Remove old documentation and other unnecessary files

[submodule "sdg-core/src/test/res/java-slicing-benchmarks"]

	path = sdg-core/src/test/res/java-slicing-benchmarks

	url = https://kaz.dsic.upv.es/git/program-slicing/java-benchmarks.git

# Default ignored files

/shelf/

/workspace.xml

{

    "latex-workshop.view.pdf.viewer": "tab"

}

 No newline at end of file

\begin{thebibliography}{99}

\bibitem{cn}

	Citation needed

\bibitem{horwitz03}

	Matthew Allen, Susan Horwitz.

	\textsl{Slicing Java Programs that Throw and Catch Exceptions.}

	2003.

\bibitem{weiser79}

	Mark D. Weiser.

	\textsl{Program Slices: Formal, Psychological, and Practical Investigations of an Automatic Program Abstraction Method.}

	1979.

\bibitem{weiserPS}

	M. Weiser.

	\textsl{Program Slicing}

	1981.

%\bibitem{citekey}

\end{thebibliography}

 No newline at end of file

# TFM Diary

## 10/5/20

### Implemented

`out` nodes of method calls.

### What is left

#### Distinguish when an out node should exist

An `out` node shouldn't always exist. Primitive types cannot be modified in any way inside a method, while object types can.

An argument to a method call may be any kind of expression. In general, any literal expression, such as `1`, `null`, `""`, `true`, etc. will not have an `out` parameter. So we have to take care of the rest.

An `out` node should exist if there is a possibility to trace the value. And it only exists that possibility when the value comes from a **variable**

So, for the list of expressions that are not literal:

- `ArrayAccess (array[0])`: only if `array` is a variable

- `ArrayCreationExpr`: NO

- `ArrayInitializerExpr`: NO

- `BinaryExpr`: NO, it returns a value

- `CastExpr ((Cast) obj)`: only if `obj` is a variable

- `ClassExpr (obj.class)`: NO

- `ConditionalExpr (1?a:b)`: we'll have to check `a` and `b` expressions

- `FieldAccessExpr (obj.field)`: only if `obj` is a **variable** 

- `InstanceOfExpr`: NO

- `MethodCallExpr (foo.bar(x))`: NO

- `NameExpr`: YES

- `ObjectCreationExpr`: NO

- `SuperExpr`: NO

- `ThisExpr`: NO

- `UnaryExpr (a++)`: we'll have to check the operator (only `++`, `--` return the variable), and if `a` is a variable

## 11/5/20

### Implemented

- Added a CALL node between the node which contains the statement with the `MethodCallExpr` and the `MethodDeclaration`. This node is linked with `in` and `out` nodes. This is because if a statement has more than 1 `MethodCallExpr`, it may have repeated `in` and `out` nodes with the same variables

- Changed `out` node expression from `VariableDeclarationExpr` to `AssignExpr`, where the name of the variable is the argument expression

### WIP

- `OutNodeVariableVisitor`, a visitor that returns the expressions that should be candidate for being `out` nodes. This visitor overrides all `*Expr` methods who may contain a `NameExpr`

    - `ConditionalExpr`: Here, we have `ThenExpr` and `ElseExpr`. As we are performing a static analysis, both should be considered for `out` node in case they represented a variable (This is why `OutNodeVariableVisitor` uses a `List`, and doesn't just return a `String` representing the variable)

### Issues

- `Out` node is generated for `MethodCallExpr` as an argument

- When a `MethodCallExpr` is an argument of another, we have to link its output to the `in` node of the containing one

 No newline at end of file