Here is a chart for easy reference and convenience:
IMPORTANT NOTE: Each compiler can be slightly different in how they have implemented integers and reals.
Delphi:
INTEGER TYPE VALID RANGE
byte : 0 to 255
longint : -2147483648 to 2147483648
shortint : -128 to 127
smallint : -32768 to 32767
word : 0 to 65535
REAL TYPE VALID RANGE
comp : (-2 power of 63) + 1 to (2 power of 63) - 1
currency : -922337203686477.5808 to 922337203685477.5807
double : 5.0 x (10 power of -324) to 1.7 x (10 power of 308)
extended : 3.4 x (10 power of -4932) to 1.1 x 10 (power of 4932)
real : 2.9 x (10 power of -39) to 1.7 x (10 power of 38)
single : 1.5 x (10 power of -45) to 3.4 x (10 power of 38)
FreePascal:
See Real Types Here
See Integer Types Here
A general memory saving practise for every day programming is as follows:
Declare everything as a shortint or smallint since most numbers you will be dealing with are fairly small. When you know or see that you will be dealing with a large number, decimal number, extreme negative number, etc. then pick your best option from the chart.
However, when you restrict yourself to these small numbers without much expansion (i.e. byte) you may spend time refactoring your code too much when you decide to later expand, or when your program changes direction slightly. So the general practise is to just declare an integer or cardinal, which offers high performance for you.
When considering exporting (DLL's, DSO's), remember to check into what is safe to export. i.e. an integer in Pascal might not be the same as an "int" in another language, etc.
|
