Re: Preprocessor functionality equivalent ideas needed

[Jeffrey Carter <spam@spam.com>]

Marin David Condic wrote:

> While I agree with the general concept of separate bodies for separate 
> implementation-specific modules, I have occasionally run into situations 
> where it would have been nice if I had a compiler directive of some sort 
> to go one way or another depending on some condition. This is especially 
> true where I might need to keep code portable between two different 
> compilers. Also its a big problem when what needs to be different is in 
> the *specification* of something. (Yes, I know, "One more layer of 
> indirection...." can go fix it - but why should it be so hard?)

There are several ways to deal with needing multiple versions of the 
spec. Sometimes you can use functions instead of constants:

package P is -- Version X
    Num_Processors : constant := 4;
end P;

package P is -- Version Y
    Num_Processors : constant := 256;
end P;

becomes

package P is
    function Num_Processors return Positive;
end P;

Functions are sometimes not acceptable because the value needs to be static.

Sometimes you can have multiple versions with different names, and use 
renaming to select the appropriate one:

package P_X is
    Num_Processors : constant := 4;

    Directory_Separator : constant Character := '\';
end P_X:

package P_Y is

    Num_Processors : constant := 256;

    Directory_Separator : constant Character := '/';
end P_Y;

with P_X;
package P renames P_X;

Now the values in P are static.

And while another layer of information hiding may be harder in the short 
run, it's almost always easier in the long run.

-- 
Jeff Carter
"Sir Lancelot saves Sir Gallahad from almost certain temptation."
Monty Python & the Holy Grail
69