Tandy 1000 Basic Reference Manual page 52

Chapter 51 Basic
Concepts
When converting
double
to
single precision,
BASIC rounds the
number to 7 significant digits. For example:
A!
=
1.2345678901234567 BASIC stores as 1.234568
A!
=
1.3333333333333333 BASIC stores as 1.333333
When converting
single to double precision, BASIC adds
trailing
zeroes to the right of the original value. If the original value has
an exact binary representation in single precision format, the re-
sulting value is accurate. For example:
A#
=
1.5 BASIC stores as 1.50000000000000
However, if a number does not have an exact binary representa-
tion, the conversion creates a n erroneous value. For example:
A#
=
1.3 BASIC stores as 1.299999952316284
You should not use such conversions in your program because
most fractional numbers do not have exact binary representa-
tions. You can avoid this by forcing the constant to be double pre-
cision, such as:
A#
=
1.3#
or A#
=
1.3D
which BASIC stores as 1.3.
If you must convert from single to double precision, the following
programs show a special technique.
Type and run the following program:
1 0 A !
=
1 . 3
2 0 A #
=
A !
3 0
PRINT
A #
BASIC prints 1.299999952316284
Now type and run this program:
1 0 A !
=
1 . 3
2 0 A #
=
V A L ( S T R S ( A ! ) )
3 0
PRINT
A #
BASIC prints 1.3. Converting the single precision number t o a
string before converting it to a double precision value causes
BASIC to store the value accurately.
Note: BASIC cannot automatically convert numeric
data to string data or vice versa. This results in a
"Type mismatch" error. Use the VAL and STR$ func-
tions to accomplish this kind of conversion.
50