This is not a really issue: I think for gmin it is correct not to have a static, fixed value which is set one times and then never changed. The problem has two sides:
All other measures, like "shrink" in this case, are static and should never change in simulation. So they are not problematic.
The true problem with "gmin" is not the define of a macro, it is the place where it is used.
warning[L015]: call to $simparam in a constant is evaluted before the simulation
--> /home/dietmar/Projects/spice/Verilog-A/VA-Models/code/r3_cmc/vacode/r3_cmc_macros.include:175:20
175
`define SIMPARSHRI $simparam("shrink",0.0)
^^^^^^^^^^^^^^^^^^^^^^^ call to $simparam in a constant
= help: the value of paramaeters like "gmin' or "sourceScaleFactor" may vary between iterations
= variant_const_simparam is set to warn by default
This is not a really issue: I think for gmin it is correct not to have a static, fixed value which is set one times and then never changed. The problem has two sides: