TP3/shell/tests/fixtures/t63_record.pas

program t63_record;

{ t63 -- record TYPES: field list, non-compacted layout, and the size.  CALC's
  record is CALC.PAS:126:

      CellRec = record
        CellStatus : set of Attributes;   -- 1:  set of 0..5, one byte
        Contents   : String[70];          -- 71: length byte + 70
        Value      : Real;                -- 6
        DEC, FW    : 0..20;               -- 1+1: a subrange with both bounds
                                                 in 0..255 is ONE byte
      end;                                -- = 80

  Section 3 of CALC-COMPILE.md measured CellRec at 80 from the original's data
  report, so "data=84" here (CellRec 80 + the two @@T temporaries) IS the
  assertion that the layout is right: a difference of one byte in any field
  moves it.  Subrange sizing is load-bearing for the same reason: "0..20" must
  be one byte, per RESUME-TP3.md section 2, or CellRec would be 82 and Sheet
  240 bytes too big.

  Field SELECTION ("c.DEC", "Sheet[FX,FY].CellStatus") is the next increment;
  today it is a loud error, not a wrong address. }

type
   Attributes = (Constant, Formula, Txt, OverWritten, Locked, Calculated);
   CellRec = record
      CellStatus : set of Attributes;
      Contents   : String[70];
      Value      : Real;
      DEC, FW    : 0..20;
   end;
var
   c : CellRec;
begin
   writeln ('a CellRec is 80 bytes')
end.