GNAT GPL, generics

GNAT GPL, generics

[Dmitry A. Kazakov]

It seems that in GNAT GPL generics are even more buggy than they were in
3.15p. The most interesting new "feature" is that you can have spec. and
body compiled and yet get errors in them upon instantiation, just like in
C++ templates! Another is that declarations in formal packages of a generic
child can somewhat hide (without being "use"-ed!) declarations of the
parent. It is almost impossible to use parameters of parameters:

generic
   with package A (<>);
   with package B (A.Something);
package C is

It will be a total mess in the children of C.

Does anybody have experience with?

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: GNAT GPL, generics

[Dmitry A. Kazakov]

On Sun, 2 Oct 2005 11:15:08 +0200, Dmitry A. Kazakov wrote:

> generic
>    with package A (<>);
>    with package B (A.Something);
> package C is

I mean:

generic
   with package AA is new A (<>);
   with package BB is new (AA.Something);
package C is

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: GNAT GPL, generics

[Dmitry A. Kazakov]

Here is one reduced example:

generic
   type T is digits <>;
package A is
end A;
-----------------------------
with A;
generic
   with package AA is new A (<>);
package B is
end B;
----------------------------
with B;
generic
   with package BB is new B (<>);
package C is
   X : BB.AA.T; -- This is OK
end C;
-------------------------------------------------------
Now add E:
-----------------------------
with A;
generic
   with package AA is new A (<>);
package E is
end E;
--------------------------
with B, E;
generic
   with package BB is new B (<>);
   with package EE is new E (BB.AA);
package D is
   X : BB.AA.T; -- "T" is not a visible entity of "AA"
end D;

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: GNAT GPL, generics

[Alex R. Mosteo]

Dmitry A. Kazakov wrote:
> It seems that in GNAT GPL generics are even more buggy than they were in
> 3.15p. The most interesting new "feature" is that you can have spec. and
> body compiled and yet get errors in them upon instantiation, just like in
> C++ templates! 

This isn't specially new, albeit I don't know if it is more frequent. I 
had an experience like this with 3.15p IIRC. Fortunately it was quite 
rare, it's the only time I've seen it.