On Mon, 23 Feb 2026, Avinash Jayakar wrote:

> On Mon, 2026-02-23 at 13:01 +0100, Richard Biener wrote:
> > On Mon, 23 Feb 2026, Avinash Jayakar wrote:
> > 
> > > 
> > > 
> > > On Mon, 2026-02-23 at 11:21 +0100, Richard Biener wrote:
> > > > On Mon, Feb 23, 2026 at 10:53 AM Avinash Jayakar
> > > > <[email protected]> wrote:
> > > > > 
> > > > 
> > > > I wonder whether match.pd matches with (cond^ ...) similar to
> > > > the saturating arithmetic patterns can be used to make the
> > > > matchings
> > > > more maintainable?
> > > > 
> > > Initially I tried to write simplify pattern in match.pd, that would
> > > recognize a >= b ? (a > b) : -1 to a == b? 0 : (a < b) -1: 1. Which
> > > enabled optimize_spaceship method. It did work when source was
> > > written
> > > with ternary operator.
> > > But one problem was that if source is written with if-else instead
> > > of
> > > ternary, it did not work.
> > > if (a >= b) return (a > b) else return -1;
> > 
> > I was suggesting a
> > 
> > (match (spaceship_lt_gt @0 @1 ...)
> >  (...))
> > 
> > and to call that from the optimize_spaceship function.
> Sure, I will try this. 
> For the test cases, is it ok to add a check_effective_target_spaceship
> in target-supports.exp?

Sure.

> > 
> > Richard.
> 

-- 
Richard Biener <[email protected]>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)

Reply via email to