Floating point to string conversion

The default formatting of a DECIMAL(P), SMALLFLOAT and FLOAT adapts to the significant digits of the value.

Floating point decimal types (like DECIMAL(5)) can store a large range of values, with a variable number of digits after the decimal point: For example, a DECIMAL(5) can store 12345 as well as 0.12345. See DECIMAL(p,s) for more details about floating point decimal types.

With Genero 2.50, the conversion to string from a DECIMAL(P), FLOAT and SMALLFLOAT has been revised, to keep all significant digits and avoid data loss.

Note: This behavior change is related to the bug fix FGL-3915.

Before Genero 2.50, floating point decimals converted to strings were formatted with 2 decimal digits by default, which could lead to data loss. See following example using a DECIMAL(12):

MAIN
   DEFINE str STRING, dec12, dec12_bis DECIMAL(12)
   LET dec12 = 10.12999
   LET str = dec12
   DISPLAY str
   LET dec12_bis = str
   DISPLAY (dec12 == dec12_bis)
END MAIN

Prior to Genero 2.50, the above code would display:

10.13
     0

Starting with Genero 2.50, all significant digits are kept, which allows proper decimal data serialization:

10.12999
     1

Prior to Genero 2.50, floating point decimal values conversion of huge values could also lose digits in the whole part of the number; the width of the result was never longer than p + 2. Starting with Genero 2.50, all significant digits of a floating point decimal are kept in the result string:

   Values            Vers<2.50       Vers>=2.50
-------------------------------------------------
   1.23456e123       1.23456e123     1.23456e123
   1.23456e40        1.235e40        1.23456e40
   123.456           123.46          123.456
   123456.0          123456.0        123456.0
   0.123456          0.12            0.123456
   0.0123456         0.01            0.0123456
   0.00123456        0.00            0.00123456
   1.23456e-08       0.00            1.23456e-08

If you expect that any DECIMAL(P) to string conversion rounds to 2 digits, define the following FGLPROFILE entry:

fglrun.decToCharScale2 = true

Note: Do not use this configuration parameter unless you have migration issues. This configuration parameter applies only to DECIMAL(P) types, FLOAT and SMALLFLOAT conversions to string is not impacted.