@stefanoaz the tcleval(...) argument is forwarded by xschem to the tcl parser for evaluation, after substituting the "@var" xschem variables. Normally xschem does only variable substitution and does not do a tcl evaluation on parameters that are not enclosed in a tcleval(...) construct, this is for performance reasons.
I think Cadence does something similar, perhaps in a more comprehensive way (may be resolving recursive param definitions). If you know the equation (may be more accurate and more complex than the simple
rho*L/W
) you can add the computed R to the generic resistors too.