Strange tuple in initial equation in Modelica.Blocks.Examples.Filter

[Peter Aronsson]

Modelica.Blocks.Examples.Filter returns this equation in the flat model:

(CriticalDamping.r, {}, {}, {}) = Modelica.Blocks.Continuous.Internal.Filter.roots.lowPass({CriticalDamping.cr[1], CriticalDamping.cr[2], CriticalDamping.cr[3]}, {}, {}, CriticalDamping.f_cut);

Shouldn't this instead be:

(CriticalDamping.r, , , ) = Modelica.Blocks.Continuous.Internal.Filter.roots.lowPass({CriticalDamping.cr[1], CriticalDamping.cr[2], CriticalDamping.cr[3]}, {}, {}, CriticalDamping.f_cut);

[Adrian Pop]

I'll have a look.

[Adrian Pop]

Is not strange.
It comes from equation:

        if filterType == Modelica.Blocks.Types.FilterType.LowPass then
          (r, a, b, ku) = Internal.Filter.roots.lowPass(cr, c0, c1, f_cut);

and for component:

        Modelica.Blocks.Continuous.Filter CriticalDamping(
          analogFilter = Modelica.Blocks.Types.AnalogFilter.CriticalDamping, 
          normalized = normalized, 
          init = init, filterType = filterType, 
          order = order, 
          f_cut = f_cut, 
          f_min = 0.8 * f_cut);

filterType is LowPass which means that equation will be selected.

components a, b, ku are arrays of 0 size, which mean they disappear from the DAE.

        parameter Real[na] a(each fixed = false);
        parameter Real[na] b(each fixed = false);
        parameter Real[na] ku(each fixed = false);

I don't know if it would be better to put them as empty instead of {}.

[Peter Aronsson]

Yes, I realized that when I wrote the ticket, but wouldn't it be nicer from a backend perspective if those were rewritten to "empty slots" instead, which is standard Modelica? That way, the backend doesn't have to deal with all possible empty arrays that could be in there, e.g.

{}, {{}}, {{{}}}

And the backend still needs to take care of the case

(a,,) = foo(b,c)

, so it would simplify the backend alot.

This is not supercritical for us right now, since I've made a workaround for now, but all attempts at simplifying stuff are good, I think ;-)

[Adrian Pop]

I think that we should actually keep all zero size components and zero size equations/algorithms in the DAE and they can be removed by the back-end.
What do you think about this?

[Martin Sjölund]

Postponed until 1.9.1