols. Are you by chance mixing two differently compiled object files with
> either static libgfortran linking or using some fancy option like -mrtd?
>
The method I use when rebuilding PLplot is:
- Clean the entire directory or create a new one
- Run CMake with the (hopefully) correct environment set up
- Run make to build everything according to the generated Makefiles
- Run the example programs
> You wrote elsewhere that you use equation.com's GCC compiler. I think you
> should consider trying another one. I don't have any experience with their
> compiler myself, but I know that equation.com's MinGW64 compiler crashes
> when compiling the polyhedron benchmark while a build from mingw64.sf.net
> does not.
>
Good to know that.
One option I still have to investigate is whether the compilers that come with
MinGW pose the same difficulty or not. Now I use the separate installation.
>> I will try and concoct a small example, if there should be a need for
>> that. My main goal is to be able to compile and link these examples
>> _without_ any special options. I hope my decription is clear enough.
>
> I think a small example would be great. (After you have checked that it is
> not due to the equation.com issue.) Additionally, it would be helpful to
> know whether there are multiple versions of GCC installed and which (compile
> and) link options do you use. Do you create first a library (static? DLL?)
> and then link it? Or do simply link simple .o files consisting of the
> "library" and the test program?
>
I will try and do so, but it will be a few days before I can get around to that.
Regards,
Arjen
More information about the Fortran
mailing list