Le forum de XCAS

Hi!

The output from 'lvar' may sometimes be counterintuitive (or there is simply a bug). For example, the commandreturns [exp(x),exp(1)]. According to the help index, 'lvar' returns a list of variables with rational dependence, but the second element is a real constant...

Yes, but this is still a generalized variable, I mean it behaves like a normal variable if you want to make polynomial or rational exact operations on an expression.
You can run evalf on the output of lvar and lvar again to avoid this.