This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] [tuples] some aux functions
> I see. So in theory we could just add two new predicate codes
> 'TRUE' and 'FALSE', which regardless of LHS or RHS evaluate
> to true or false.
>
> Of course using 0 == 0 and 1 == 0 (or similar) would do also, though
> that raises the question of what is the canonical form. Is there
> a gimple_set_cond_predicate() that accepts a tree (thus a result
> from fold) that might be a simple 1 or 0?
My personal preference would be to not add these predicates. We would
have to check everywhere that they are being used instead of 1 == 1
for example.
Down the road we could declare one form as being canonical (1 == 1 for
example), and add checks to make sure it is the only one used. For now
I think it is best to just accept all "trivial" ones.
> Richard.
>
Cheers,
--
Rafael Avila de Espindola
Google Ireland Ltd.
Gordon House
Barrow Street
Dublin 4
Ireland
Registered in Dublin, Ireland
Registration Number: 368047