Hi! Three weeks ago I've removed the 'level' argument of the eval() functions (on the C++11 branch). After all, if an object derived from basic is evaluated, it must evaluate its own structure, but not the one of expressions it contains - these are already evaluated. (This had become possible after elusive bugs in the automatic evaluation were fixed.) Now I wonder what the 'level' argument of evalf() and normal() are good for. The situation is slightly less clear than with eval(): After all, it *might* be useful to evalf/normal the expression's top level only without recursing down the directed acyclic graph. The question is: *Is* it a realistic use case to evalf/normal the expression to some level only? I do not see it. If anyone cares for the 'level' argument of evalf and/or normal, please respond and explain what it is used for. Otherwise, the 'level' argument will be removed by the end of this month. (Fact: This removal does not affect any of the regression tests.) All my best, -richy. -- Richard B. Kreckel <http://in.terlu.de/~kreckel/>