> - Rather than recognizing Fortran specially in read_array_type, I'd
>   rather have a new member of 'struct language_defn' giving the array
>   ordering for that language.
> 
> - I'd like to have the logic that chooses a byte order pulled out of
>   read_array_type into its own function that takes a die, a cu, and
>   returns a DW_ORD_ value.  That function would take care of looking
>   for a DW_AT_ordering attribute, recognizing the appropriate versions
>   of GNU F77, and so on.  read_array_type should call that function
>   and build the type appropriately.

Ok, here's round two!   It passes my Fortran test case ok, but I haven't
run through the normal tests.  I've had to guess that all other
languages are row major (where it actually is relevant..) and I cannot
think of any exceptions other than Fortran.

2004-08-16  David Lecomber  <david@streamline-computing.com>
 
        Fix PR gdb/648
        * language.h (enum array_ordering): New enum.
        * language.h (struct language_defn): New la_array_ordering
attribute.
        * language.c (unknown_language_defn, auto_language_defn)
        (local_language_defn): Ditto.
        * ada-lang.c (ada_language_defn): Ditto.
        * c-lang.c (c_language_defn, cplus_language_defn)
        (asm_language_defn, minimal_language_defn): Ditto.
        * f-lang.c (f_language_defn): Ditto.
        * jv-lang.c (java_language_defn): Ditto.
        * m2-lang.c (f_language_defn): Ditto.
        * objc-lang.c (objc_language_defn): Ditto.
        * p-lang.c (pascal_language_defn): Ditto.
        * scm-lang.c (scm_language_defn): Ditto.
        * eval.c (evaluate_subexp_standard): Assume Fortran arrays are
        oriented large to small in type structure.