Question

Why variables are declared as TStrings and created as TStringList?

eg: the var sl is declared as TStrings but created as TStringList

var
  sl : TStrings;
begin
  sl := TStringList.Create;

  // add string values...
  sl.Add( 'Delphi' );
  sl.Add( '2.01' );

  // get string value using its index
  // sl.Strings( 0 ) will return
  //   'Delphi'
  MessageDlg(
    sl.Strings[ 0 ],
    mtInformation, [mbOk], 0 );

  sl.Free;
end;

Answer 1

To my mind that is rather pointless albeit completely harmless. You could perfectly well declare sl to be TStringList and I would always do it that way. For a reader of the code it makes the list of local variables easier to understand.

In this code sl is always assigned a TStringList instance and so there’s nothing to be gained from declaring sl to have the base class type of TStrings. However, if you had code that assigned a variety of different types of TStrings descendants to the variable, then it would make sense to declare it as TStrings.

The situations when you might declare a variable to be of type TStrings would typically be when the code was not explicitly creating the instance. For example a utility method that received a string list as a parameter would be more useful if it accepted a TStrings since then any descendant could be passed to it. Here’s a simple example:

procedure PrintToStdOut(Strings: TStrings);
var
  Item: string;
begin
  for Item in Strings do
    Writeln(Item);
end;

Clearly this is of much greater utility when the parameter is declared to be TStrings rather than TStringList.

However, the code in the question is not of this nature and I believe that it would be ever so mildly improved if sl was declared to be of type TStringList.