[Patch, F03, RFC] FINAL support in 4.8

[Janus Weil]

Hi all,

there has been quite some activity on FINAL for a while (mainly by
Alessandro and Tobias), which recently stagnated a bit. I think it
would be very important to get this into trunk for the 4.8 release
(the end of stage 1 is coming soon, I guess).

To get the balling rolling again, I'm attaching a new version of the
FINAL patch, based on Tobias' 2012-09-18 version from
https://userpage.physik.fu-berlin.de/~tburnus/final/. At this point I
have only re-diffed it to apply cleanly to the current trunk, and
fixed the scan-tree-dump regressions. Only two testsuite regressions
are remaining now:

FAIL: gfortran.dg/class_array_14.f90  -O0  execution test
FAIL: gfortran.dg/finalize_6.f90  -O  (test for excess errors)

The second one is not dramatic, one only needs to put some more care
into throwing error messages with -std=f95 (I will do that soon). The
first one, however, throws segfaults at runtime (invalid memory
reference). I don't directly see what goes wrong there.

In any case, I would say we should get the patch into trunk ASAP, even
if the implementation of FINAL is not 100% complete (on Tobias'
website, quite a number of to-do's is listed, some of which may still
take a significant amount of effort). AFAICS, the patch already fixes
PR 46321, i.e. polymorphic deallocation (at least in a basic form),
which for me is enough justification to get it in. That is quite some
progress, even if not all corner-cases of finalization will make it
into 4.8.

So, appart from the two left-over testsuite regressions above, are
there any "must-haves" that need to be included in the patch before it
can hit the trunk? Further improvements and bugfixes can follow later,
I would say. Any comments/reviews/contributions are highly welcome ...

Cheers,
Janus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: final-2012-10-24.diff
Type: application/octet-stream
Size: 43371 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20121024/db969a8e/attachment.obj>