Re: HELP:Declaration in ADA?!!

[kst@thomsoft.com (Keith Thompson)]

Tuyet-Tram DANG-NGOC <dangngoc@ens-info.uvsq.fr> wrote:

>Hi there,

>I have a stupid question to ask.
>DOES SOMEONE KNOW HOW TO DECLARE LONG_INTEGER IN ADA?
>All the books I've read here does not answer this metaphysical question, they
>just take it as if everyone should know how to declare it. In a wonderful
>book, I've seen how to declare LONG_REAL by typing:
>   type LONG_REAL is DIGIT 14;
>But how to do it for LONG_INTEGER?
>Can someone answers me pleeeeaaaase, i've a big and horrible project to finish
>for very soon.

Probably the "big and horrible project" is either finished or overdue
by now, but I'll jump into the fray anyway.

The type Long_Integer, if it exists, is predefined in package Standard.
Some compilers support a type of this name, some don't.  Of those that
do, not all declare it with the same bounds (though -2**31 .. 2**31-1
is most common).

If the compiler you're using doesn't predeclare Long_Integer, you can
declare your own integer type of that name, but it's almost certainly a
bad idea.  Anyone else reading your code will naturally assume that the
name Long_Integer refers to the predefined type, not to a user-defined
type.  Please don't introduce this kind of confusion if you don't have to.

If your compiler does predeclare Long_Integer, DON'T USE IT!  (Sorry about
the shouting.)  Ada provides excellent facilities for declaring your
own integer types of whatever size and/or range you want, subject to
the limitations of the implementation.  Using Long_Integer directly is
inherently non-portable.  (The Ada 95 standard does recommend providing
a type Long_Integer if the implementation supports a type of at least
32 bits, but this is only a recommendation.)

If you want to declare the longest possible integer type, you can do it
like this:

    type Longest_Integer is range System.Min_Int .. System.Max_Int;

If you want a 32-bit type, assuming 2's-complement hardware, you can do
it like this:

    type Integer_32 is range -2**31 .. 2**31-1;

(or use Interfaces.Integer_32 if it's provided).  (There are ways to
declare this without assuming 2's-complement, but that's beyond the
scope of what I feel like thinking about at the moment.)

The only valid reason I can think of for using Long_Integer directly in
user-level code is if you're trying to port a program from a compiler
that supports Long_Integer to one that doesn't, and the program makes
extensive use of Long_Integer.  In this case, it might be necessary
to declare your own type Long_Integer whose range is compatible with
the range of the predefined Long_Integer on the original compiler.
Do this only if you don't have enough time or other resources to remove
the references to Long_Integer, add lots of apologetic comments to the
type declaration, and wash your hands afterward.  8-)}  Then track down
the person who wrote the original code and make them read this article.

(Note that I specifically said user-level code.  I think the GNAT runtime,
for example, uses Long_Integer directly.  In that context, it's not
unreasonable to depend on the characteristics of the compiler.)

-- 
Keith Thompson (The_Other_Keith) kst@thomsoft.com <*>
TeleSoft^H^H^H^H^H^H^H^H Alsys^H^H^H^H^H Thomson Software Products
10251 Vista Sorrento Parkway, Suite 300, San Diego, CA, USA, 92121-2718
This sig uses the word "Exon" in violation of the Communications Decency Act.