Incompatible types error detection bug - Gpc

2 Feb 2006


      GPC fails to detect incompatible type errors between two or more
new-pointer-type defined types where the same domain-type is used in
new-type type-denoters.  An example demonstrating the bug:
{$extended-pascal}
program grp12(output);
type
  t1 = ^integer;
  t2 = ^integer;
var
  x: t2;
procedure foo(p: t1);
begin
end;
begin
foo(x); {WRONG - incompatible types}
writeln('FAIL');
end.
Although both t1 and t2's new-pointer-type type-denoter have the same
domain-type, t1 and t2 are distinctly different types.  Therefore, t1
and t2 aren't assignment-compatible types (per ISO 10206 paragraph
6.4.6) and the foo(x) call is illegal since the actual-parameter type
(t2) isn't assignment-compatible with the formal-paramter type (t1) as
required by ISO 10206 paragraph 6.7.3.2.
The compiler I used was the latest version of Mac OS X GPC that Adriaan
built and made available for downloading.  The specs:
Reading specs from /Developer/Pascal/gpc345u2/lib/gcc/powerpc-apple-darwin8/3.4.5/specs
Configured with: ../gcc-3.4.5/configure --enable-languages=pascal,c
--enable-threads=posix --target=powerpc-apple-darwin8
--host=powerpc-apple-darwin8 --build=powerpc-apple-darwin8 --prefix=/Developer/Pascal/gpc345u2
Thread model: posix
gpc version 20051116, based on gcc-3.4.5
P.S. This bug was originally found by Mike Hore and discussed on the
MacPascal mailing list.  The above example is a distilled version of the
problem he found while working with  Mac OS X GPC and the Mac OS X GPCPInterfaces.
Gale Paeper
gpaeper@empirenet.com