https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68226
janus at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |ice-on-valid-code
Summary|[OOP] ICE on wrong |[4.9/5/6 Regression] [OOP]
|assignment of |ICE on assignment of
|pointer-valued function to |pointer-valued function to
|allocatable |allocatable
--- Comment #3 from janus at gcc dot gnu.org ---
(In reply to Dominique d'Humieres from comment #2)
> Confirmed for 4.9.3 up to trunk (6.0). Version 4.8.5 gives the following
> error
>
> a = f()
> 1
> Error: Can't convert CLASS(t) to TYPE(t) at (1)
This means the ICE is technically a regression, even if this error message is
probably not correct, either.
I would guess the code is actually valid!
