In Delphi, what's the keyword for the maximal value for a floating point type?

Question

Note: This question has been edited because my issue became something completely different. You'll find the direct answer to the question What's the keyword for the maximal value for a floating point type? in MBo's answer - use System.Math's constants.

Delphi provides multiple floating point value types. There's Single, Double and Extended to start with.

Last week I looked up what the keyword is for the maximum floating point value and found this: http://www.delphibasics.co.uk/RTL.asp?Name=Infinity

var
  float1, float2 : single;
begin
  float1 := Infinity;
  float2 := 23;

  ShowMessage('float1 = '+FloatToStr(float1));
  ShowMessage('float2 = '+FloatToStr(float2));
  ShowMessage('float1 - float2 = '+FloatToStr(float1 - float2));
  ShowMessage('-float1 = '+FloatToStr(-float1));
end;

With supposed output

float1 = INF
float2 = 23
float1 - float2 = INF
-float1 = -INF

I have a form with a component on it. This component has a couple attributes. If I use the property editor to put Infinity as value for the floating point attribute, I get the error that it is an invalid floating point value. Same goes for INF.

Then I typed 9.9E308 into the MaxValue field. It transformed into INF.

Feeling rather pleased about this, I saved my form and went on with my work. I compiled the form later that day, succesfully testing that the form could hold a value of 1E308.

The next day...

Opening the .dfm file in notepad gives me this:

object gridDoubleEditor: TdgmrGridDoubleEdit
  Tag = 0
  WantKeyLeftRight = False
  WantKeyUpDown = False
  WantKeyHomeEnd = False
  WantKeyPriorNext = False
  EmptyString = '--'
  EmptyValue = -1E300
  MinValue = -1E300
  MaxValue = INF
  NumberOfDecimals = 2
  Alignment = taRightJustify
  Left = 336
  Top = 208
end

How come it automatically changes 9.9E308 to INF, compiles succesfully, saves succesfully, but fails to load? Is this a bug in my IDE? I'm using Embarcadero® Delphi® XE5 Version 19.0.14356.6604, 32 bits on Windows 7.

Answer 1

System.Math unit contains Infinity constant to represent IEEE754 Inf pseudovalue(1.0/0.0). There are also a lot of MaxDouble, MaxExtended etc constants in this unit (look at Variables section here). This short test works as needed, so I suspect your component may have some kind of range checking or have property editor that don't understand unknown identificators.

  TTest = class(TComponent)
  private
    FValue: Double;
    procedure SetValue(const Value: Double);
  public
    property Value: Double read FValue write SetValue;
  end;


procedure TForm1.Button26Click(Sender: TObject);
begin
  with TTest.Create(Self) do try
    Value := Infinity;
    Caption := FloatToStr(Value);
  finally
    Free;
  end;
end;

P.S. Do you need to use Infinity as field value, or just to ensure that field value > -1E300?

P.P.S. Why you are still using INF, not Infinity? BTW, I've found one component with float published property - Object Inspector don't understand Infinity - so property editor is not designed for these purposes, as I've already supposed, but I can assign Infinity in code.