[BE] Fixing differentiation of synchronous states

Belonging to [master]:
  - #2269
  - OpenModelica/OpenModelica-testsuite#874

Author: Willi Braun

Compiler/BackEnd/Differentiate.mo

@@ -1336,10 +1336,17 @@ algorithm
 
       String s1, s2, serr, matrixName, name;
 
-    case (e as DAE.CALL(path=Absyn.IDENT(name = "previous")), _, _, _, _) equation
-      tp = Expression.typeof(e);
-      zero = Expression.makeZeroExpression(Expression.arrayDimension(tp));
-    then (zero, inFunctionTree);
+    /* with previous are the actaully states marked in synchronous */
+    case (e as DAE.CALL(path=Absyn.IDENT(name = "previous"), expLst = {DAE.CREF(componentRef=cr, ty=tp)}),
+           _, BackendDAE.DIFFINPUTDATA(independenentVars=SOME(timevars),matrixName=SOME(matrixName)),
+          BackendDAE.GENERIC_GRADIENT(), _) equation
+
+      cr = ComponentReference.makeCrefQual(DAE.previousNamePrefix, tp, {}, cr);
+      (_::_, _) = BackendVariable.getVar(cr, timevars);
+      cr = createSeedCrefName(cr, matrixName);
+
+      res = DAE.CREF(cr, tp);
+    then (res, inFunctionTree);
 
     case (DAE.CALL(path = path as Absyn.IDENT(name = "der"),expLst = {e},attr=attr), _, _, BackendDAE.DIFFERENTIATION_TIME(), _)
       then

Compiler/BackEnd/SymbolicJacobian.mo

@@ -1654,6 +1654,7 @@ continuous states: der(x) > dependent and x > independent
 clocked states: previous(x) > independent and x > dependent
 "
   input list<BackendDAE.Var> independentVars;
+  input Boolean createpDerStates = true;
   output list<BackendDAE.Var> outVars = {};
   output list<DAE.ComponentRef> outCrefs = {};
 protected
@@ -1664,9 +1665,12 @@ algorithm
       var :=  BackendVariable.createClockedState(v);
       outVars := var::outVars;
       outCrefs := var.varName::outCrefs;
-    else
+    elseif createpDerStates then
       outVars := BackendVariable.createpDerVar(v)::outVars;
       outCrefs := v.varName::outCrefs;
+    else
+      outVars := v::outVars;
+      outCrefs := v.varName::outCrefs;
     end if;
   end for;
   outVars := listReverse(outVars);
@@ -2000,7 +2004,7 @@ algorithm
 
       BackendDAE.Shared shared;
       BackendDAE.Variables  globalKnownVars, globalKnownVars1;
-      list<BackendDAE.Var> diffedVars "resVars", seedlst;
+      list<BackendDAE.Var> diffedVars "resVars", seedlst, indepVars;
 
       DAE.FunctionTree funcs;
 
@@ -2013,9 +2017,20 @@ algorithm
 
         comref_vars = List.map(inDiffVars, BackendVariable.varCref);
         seedlst = List.map1(comref_vars, createSeedVars, inName);
+        indepVars = createInDepVars(inDiffVars, false);
+
+        if Flags.isSet(Flags.JAC_DUMP) then
+          print("Crete symbolic Jacobianis from:\n");
+          print(BackendDump.varListString(indepVars, "Independent Variables"));
+          print(BackendDump.varListString(diffedVars, "Dependent Variables"));
+          print("Basic equation system:\n");
+          print(BackendDump.equationListString(BackendEquation.equationSystemsEqnsLst(reducedDAE.eqs), "differentiated equations"));
+          print(BackendDump.varListString(BackendVariable.equationSystemsVarsLst(reducedDAE.eqs), "related variables"));
+          print(BackendDump.varListString(BackendVariable.varList(reducedDAE.shared.globalKnownVars), "known variables"));
+        end if;
 
-        // Differentiate the ODE system w.r.t states for jacobian
-        (backendDAE as BackendDAE.DAE(shared=_), funcs) = generateSymbolicJacobian(reducedDAE, inDiffVars, inDifferentiatedVars, BackendVariable.listVar1(seedlst), inStateVars, inInputVars, inParameterVars, inName);
+        // Differentiate the eqns system in reducedDAE w.r.t. independents
+        (backendDAE as BackendDAE.DAE(shared=_), funcs) = generateSymbolicJacobian(reducedDAE, indepVars, inDifferentiatedVars, BackendVariable.listVar1(seedlst), inStateVars, inInputVars, inParameterVars, inName);
         if Flags.isSet(Flags.JAC_DUMP2) then
           print("analytical Jacobians -> generated equations for Jacobian " + inName + " time: " + realString(clock()) + "\n");
         end if;