Backend adds = 0.0 bindings for no good reason

[Jens Frenkel]

For the model Modelica.Mechanics.MultiBody.Examples.Loops.EngineV6_analytic
the frontend evaluates
engine.cylinder4.jointRRP.jointUSP.prismatic.positiveBranch:PARAM(fixed = false final = true ) .Boolean type

to a real value instead of a boolean

[Jens Frenkel]

• add workaround in r12679

[Martin Sjölund]

Hmmm. Not a pure front-end issue since this is not Ceval'ed in the front-end...

engine.cylinder4.jointRRP.jointUSP.prismatic.positiveBranch = Modelica.Mechanics.MultiBody.Joints.Internal.PrismaticWithLengthConstraint.selectBranch(engine.cylinder4.jointRRP.jointUSP.prismatic.length, {engine.cylinder4.jointRRP.jointUSP.prismatic.e[1], engine.cylinder4.jointRRP.jointUSP.prismatic.e[2], engine.cylinder4.jointRRP.jointUSP.prismatic.e[3]}, engine.cylinder4.jointRRP.jointUSP.prismatic.s_offset + engine.cylinder4.jointRRP.jointUSP.prismatic.s_guess, {engine.cylinder4.jointRRP.jointUSP.prismatic.r_a[1], engine.cylinder4.jointRRP.jointUSP.prismatic.r_a[2], engine.cylinder4.jointRRP.jointUSP.prismatic.r_a[3]}, {engine.cylinder4.jointRRP.jointUSP.prismatic.r_b[1], engine.cylinder4.jointRRP.jointUSP.prismatic.r_b[2], engine.cylinder4.jointRRP.jointUSP.prismatic.r_b[3]});

It must be either inline, ceval or simplify called from the back-end. I'll have a closer look.

[Martin Sjölund]

lowerKnownVar is executed for this variable, and has an extra = 0.0 binding added to it. I will see what happens if I remove this automatic adding of bindings...

[Martin Sjölund]

The testcase I will add (current omc simulates this as i=1, r=1)

model M
  parameter Integer i(start=1,fixed=false);
  Real r = i;
initial equation
  i = 2;
end M;

[Martin Sjölund]

r12686 fixes this problem. My pretty testcase could not be added because we still do not support initial equations.

If the rest of the backend wants fixed=true parameters to have the binding added this can be done, but requires some more work (check fixed=true, check input type, etc).