Hi, Our FAQ [1] was giving the impression that tree-traversals for complex operations are best being written as functions that recursively call themselves and contain lots of silly little ifs that distinguish between the different object types. This was rubbish. I have polished up an old example (an actual prototype of a simplifier in a certain domain) and rewritten the items "How do I simplify things like sin^2+cos^2 to 1?" and "How do I write a tree-traversal algorithm?" from scratch to take a more elegant and more powerful point of view. If (after careful study, please) people find something unclear therein or even actual errors, I would very much like to hear about that. I am confident that people writing applications in GiNaC will benefit greatly from these these two rewritten sections. For the curious, the old FAQ is still available [2]. enjoy! -richy. [1] <http://www.ginac.de/FAQ.html> [2] <http://www.ginac.de/FAQ_old.html> -- Richard B. Kreckel <Richard.Kreckel@Uni-Mainz.DE> <http://wwwthep.physik.uni-mainz.de/~kreckel/>