Boundaries of arrays with deferred shape

[Daniel Kraft <domob1812 at gmail dot com>]

Hi,

I already feel I've got the auto-deallocation thing near finishing with 
finalization, and then only the deallocation of temporary function 
results after they've been used will be left.

But at the moment I've got this problem:  When finalizing a derived-type 
array, I'm building a DO-loop to scalarize finalization of 
finalizable/ALLOCATABLE components of the elements of the array.  That's 
working basically at the moment, but in the loop-builder I'm relying on 
as->type==AS_EXPLICIT because I'm currently accessing the boundaries in 
the as->lower/as->upper arrays.

I need to handle arrays with deferred shape (like allocated ones) here, 
too.  A quick gdb check showed me that in this case lower/upper really 
contains NULLs.  How can I access (during resolution, not trans!) the 
array boundaries of *every* array expression?  Is this possible or do I 
need some magic here, like generating intrinsic-calls or other "ugly" 
things?

BTW, at some place I need to generate something like

IF (ALLOCATED (array)) THEN
  do-some-finalization-stuff
END IF

For building the ALLOCATED-expression, at the moment I'm doing a 
gfc_get_sym_tree ("allocated", ...) call which looks really, really 
wrong to me; but so far that's the only partially working solution I 
came up with (works basically, but I get some testsuite failures that 
look like being caused by that one).  How can I do this correctly?  I 
want a gfc_expr* representing this ALLOCATED-call given array as 
gfc_expr*...

Thank you very much, hopefully I can soon submit a really working 
finalization patch :)

Daniel

--
Done:     Bar-Sam-Val-Wiz, Dwa-Elf-Hum-Orc, Cha-Law, Fem-Mal
Underway: Ran-Gno-Neu-Fem
To go:    Arc-Cav-Hea-Kni-Mon-Pri-Rog-Tou