[Bug c/16409] [4.0 regression] ICE in size_binop, at fold-const.c

["jsm28 at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2004-09-27 22:14 -------
Testing a patch (which gives an error for initializing an incomplete
array whose element type is a VLA type; I don't think there's any
other reasonable interpretation given that VLAs may not be initialized,
and such incomplete arrays ought to count as VLAs).

Although I personally consider a regression from previous mainline to
count as a regression, I didn't think that was policy as regards
release management and setting target milestones: I thought that only
regressions from previous releases counted there.  (The testcases in
this bug never worked in any release.)

Also included in my patch are fixes and tests for some other cases that ICE:

const int i = 1;
void foo() { char *p = (char [i]){ "" }; }


const int i = 1;
void foo() { void *p = (char [][i]){ "" }; }


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2004-07-07 17:58:18         |2004-09-27 22:14:35
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16409