differentiate delay expr

[Willi Braun]

Differentiation of delay expression is not implementet, yet.
All delay expression get an index in BackendDAECreate,
but we don't know the next index while differentiation.

[Andreas Heuermann]

Example for failing model:

model miniDelay
  Real x(start=0, fixed=true);
  Real entryTime;
equation
  entryTime = delay(time, 0.1, 0.1);
  der(x) = entryTime;

annotation(
    __OpenModelica_commandLineOptions = "-d=newInst --generateSymbolicJacobian");

end miniDelay;

[Andreas Heuermann]

If the delay operator is only depending on time (like in miniDelay) we could simply set the derivative to zero. Only if the operator depends on states we need to actually compute a derivative.

model stateDelay
  Real x,y(start=0, fixed=true);
equation
  x = sin(time)+y;
  der(y) = delay(x, 0.1, 0.1);
end stateDelay;

Problem in this case: We need to store the values of the expression for each seed. Maybe a simple solution would be to use differential quotients to approximate the derivative in stead of storing values for all directional derivates for the Jacobian.