I can confirm the ICE with 4.7. Trunk gives:

    select type(an => carr%c)
                      1
Error: Component to the right of a part reference with nonzero rank must not have the ALLOCATABLE attribute at (1)

I have not checked 4.6.

AFAICT this has been fixed by revision 187192 (pr41600). I don't think this is a regression: I get the ICE for 4.5.3, 4.6.3, and 4.7.2 (CLASS is not part of 4.4).

I don't know the best way to resolve this PR: WONTFIX or FIXED?

(In reply to comment #2)
> I don't know the best way to resolve this PR: WONTFIX or FIXED?

FIXED I would say (provided the behavior on trunk is ok).

(In reply to comment #2)
> AFAICT this has been fixed by revision 187192 (pr41600). I don't think this is
> a regression: I get the ICE for 4.5.3, 4.6.3, and 4.7.2 (CLASS is not part of
> 4.4).

Interesting, I get no ICE with 4.6.3 (I used the installed compiler on Debian, maybe it is a somehow patched version? Probably a bad idea to use an installed compiler to test this ...):

% uname -a
Linux dellschleppa 3.2.0-4-686-pae #1 SMP Debian 3.2.39-2 i686 GNU/Linux

% cat /etc/debian_version 
7.0

% gfortran-4.6 -v polymorph2.f08
Driving: gfortran-4.6 -v polymorph2.f08 -l gfortran -l m -shared-libgcc
Using built-in specs.
COLLECT_GCC=gfortran-4.6
COLLECT_LTO_WRAPPER=/usr/lib/gcc/i486-linux-gnu/4.6/lto-wrapper
Target: i486-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 4.6.3-14' --with-bugurl=file:///usr/share/doc/gcc-4.6/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-4.6 --enable-shared --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.6 --libdir=/usr/lib --enable-nls --with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-gnu-unique-object --enable-plugin --enable-objc-gc --enable-targets=all --with-arch-32=i586 --with-tune=generic --enable-checking=release --build=i486-linux-gnu --host=i486-linux-gnu --target=i486-linux-gnu
Thread model: posix
gcc version 4.6.3 (Debian 4.6.3-14) 
COLLECT_GCC_OPTIONS='-v' '-shared-libgcc' '-mtune=generic' '-march=i586'
 /usr/lib/gcc/i486-linux-gnu/4.6/f951 polymorph2.f08 -quiet -dumpbase polymorph2.f08 -mtune=generic -march=i586 -auxbase polymorph2 -version -fintrinsic-modules-path /usr/lib/gcc/i486-linux-gnu/4.6/finclude -o /tmp/ccUKC6Je.s
GNU Fortran (Debian 4.6.3-14) version 4.6.3 (i486-linux-gnu)
        compiled by GNU C version 4.6.3, GMP version 5.0.5, MPFR version 3.1.0-p10, MPC version 0.9
GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
GNU Fortran (Debian 4.6.3-14) version 4.6.3 (i486-linux-gnu)
        compiled by GNU C version 4.6.3, GMP version 5.0.5, MPFR version 3.1.0-p10, MPC version 0.9
GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
polymorph2.f08:20.12:

            c = an
            1
Error: Incompatible ranks 0 and 1 in assignment at (1)

> > AFAICT this has been fixed by revision 187192 (pr41600). I don't think this is
> > a regression: I get the ICE for 4.5.3, 4.6.3, and 4.7.2 (CLASS is not part of
> > 4.4).
>
> Interesting, I get no ICE with 4.6.3 (I used the installed compiler on Debian,
> maybe it is a somehow patched version? Probably a bad idea to use an installed
> compiler to test this ...):

Sorry for the noise, the comment #3 was intended for pr56730. Indeed for this test I see the ICE for 4.7 only.

(In reply to comment #5)
> Sorry for the noise, the comment #3 was intended for pr56730. Indeed for this
> test I see the ICE for 4.7 only.

... which means it's a 4.7 regression after all.

If there is an easy fix, it should certainly be applied to the 4.7 branch (otherwise one can probably forget about it, since it's "only" an ICE-on-invalid).

Slightly reduced test case:

program polymorph2

    type ta
        integer i
    end type

    type ca
        class(ta), allocatable :: c
    end type

    type(ca) carr(2)

    select type(an => carr%c)
        type is (ta)
    end select

end program

(In reply to comment #1)
> Trunk gives:
> 
>     select type(an => carr%c)
>                       1
> Error: Component to the right of a part reference with nonzero rank must not
> have the ALLOCATABLE attribute at (1)

That error message was apparently fixed for CLASS components (PR53191) with this commit:

http://gcc.gnu.org/viewcvs/gcc?view=revision&revision=187192


Not sure if it's worth backporting though.

GCC 4.7.3 is being released, adjusting target milestone.

Fixed in 4.8.0?

> Fixed in 4.8.0?

I cannot answer for 4.8.0, but it is fixed for 4.8.3.