PATCH: RTL_EXPR vs. inlining-on-trees

Richard Kenner kenner@vlsi1.ultra.nyu.edu
Sat Mar 4 18:12:00 GMT 2000


This removal will have a very large cost for Ada.

Can you say more about why you think this feature is problematical?
It would seem that when you have tree-based inlining you *also* want to
have the scope: shouldn't all the temporaries from one inlined function
be allowed to overlap those from another inlined function?

The Ada issue is when you want to generate code at one point but insert
it into the tree in another you do this with an RTL_EXPR.  Any temporaries
you make during that execution are part of that code and are not
global within the frame.

If there's a bug here, let's fix it, but removing this critical functionality
seems the wrong approach to me.


More information about the Gcc-patches mailing list