drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jacques Nadeau <jacq...@apache.org>
Subject Re: Why does GroupScan have min, max and enforceWidth?
Date Tue, 07 Jul 2015 22:40:15 GMT
Ok, thanks.  I'll try to get a patch up for it.

J

On Tue, Jul 7, 2015 at 3:38 PM, Sudheesh Katkam <skatkam@maprtech.com>
wrote:

> No nuance. My bad, I did not look into the details of how the parallelizer
> arrives at the final value of width using minWidth and maxWidth. So the
> method can be safely removed. minWidth > 1 is a sufficient check.
>
> > On Jul 7, 2015, at 3:26 PM, Jacques Nadeau <jacques@apache.org> wrote:
> >
> > Agreed.  It seems like the same could be accomplished by simply
> determining
> > whether minWidth is greater than 1.
> >
> > On Tue, Jul 7, 2015 at 3:24 PM, Jinfeng Ni <jinfengni99@gmail.com>
> wrote:
> >
> >> Sudheesh may have better idea.  From the code, seems enforceWidth() is
> used
> >> only in ExcessiveExchangeIdentifier.  The purpose is to prevent
> >> ExcessiveExchangeIdentifer from removing Exchange operator, when we want
> >> the query of distributed system table (memory, threads) to run in
> >> distributed mode.
> >>
> >> select * from sys.memory;
> >>
> >>
> >>
> >> On Tue, Jul 7, 2015 at 3:01 PM, Jacques Nadeau <jacques@apache.org>
> wrote:
> >>
> >>> Hey Guys,
> >>>
> >>> In reviewing GroupScan [1], I noted that we have an enforceWidth method
> >> in
> >>> addition to getMin and getMax.  If we set min == max, wouldn't that
> >>> accomplish the same task?  Or am I missing a nuance here?  It seems
> like
> >>> the location referenced in the javadoc could just as well look at
> whether
> >>> getMin() > 1.  Thoughts?
> >>>
> >>> thanks,
> >>> Jacques
> >>>
> >>> [1]
> >>>
> >>>
> >>
> https://github.com/apache/drill/blob/master/exec/java-exec/src/main/java/org/apache/drill/exec/physical/base/GroupScan.java#L61
> >>>
> >>
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message