Variables out of scope

[jlaybar]

This code should return global x and y instead of a and b

Matrix a = Row(0,0);
Matrix b = Row(1,1);

Matrix x = Row(2,2,2);
Matrix y = Row(3,3,3);

Matrix fun1( Matrix q, Matrix p)
{
WriteLn( " q:<<q+?"+NL+"p:<<p+?");
WriteLn( "-------------------------------------------------------------");
WriteLn( " TOL SE VUELVE MAJARA Y ASIGNA MAL LAS VARIABLES
GLOBALES"+NL);
WriteLn( "Asignacion Incorrecta: "+NL+"x:<<x+?"+NL+"y:<<y+?");
WriteLn( "-------------------------------------------------------------");

p+q+x+y

};

Set fun2( Matrix p, Matrix q)
{

Matrix fun3( Matrix x, Matrix y)

{

fun1(x,y)

};

Matrix out = fun3(p,q);

WriteLn( "q:<<q+?"+NL+"p:<<p+?");
WriteLn( "Asignacion Correcta:"+NL+"x:<<x+?"+NL+"y:<<y+?");

p,q,x,y, out?

};

Set salida = fun2(a,b);

[danirus]

We can follow the same problem with this short code:

Real fun1(Real p1) { x };

Real fun2(Real p2)
{

Real fun3(Real x) { fun1(x) };
fun3(p2)

};

Real a = 1;
Real x = 2;
Real salida = fun2(a);

Variable salida must be "2" the value of global variable "x", but actually
returns "1" the value of "p2" (assigned to "x" parameter of fun3 in its stack).

[danirus]

I have just committed some changes in the core of Tol to give a solution for
this bug.

If you are trying to resolve the value of variable "salida", executing while
reading the code above this messages, practically is ovious that the value
returned must be "2" but not "1".

Implementation Info:

BUserFunction has a new attribute called fInsideLevel_ that stores the level_
used inside a BUsedFunction. When Tol evaluates function's code, all variables
needed are searched into the stack_, but only those with a level_ <= to
fInsideLevel_ are accesible to the function.

[danirus]

Now appears another problem related with the Scope of variables with the same
fInsideLevel_. A better solution needs a reimplementation of the common stack_
that implements local scopes.

Code has been returned back to previous state in CVS, until we have a definitive
solution. The intermediate solution is also there, not erased.

Daniel

[danirus]

This bug should be already clossed.
The solution was implemented, but someone of us forgot to close it.

[danirus]

* Bug 181 has been marked as a duplicate of this bug. *

[Víctor de Buen Remiro]

Problem happens again almost a year ago
I cannot find any version of TOL where this problem is fixed.

[Víctor de Buen Remiro]

It has been a confussion because now scope is DYNAMIC and this is not an error