On Mon, Nov 03, 2003 at 02:04:33PM +0100, Lars Gullik Bj�nnes wrote: > Andre Poenitz <[EMAIL PROTECTED]> writes: > > | The approach is fine with me. And as this is more or less the same > | scheme as used in src/m* this would be a major step forward on inset > | unification. > > I must admit that I am not a fan of hiding _all_ arguments inside a > huge object (a we-might-need-this-store-it-here object)
I know. That's why I did not propose to move the non-const ostream argument to the struct as well and I even would accept such a change for mathed if the resulting code was as simple as the one currently used there. I am not feeling too comfortable with that 'big argument struct' myself, but it has turned out to be simple, maintainable and fast. Andre'
