Question on tree-walking and mutually-recursive types

[Daniel Berlin]

On Jun 28, 2004, at 11:37 PM, Richard Kenner wrote:

>     I'm confused.
>     Why can't you just use walk_tree_without_duplicates to do your tree
>     walk in the caller, instead of polluting walk_tree?
>
>     Or is this common code that requires a walk_tree instead (it's not
>     clear), and if so, what code exactly?
>
> It's *all* the tree walking that can run into this.  I'm seeing it for
> the walk using check_pointer_types_r, but it could occur on any of 
> them.
>
Okay, well, then rather than pollute walk_tree with what looks like 
completely ada specific code, why not generalize it so that there is a 
tree bit used to tell walk_tree not to recurse further (or to use 
duplicate detection starting with this node), and have the Ada frontend 
set that bit on these trees.
Call it TREE_DONT_WALK or TREE_MAY_CONTAIN_CYCLES or something.
That way, you don't have special cased language code in a middle end 
function.
--Dan