Thursday, April 10, 2003 Hans Hagen wrote: HH> \startuseMPgraphic{mpos:par:columnset} HH> path q ; picture r ; color framecolor, backgroundcolor ; HH> cornerradius := 5pt ; offset := 7.5pt ; HH> framecolor := red ; framethickness := 2pt ; HH> backgroundcolor := lightgray ; HH> for i := 1 upto nofmultipars : HH> multipars[i] := multipars[i] enlarged offset ; HH> if multilocs[i] = 1 : % top HH> q := multipars[i] bottomenlarged (4cornerradius) ; HH> elseif multilocs[i] = 2 : HH> q := multipars[i] bottomenlarged (4cornerradius) ; HH> q := q topenlarged (4cornerradius) ; HH> else: HH> q := multipars[i] topenlarged (4cornerradius) ; HH> fi ; HH> q := q smoothed (2cornerradius) ; HH> r := image HH> ( fill q withcolor backgroundcolor ; HH> draw q withcolor framecolor HH> withpen pencircle scaled framethickness ; ) ; HH> clip r to boundingbox multipars[i] ; HH> addto currentpicture also r ; HH> endfor ; HH> \stopuseMPgraphic HH> watch the usage of "enlarged" and the "smoothed" I think I see; instead of drawing only the needed parts you draw the whole frame and then cut the unwanted parts off. Is this more efficient than drawing only the needed parts? And what about the way to determine when there is a split or not? Could you please put a boolean that reveals it, in the next beta? (even if I'm still convinced that using a multilocs of 0 would be enough ... ok, it would break backwards compatibility, but how many people are using this mechanism already?) -- Giuseppe "Oblomov" Bilotta