Issues for effects with points-to (in any order)

1. Duplicate effects, same semantics, same print-out, different internal representation: a first attempt at merging them led to the merge of A(I) and A(I+1). See call09.c and call10.c

2. Subscript list: extra [0] subscript in array05.c, point_to02.c (very small piece of code). See also Rice/pointer10.c. See also Effects/Old.sub/call19.c.

3. Buggy analyzed codes with NULL or undefined pointer parameters lead to core dumps. See call25.c, call26.c.

4. Points-to information in the caller not translated as points-to information in the callee: call27.c

5. Segfault detected one statement too late: dereferencing05.c. Also
nowhere01.c, null_pointer01.c. This might be normal but the
information about the segfault is only know after segfaulting.

6. Bug for points-to analysis in declaration initialization: statement01_dec.c

7. C intrinsics effects: io_intrinsics.c, statement04.c, statement05.c

8. Effects are unduly merged and imprecise: varargs04.c

10. Effect translation issue for global variables. There
initializations should be taken into account for the "main"
modules. See call03.c.

11. Backward translation fails in array03.c, call05.c, in call06.c,
call07.c, call08.c, call14.c, call20.c, call21.c, call24.c,
inter02.c. Also in Pointers/Mensi.sub/chain.c and tableau_dynamique.c

12. Segfault for call16.c, call19.c; for call16.c, the caller itself
is a callee and callee's and caller's stubs must be distinguished

13. Points-to consistency issue: chap2_exp3.c (and the code is buggy)

14. Points-to analysis: pips internal error; see fulguro01.c

15. Points-to analysis: typing bug in inter02.c

16. Points-to analysis core dumps after a bug is detected:
pointer_modif02.c. I'd like to fix the bug before I fix the source
code of pointer_modif02.c.

17. Interprocedural effects not translated:
Pointers/Mensi.sub/struct_inter03.c

19. call29: A write effect is lost because the callee does not return

20. call30: The free is not taken into account properly by the gen and
kill sets of the call site.


Next bug is 21

----------------------------------------------------------------

Problemes solved:

9. Error falsely detected for call03.c?

-> Tpips script changed.

18. Inexistant effects are not removed as they should. w is NULL and a
branch is never taken in Rice/pointer13.c. However, some memory
effects are still computed although the points-to information is
"bottom".

-> In fact, memory effects ignore continuation issues... Not sure this
   is the correct answer.