3rd ICE on PR20541 - allocatable components of derived types (TR15581))

[Erik Edelmann]

On Sun, Jul 30, 2006 at 06:45:04PM -0400, Jack Howarth wrote:
> Erik,
>   If I understand the proposed patch correctly, it shouldn't impact anything
> else in gfortran other than to implement a previously nonexistent feature.
> If that is true, even as a new feature, it should still be allowed at this
> point since it can't cause regressions elsewhere.

Well, yes, almost all of the code in the patch is hidden by

if (allocatable components) 
  {
    ...
  }

so the probability that it will cause regressions is small.  But
it's still a pretty large patch.


        Erik