On 05/25/2011 03:45 PM, Jason Merrill wrote:
> While looking at this, I also noticed that print_node expects everything to
> have TREE_TYPE, which is no longer correct.
Technically, I think this is not true; everything inherits from tree_common;
or at least tree_typed. (I had the bit of print_node fixing in one of my
patches; thanks for getting it in ahead of me!) Once the STATEMENT_LIST and
BLOCK slimming patches are in, there will indeed be things that don't have
TREE_TYPE. If something doesn't have TREE_TYPE, then the Right Thing to do
would be to fix the tree structure so that it only inherits from tree_base.
Like the following (totally untested) patch. WDYT? (It's certainly possible
that other C++-specific trees could be modified in the same way;
tree_static_assert jumps out, for instance.)
-Nathan
gcc/cp/
* cp-tree.h (struct tree_argument_pack_select): Inherit from
tree_base.
diff --git a/gcc/cp/cp-tree.h b/gcc/cp/cp-tree.h
index ada01fb..a315986 100644
--- a/gcc/cp/cp-tree.h
+++ b/gcc/cp/cp-tree.h
@@ -534,7 +534,7 @@ struct GTY (()) tree_static_assert {
};
struct GTY (()) tree_argument_pack_select {
- struct tree_common common;
+ struct tree_base base;
tree argument_pack;
int index;
};
