This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
Fortran doesn't set TYPE_NONALIASED_COMPONENT. The problem with fortran is that it uses T[1] * and T * willy-nilly to access either T or T[1]. So fortran requires that the element alias set is a subset of the array alias set and that the array alias set is a subset of the element alias set.
See my correction.
Note that disambiguating whole-array accesses and array element accesses via TBAA is rarely useful, instead offset based disambiguation should do the trick here in 99% of the cases.
That may well be, but why treat arrays and records differently?
Because they *ARE* entirely different, because of being able to reference A(J), there is no analog for records!
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |