We consider general linear methods using the formulation introduced by Burrage
and Butcher [1] in 1980. Even though this formulation includes a
large class of methods, it has proved difficult to find genuinely new methods
with practical advantages. The methods with inherent Runge-Kutta stability,
however, seem to have considerable potential for the identification of
practical and efficient methods. In this talk we will discuss several
implementation issues, such as error estimation, stepsize change, continuous
solutions and variable order for these methods.