Parameter has neither value nor start value, and is fixed

[Christian Schubert]

Hi,

I always wondered about those strange warnings and created a simple testmodel, which reproduces this error (see attachment)
Basically, if you have

model A
  parameter Integer n = 2;
  parameter Real a[n](each fixed = false);
initial equation
  a = fill(1.0, n);
end A;

you'll get the warning although (from my point of view) everything should be fine.
Morover, the warning is misleading.
Yes, the parameter doesn't have value nor start value. But where does it say it is fixed?
I think that case might even be handled correctly by Lennart's new initilization?

[Adrian Pop]

I've seen this too in some cases. Is a bug.
In some cases even if you have a binding it seems it gives the message.
I'll look into this when time permits.

Cheers,
Adrian Pop/

[Christian Schubert]

Is it only the warning that is wrong, or is there a chance of something worse caused by that?

[Jens Frenkel]

I guess this task should have a higer prioriti. Because of final parameteres with no binding the backend get in trouble. For example if the final parameter is replaced with its default falue index reduktion will fail sometimes.

[Adrian Pop]

As far as I've seen only the warning is wrong.
The bindings do appear in the flat Modelica code.
Also, in some cases:

• we move the bindings from the parameters into the equation section
• we instantiate the component without the binding just to get its type

and these might trigger the wrong warning.

Cheers,
Adrian Pop/

[Jens Frenkel]

Good to know

we move the bindings from the parameters into the equation section 

I am sure this is not handled in the backend if the parameter keeps fixed=true.

[Jens Frenkel]

For example the model Modelica.Electrical.Spice3.Examples.FourInverters has a binding for

mp1.c1

final parameter Mos1.Mos1Calc c1 = Mos.mosCalcInitEquations(

        p,

        C,

        vp,

        m) "Precalculated parameters"; 

but the binding is neither at the parameter (it is a record) nor in the equation section.

[Jens Frenkel]

see also Ticket #1917

[Adrian Pop]

Ok. This seems a bit serious.
I'll have a look at it asap.

Cheers,
Adrian Pop/

[Adrian Pop]

Fixed in revision: r14351 [with additional fixes in r14354, r14355, r14357, r14365, r14366 :)]. The model in this ticket is fixed, but I think there are more left.