3.4 Using Math.h++ with LAPACK.h++

If you do not use Rogue Wave's LAPACK.h++ library, please skip this section. If you do use LAPACK.h++, and you plan to use Math.h++ with LAPACK.h++ and as an independent library, you must build two versions of Math.h++. We recommend that you install and build the two versions of Math.h++ in different Rogue Wave root directories.

To avoid problems, we recommend that you install and build the stand-alone version of the Math.h++ library, and then build and test the examples to insure that it works properly. Next, install the version of Math.h++ that will be used with LAPACK.h++ in a different Rogue Wave root directory than the stand-alone version of the library. After you build the Math.h++ library, install and build the LAPACK.h++ library. Finally, build and run the Math.h++ and the LAPACK.h++ examples to test the libraries.

Section 3.2 explains how to compile Math.h++. Section 3.3 explains how to compile the examples.The LAPACK.h++ Build Guide explains how to compile LAPACK.h++ and the LAPACK.h++ examples.

As SPM compiles Math.h++, the build process will display questions that determine whether this version will be used with LAPACK.h++. In particular, when SPM builds the Math.h++ examples, you will be asked for the location of the BLA modules you plan to use. This allows SPM to specify that location to the linker.

After you compile Math.h++, its BLA routines will be located in:

<rw_root>\workspaces\<OS+ver>\<COMPILER+ver>\<buildtype>\lib\<libname>

where <libname> is the actual name of the BLA library.

For example, if you use a Unix compiler to build a multithreaded, static BLA library, which does not use the Standard C++ Library or provide debugging, the resulting library name is:

libbla4s.a

The same library built on Windows NT or Windows 95 is named:

bla4s.lib

Section 2.4 describes how libraries are named. Substitute bla for mth to determine the name that the Math.h++ build process will give the BLA library.