GCC 3.1.1 Status

[Kriang Lerdsuwanakij]

>The middle one is not; we really need to fix the rest of the code to
>be robust in the face of a field with an incomplete type.  Otherwise,
>our error-recovery is going to be horrible.

  OK, I'll work on the suggested fix.

I still have another patch intended for 3.1.1 at
   http://gcc.gnu.org/ml/gcc-patches/2002-06/msg00730.html
waiting for approval.  It fixes bad code generation for
multi-dimensional array constructor.  For the build_vec_init part,
though not a regression, I think it's a nasty bug and it looks safe for 
3.1.1.

But for 3.2, I plan to change build_vec_init by collapsing
multi-D array into 1-D array and working with this 1-D array the entire
function.  The code dealing with the multi-D INIT tree structure could get
messy, unsuitable for 3.1.1.

--Kriang