The following invalid testcase triggers an ICE on trunk (and GCC 4.1.x): ==================================== struct A;
struct B { A a; }; B b[1] = (B[]) { 0 }; ==================================== bug.cc:5:5: error: field 'a' has incomplete type bug.cc:8:20: error: initializer for 'B' must be brace-enclosed bug.cc:8:20: internal compiler error: tree check: expected class 'type', have 'exceptional' (error_mark) in digest_init_r, at cp/typeck2.c:844 Please submit a full bug report, [etc.] If one actually defines A, the compiler goes into an infinite loop (since GCC 4.1.0): ==================================== struct A {}; struct B { A a; }; B b[1] = (B[]) { 0 }; ==================================== bug.cc:8:20: error: initializer for 'A' must be brace-enclosed bug.cc:8:20: error: initializer for 'A' must be brace-enclosed bug.cc:8:20: error: initializer for 'A' must be brace-enclosed [etc.] -- Summary: [4.3/4.4/4.5 Regression] Trouble with invalid array initialization Product: gcc Version: 4.5.0 Status: UNCONFIRMED Keywords: ice-on-invalid-code, error-recovery, monitored, compile- time-hog Severity: normal Priority: P3 Component: c++ AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: reichelt at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42058