It seems that code like
a = obj%func () ! PROCEDURE, NOPASS :: func => target_func
misses some checks (for instance, that a and the result of func have the same
rank) that are performed for the "equivalent"
a = target_func ()
The attached test program not only misses a diagnostic for this, but ICEs.
I've not yet looked in detail at the ICE, but I suppose it's very likely caused
because it assumes that both ranks are equal or something like this and the
assumption failed because no error was emitted for this.
--
Summary: Checks when assigning from a type-bound procedure broken
Product: gcc
Version: unknown
Status: UNCONFIRMED
Keywords: ice-on-invalid-code
Severity: normal
Priority: P3
Component: fortran
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: domob at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37429
