FWD » Core Development » Base Language

Testcases are checked into old testcases and attached.

Interestingly, the code using the fields as the "model" (initializer_var_like_field.p) properly initializes the field holding the date given "1.1" as the initial value:

   field f5b as date init 1.1
...
/*def var vf5a like tt1.f5a.*/
def var vf5b like tt1.f5b.

and becomes:

   @LegacySignature(type = Type.VARIABLE, name = "vf5b")
   /*def var vf5a like tt1.f5a.*/                             <-- BTW, this comment should be above the LegacySignature
   date vf5b = UndoableFactory.date();

Whereas the variable being the "model" for the field (initializer_var_like_field.p):

def var vf5b as date init 1.1.
..
   /*field f5a like vf5a*/
   field f5b like vf5b

becomes:

   @LegacySignature(type = Type.VARIABLE, name = "vf5b")
   /*def var vf5a as date init 0.*/   
   date vf5b = UndoableFactory.date(date.fromLiteral("1.1"));

The 4GL has "1.1" become the data "1/1/yy" where yy is the current year (23).

In walking through the code, the fromLiteral ends up constructing using date() since the format isn't recognized:

   public static date fromLiteral(String str)
   {
      Matcher matcher = literalPattern.matcher(str);
      if (!matcher.matches())
      {
         return new date(); // format not recognized
...

That ends up just returning a date containing unknown, and then that ends up going through the deepAssign:

Daemon Thread [Conversation [00000005:bogus]] (Suspended)    
    date.deepAssign(date) line: 2567    
    date.<init>(date) line: 668    
    TypeFactory.date(date) line: 575    
    UndoableFactory.date(date) line: 486    
    InitializerLikeTest.<init>() line: 84    
    InitializerLikeTestConstructorAccess.newInstance() line: not available    
...

For the case where the field is the model, we don't go through the same path and the output of the variable is the correct "01/01/23":

Daemon Thread [Conversation [00000006:bogus]] (Suspended)    
    TypeFactory.date() line: 521    
    UndoableFactory.date() line: 422    
    InitializerLikeTestReverse.<init>() line: 84    
    InitializerLikeTestReverseConstructorAccess.newInstance() line: not available    
...

In any case, what we seem to have are fields in the table that are formatted more correctly than in the variables (initializer_field_like_var.p on left initializer_var_like_field.p on right):