[Bug fortran/51605] internal compiler error gfc_trans_block_construct, at fortran/trans-stmt.c:984

["burnus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51605

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |rejects-valid

--- Comment #6 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-12-18 14:11:00 UTC ---
(In reply to comment #5)
> Therefore I believe it's conforming to point to the associating entity with a
> typed pointer.
> (integer_ptr => symbol_ptr)

I should have read the test case more carefully. The associate name is on the
right and not on the left hand side. I concur that in this case the associate
name should get the target attribute. Thanks for pointing that out!


> I'm using the gfortran I'm using because it had a Mac installer.  I thought
> 4.6.2 was fairly recent.

Yes, 4.6.2 is the latest release; however, the 4.6 branch is now 9 months old
and thus misses all the changes for 4.7, which will be released in March. [The
trunk (= main development line) is currently in the stabilization and bug
fixing Stage 3.]

As Fortran's polymorphism support is quite complicated, implementing it takes a
while and the current implementations are still incomplete and buggy. (That's
the case for all compilers, though the stability and completeness varies.)

In case of GCC/gfortran, 4.7 now supports constructors (DT name = generic
function name) and - since 2011-12-11 -polymorphic arrays. Additionally,
several other bugs were fixed. See http://gcc.gnu.org/wiki/OOP for an overview.

There is a fairly new 4.7 binary available for Darwin (cf.
http://gcc.gnu.org/wiki/GFortranBinaries#MacOS ) but the build is almost 4
weeks old and thus does not yet support polymorphic arrays. My idea was that I
will asked for a new build soon, but only after a few additional OOP bugs have
been fixed.