Re: GNAT-Problem '+'-operator with Count

[bobduff@dsd.camb.inmet.com (Bob Duff)]

In article <2ee30bbc.4a534e4f4445@jsnode.equinox.gen.nz>,
Joseph Skinner <joe@jsnode.equinox.gen.nz> wrote:
>In article <3bn74o$go5@tucs6.rz.tu-cottbus.de>,
>Andreas Krohn <ak@informatik.tu-cottbus.de> wrote:
>>   148. procedure solve_matrix(source: in matrix;target: in out matrix) is 
>>   149.     x,y,i,j:Count;    --akt. Pos. in Matrix                        
>>   150.     nachbarn:INTEGER;          --Anzahl der Nachbarn jeder Zelle   
>>   151.     a,b:Count;                                                     
>>   152.                                                                    
>>   153. begin                                                              
>>   154.   for i in feldmin..feldmax loop                                   
>>   155.     for j in feldmin..feldmax loop                                 
>>   156.       nachbarn:=0;          --neue Zelle 0 setzen                  
>>   157.       for x in 1..3 loop   --nachbarn ermitteln                    
>>   158.         for y in 1..3 loop                                         
>>   159.           a:=i+x-2;                                                
>>                      |                                                    
>>        invalid operand types for operator "+"                             
>>z                                                                          
>>z   160.                  b:=j+y-2;                                        
>>                      |                                                    
>>        invalid operand types for operator "+"                             
>>z                                                                          
>>z   161.                  if source(a,b)=TRUE then nachbarn:=nachbarn+1;   
>>   162.           end if;                                                  
>>   163.         end loop;                                                  

The program fragment doesn't show enough to be sure.  In particular,
what type are feldmin and feldmax?  This type will be used as the type
of i and j.  However, x and y are of type Integer, since there's a
special rule saying that if the bounds are universal_integer
(e.g. literals), then it uses Integer.  So if feldmin and feldmax are
not of type Integer, then you've got a type mismatch.  If that's the
problem, try this:

    for x in Whatever_Type_It_Is range 1..3 loop ...

I've gotten into the habit of *always* giving the type name in a for
loop, since I think the default-to-Integer rule confuses things, and
anyway it makes the code more readable.

>At a guess the problem is that the definitions for '+' is not visible to
>the program. Check out your with and use statements to make sure that
>the definition is visible.
>
>On the other hand I may be spouting BS and I'm sure that someone else
>will tell me if I am.

Hard to say.  My answer's a guess, too.  Anyway, in an extremely polite
forum such as this, who would ever accuse anyone of "spouting BS"?
Instead, I respectfully submit that my guess is more likely correct than
yours.  ;-)

- Bob
-- 
Bob Duff                                bobduff@inmet.com
Oak Tree Software, Inc.
Ada 9X Mapping/Revision Team (Intermetrics, Inc.)