We have just described what a problem is conceptually. We now need a concrete way to describe how a problem is to be specified by the user. Ideally, we would like users already using scientific software packages from C or Fortran to be able to switch to NetSolve with no modification to their code. From this viewpoint, when describing a problem as in the preceding subsection, we also describe what we call its format. This is in effect describing what the calling sequence to NetSolve for this problem should be from C or Fortran. Moreover, a problem can have different calling sequences, and the user can chose between them.