> I think groff's rationale for requiring user-entry of italic > correction is sound. Unlike pair-wise kerning, there are as > many places where italic correction isn't desirable as there > are ones where it is.
But Gunnar and Dave are correct in claiming that it shouldn't be necessary to request it again and again for the same character combination. Specifying it once should be enough. I guess it's only because the necessity for italic correction arises not too often (compared to kerning within the same font) that the "90% is good enough" approach works.