[Patch, fortran] PR15809 (PR19276) - ICE and segfault with au toma tic character, dummy pointer arrays

[THOMAS Paul Richard 169137]

Janne,

> > PR19276 (I note that I have not committed the other two 
> fixes to gcc-4.1; I
> > will do so in the next 24hours.)

...well, 48 hours.

> 
> Thanks, it's a bit cleaner than my original version. See a few
> comments in the patch though.

OK - responses below.

> This is OK I guess.
> 
> Actually, I'm not sure this is what we want to do. The problem is that
> there's only 24 bits in the dtype for the length, so it breaks for
> bigger character variables. There was some PR:s about this, and a
> number of intrinsics as well as transfer_array were rewritten to take
> an extra argument for the string length instead of using the
> descriptor. Or perhaps I misunderstood what this part of the patch
> does?

Within the procedure, there is a need for the size information for the array
elements to be available to build the array descriptor dtype. This is needed
in assignments of GFC_DESCRIPTOR_TYPE_Ps and their io, for example,
otherwise segfaults occur.  The latter could be fixed, I guess, but I am not
sure that I want to get into the former just yet.

At some stage array representations will have to be reformed. How about I
hang a TODO patch here? 

> This comment is not correct. If an intrinsic is a derived type
> component, it is scalarized, even with this patch.

You are quite right.  Apologies.
> 
> And yes, the updated testcase you sent by private mail works correctly
> on Linux.

Good.  I need to understand why the original worked on Cygwin. It should not
have.

Subject to these corrections and comments, is the patch OK to go?

Paul