------- Additional Comments From tkoenig at gcc dot gnu dot org 2005-07-15 20:25 ------- (In reply to comment #16) > This is one bad nasty ugly bug.
Yep. This "works" (for small values of "works") if the IMPLICIT NONE is omitted. The problem appears to be that resolve_symbol() doesn't find the generic function, but sets the symbol to a variable of default type using gfc_set_default_type(). Without IMPLICIT NONE, this is harmless and will get corrected later on. With IMPLICIT NONE, gfc_set_default_type() throws an error. I can think of two solutions: a) Find a way for resolve_symbol() to look at the generic functions b) Save type error checking for later, when generics have been resolved. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18108
