This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Canonical types (1/3)
- From: Andrew Pinski <pinskia at physics dot uc dot edu>
- To: mrs at apple dot com (Mike Stump)
- Cc: iant at google dot com, doug dot gregor at gmail dot com, gcc-patches at gcc dot gnu dot org, stevenb dot gcc at gmail dot com, kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Date: Tue, 28 Nov 2006 14:35:21 -0500 (EST)
- Subject: Re: [PATCH] Canonical types (1/3)
> On Nov 28, 2006, at 10:14 AM, Richard Kenner wrote:
> >> It seems to me that all the frontends should use these fields.
> >> Hopefully we can use it as a lever to eliminate the
> >> types_compatible_p
> >> langhook.
> > They can certainly "use" them but are there any front ends besides
> > C and C++ where they'd *gain* anything from there: where two types are
> > to be considered identical even though they are different tree nodes.
> Java and Ada I suspect as the only ones that might, depends upon the
> nature of the code and depends on wether those front ends build
> duplicates or go out of their way to not do this. If they don't build
> duplicates or if they just don't exercise comptypes much, there'd be
> no benefit.
Actually with Ada (and Fortran), they get no benifit as both don't create
a tree level until parsing and semantic analysis is done. If someone was
going to rewrite the C++ front-end, this is the first thing they should do
and get away from building trees for things like templates which are not used.