[Patch, fortran] PR55763 - Issues with some simpler CLASS(*) programs

[Paul Richard Thomas <paul dot richard dot thomas at gmail dot com>]

Dear All,

Thanks to Tobias for coming up so quickly with class(*) bugs!

This patch is nearly 'obvious' and is self explanatory.  The
correction to match.c is needed to prevent the segfault that would
otherwise result.  The modifications to resolve.c are required since
select type temporaries pick up the array spec of the selector and are
never dummies.

Bootstrapped and regtested on FC17/x86_64 - OK for trunk?

Paul

2012-12-20  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/55763
    * match.c (select_type_set_tmp): Return is a derived type or
    class typespec has no derived type.
    * resolve.c (resolve_fl_var_and_proc): Exclude select type
    temporaries from 'pointer'.
    (resolve_symbol): Exclude select type temporaries from tests
    for assumed size and assumed rank.

2012-12-20  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/55763
    * gfortran.dg/unlimited_polymorphic_4.f03: New test.

Attachment: submit.diff
Description: Binary data