On Tue, Aug 12, 2003 at 04:41:25PM +0200, Cyrille Chepelov wrote: > Was the intent to have the connection points not on the centreline but on > the two solid lines? Then you need to have two connpoint_lines instead of > one -- you should then have a look at the "GRAFCET - Vergent" object (hmmm > -- I'm not sure I'd do all I did in the vergent the same way today. Ahem.) I had several intents. Yes, the one you described was one of them but it didn't feel quite... natural. Another intent was to have the second line be ortogonal to the first line (meaning fun operations to produce sinus/cosinus stuff :) but I've dropped that (too much overhead) in favor of my current implementation: second line based on the quadrant (well, quadrant + 45degrees). Much easier to implement and more clean too imo. > The only thing you'd break by pushing the rendering code into the renderers > is the renderer binary interface -- which is OK, it's been already broken > since 0.91 IIRC. On the other hand, the "double line" style becomes > accessible to other object types (one obvious candidate would be the GRAFCET > vergent -- don't ask me why I didn't make the double line style back > then). And how about arrows? Wouldn't this make things more difficult? Arrows are already a PITA for double-lines, and I've been thinking about how to implement them correctly (for instance split arrows into two half arrows and assign an arrow to each member of the double line). > However, I disagree that, should this indeed go to the renderers, this > should be another style just like dashed. I think single/double/more complex > transversal styles is orthogonal to the solid/dotted/dashed/more complex > longitudinal styles we have right now (try to challenge a sentence which says > "transversal is orthogonal to longitudinal, eh? ;-) ) Give me a year or ten to interprete this one, okay? :) Sven Vermeulen -- Save some animals, eat a vegetarian.