On Mon, 2010-01-11 at 10:24 -0600, Roman Kononov wrote:
> #g++ --version | head -1
> g++ (GCC) 4.4.3 20091228 (prerelease)
> #cat test.cc
> typedef _Decimal32 my_type;
> #gcc -c test.cc
> #g++ -c test.cc
> test.cc:1: error: '_Decimal32' does not name a type
>
> G++ is unfamiliar with the _DecimalXX types. Is it a "feature", bug or lack
> of development?
>
> #cat test.cc
> typedef __decltype(0.0DF) my_type;
> my_type foo(my_type v) { return v; }
> #g++ -c test.cc
> test.cc:2: internal compiler error: in write_builtin_type, at cp/mangle.c:1855
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.
>
> The same question.
>
> Thanks :)
>
G++ doesn't recognize _Decimal32 and friends because the C++ standards
committee decided that decimal floating point types should be provided
via classes rather than builtin types. Some of the support for those
classes is in current trunk, but a crucial change to the compiler to
allow binary compatibility between those classes and the C builtin
types wasn't approved before the 4.5 feature cutoff (see
http://gcc.gnu.org/ml/gcc-patches/2009-11/msg01321.html).
Having said that, the GNU C++ compiler knows about decimal floating
point type modes and you can get to the decimal float support via
typedefs:
typedef float dec32 __attribute__((mode(SD)));
typedef float dec64 __attribute__((mode(DD)));
typedef float dec128 __attribute__((mode(TD)));
Instead of dec32/64/128 you could use _Decimal32/64/128, but the C++
TR requires that float.h define those symbols as typedefs to the
classes so you'd run into conflicts later.
There are a few changes needed in the G++ compiler for decimal float
to work properly; see patches added in trunk revisions r152030,
r152242 (this one fixes your ICE), and r152408. In trunk, tests for
builtin decimal float support in G++ are in
gcc/testsuite/c-c++-common/dfp.
Janis
