Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Drill >> mail # dev >> Review Request 14027: Adding Limit operator end to end


Copy link to this message
-
Re: Review Request 14027: Adding Limit operator end to end
I tried that as well earlier, but still the same result.

It's creating the input and sorts, but in the end it won't call the
DrillImplementor for Rels at all.

I also tried taking out the replace, but same results too.

It wonder if the best route (cheapest) as able to skip the one copied?

I believe I followed mostly what you suggested that is follow
EnumerableLimitRel impl.

Tim
On Sun, Sep 29, 2013 at 3:01 PM, Julian Hyde <[EMAIL PROTECTED]> wrote:

>
> On Sep 28, 2013, at 1:01 AM, Timothy Chen <[EMAIL PROTECTED]> wrote:
>
> > Here is one of the trials I tried:
> >
> > -    final RelTraitSet traits > sort.getTraitSet().plus(DrillRel.CONVENTION);
> > +    final RelTraitSet traits = sort.getTraitSet();
> >     RelNode input = sort.getChild();
> >     if (!sort.getCollation().getFieldCollations().isEmpty()) {
> >       input = sort.copy(
> > -          sort.getTraitSet().replace(RelCollationImpl.EMPTY),
> > -          input,
> > -          RelCollationImpl.EMPTY,
> > +          sort.getTraitSet(),
> > +          sort,
> > +          sort.getCollation(),
> >           null,
> >           null);
> >     }
>
> The
>
>  +  sort,
>
> line should be
>
>  + input,
>
> otherwise you are creating a sort on top of a sort. Give that a try.
>
> Julian
>
>