[Aldor-l] Aldor-combiler bug

[Ralf Hemmecke]

> %1 >> #include "aldor"
>                                            Comp: 60 msec, Interp: 20 msec
> %2 >> #include "aldorinterp"
>                                            Comp: 30 msec, Interp: 0 msec
> %3 >> import from Integer
>                                            Comp: 10 msec, Interp: 0 msec
> %4 >> Set: Integer := 1
> 1 @ AldorInteger
>                                            Comp: 0 msec, Interp: 100 msec
> %5 >> a: Set Integer
>       ^
> [L4 C1] #1 (Error) `Set' is used in a type, so must be constant, and so cannot be assigned to.
> 
> %6 >> import from Set Integer
>       ^
> [L4 C1] #1 (Error) `Set' is used in a type, so must be constant, and so cannot be assigned to.

Interesting...

Try the following program...
Try for T either MachineInteger, List, String, AldorInteger, Set.

For the first 3 it works.

 >aldor -grun -laldor aaa.as
Set = 1
a   = {1}

for AldorInteger and Set it give the same error as above.

---BEGIN aaa.as
#include "aldor"
#include "aldorio"
macro T == List;
main(): () == {
	import from Integer;
	T: Integer := 1;
	a: Set Integer := [1];
	stdout << "Set = " << T << newline;
	stdout << "a   = " << a   << newline;
}

main();
---END aaa.as

For me that looks like a compiler bug, since if T==>Set then there are 
just two identifiers

Set: Integer
Set: (T:PrimitiveType) -> Join(BoundedFiniteLinearStructureType T,
			DynamicDataStructureType T) with { ... }

visible to the compiler and I cannot understand why it thinks that the 
second "Set" which is a constant function has been redefined or made 
non-constant.

But maybe the rule applies that in a given scope a variable can only 
have one type. Still, we have a variable and a constant in scope. That 
should not be a problem. Or should it? Does somebody know where this is 
explained in the AUG?

BTW, how far is the progress of open sourcing the Aldor compiler?

Thank you
Ralf