[Bug fortran/46152] ALLOCATE with type-spec fails for intrinsic types

["sgk at troutmask dot apl.washington.edu" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46152

--- Comment #14 from Steve Kargl <sgk at troutmask dot apl.washington.edu> 2010-10-26 21:01:57 UTC ---
On Tue, Oct 26, 2010 at 07:43:49PM +0000, burnus at gcc dot gnu.org wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46152
> 
> --- Comment #13 from Tobias Burnus <burnus at gcc dot gnu.org> 2010-10-26 19:43:43 UTC ---
> (In reply to comment #12)
> > New patch attached.  This passed regression testing.
> 
> The patch is OK from my side.
> 
> I assume there is still an error printed but error but the extra error causes
> that succeeding checks fail thus you removed the following line:
> 
> - allocate(tx :: x(5))  ! { dg-error "is not an accessible derived type" }
> 
> Otherwise, "is not an accessible derived type" is a nicer error message then
> "error in CLASS IS specification at" but I assume one cannot have everything -
> and the latter is also sufficiently clear.
> 

The error message "is not an accessible derived type" make no
sense to me.  In the testcase, the statement 'allocate(tx :: x(5))'
appears and there is no other appearance of 'tx'.  There is
no reason (imho) to assume that 'tx' was a derived type over
some simple typographically error (ie., a syntax error).  In the
following fixed-form program,

  program hmm
  doubleprecision, allocatable :: x
  allocate(doubleprecicion :: x)
  end program hmm 

saying 'doubleprecicion' is inaccessible derived type seems odd.

match_type_spec() now simply looks at 'tx' and asks if is a
derived type.  If yes, match_type_spec() returns MATCH_YES.  If no,
then 'tx' is compared against the intrinsic types.  'tx' clearly
is not an intrinsic to match_type_spec() returns MATCH_NO.

Also note, the name of the function is match_type_spec(),
I've removed these chunks of code

-      old_locus = gfc_current_locus;
-      if (gfc_match (" :: ") != MATCH_YES)
-       return MATCH_ERROR;
-      gfc_current_locus = old_locus;
       /* Enfore F03:C401.  */
       if (ts->u.derived->attr.abstract)
        {
@@ -2771,8 +2776,6 @@ match_type_spec (gfc_typespec *ts)
        }
       return MATCH_YES;
     }
-  else if (m == MATCH_ERROR && gfc_match (" :: ") == MATCH_YES)
-    return MATCH_ERROR;

that look for '::'.  The double-colon is not part of
a type-spec.  Now, match_type_spec() can never return a MATCH_ERROR
condition, which may mean the "is not an accessible derived type"
message may never triggered. Yes, I think it may be dead code.  

OTOH, I don't use the OO features of Fortran and I certainly will
defer to Janus if he needs this error message for something other
than allocate().  What clearly needs to go is the gfc_match_symbol()
in match_derived_type_spec().

I'll ping Janus for comment before proceeding.