Re: [PATCH 2/5] d3dcompiler: Fold addition and multiplication of uint constants.

[Matteo Bruni]

On Thu, Jul 2, 2020 at 6:47 PM Zebediah Figura  wrote:
> On 7/2/20 11:15 AM, Matteo Bruni wrote:
> > On Tue, Jun 30, 2020 at 1:56 AM Zebediah Figura  wrote:
> > > +{
> > > - struct hlsl_ir_node *instr, *next;
> > > - BOOL progress = 0;
> > > -
> > > - LIST_FOR_EACH_ENTRY_SAFE(instr, next, instrs, struct hlsl_ir_node, entry)
> > > - {
> > > -
> > >    if (instr->type == HLSL_IR_IF)
> > > -
> > >    {
> > > -
> > >        struct hlsl_ir_if *iff = if_from_node(instr);
> > > -
> > >        progress |= transform_ir(func, iff->then_instrs);
> > > -
> > >        if (iff->else_instrs)
> > > -
> > >            progress |= transform_ir(func, iff->else_instrs);
> > > -
> > >    }
> >
> > I like to have a blank line between declarations and other statements.
>
> Oops, missed that one :-/

No problem :)
> > > -
> > >    else if (instr->type == HLSL_IR_LOOP)
> > > -
> > >        progress |= transform_ir(func, loop_from_node(instr)->body);
> >
> > Unrelated to the patch: maybe we should always initialize else_instrs?
> > I think that would get rid of a bunch of checks with no drawback.
>
> Besides another heap allocation (though maybe we could just store those
> inline instead? I don't think there's any reason why not), but yes,
> probably.

Yeah, I don't see any problem with inlining them either.