You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by Ben Becker <bb...@maprtech.com> on 2013/10/23 23:35:46 UTC
Extending the expression tree materializer
The expression tree materializer currently works by looking up a SchemaPath
in a RecordBatch to get the corresponding ValueVector.
I'd like to allow for similar functionality, but substituting the
RecordBatch with a RecordBatchLoader. Note that both classes already
implement the required getValueVectorId(SchemaPath) method.
It seems there are several ways to make this work (generics, additional
interface, additional materializer, etc.), but the most appropriate
solution is not obvious to me. What's the best way to go about this?
Thanks,
Ben
Re: Extending the expression tree materializer
Posted by Ben Becker <bb...@maprtech.com>.
VectorAccessible worked great; thanks!
On Wed, Oct 23, 2013 at 2:51 PM, Jacques Nadeau <ja...@apache.org> wrote:
> Steven actually added an interface for this purpose in his latest DRILL-230
> patch called VectorAccessible. If you want to base on top of his branch,
> you should probably be good to go.
>
>
> On Wed, Oct 23, 2013 at 2:35 PM, Ben Becker <bb...@maprtech.com> wrote:
>
> > The expression tree materializer currently works by looking up a
> SchemaPath
> > in a RecordBatch to get the corresponding ValueVector.
> >
> > I'd like to allow for similar functionality, but substituting the
> > RecordBatch with a RecordBatchLoader. Note that both classes already
> > implement the required getValueVectorId(SchemaPath) method.
> >
> > It seems there are several ways to make this work (generics, additional
> > interface, additional materializer, etc.), but the most appropriate
> > solution is not obvious to me. What's the best way to go about this?
> >
> > Thanks,
> > Ben
> >
>
Re: Extending the expression tree materializer
Posted by Jacques Nadeau <ja...@apache.org>.
Steven actually added an interface for this purpose in his latest DRILL-230
patch called VectorAccessible. If you want to base on top of his branch,
you should probably be good to go.
On Wed, Oct 23, 2013 at 2:35 PM, Ben Becker <bb...@maprtech.com> wrote:
> The expression tree materializer currently works by looking up a SchemaPath
> in a RecordBatch to get the corresponding ValueVector.
>
> I'd like to allow for similar functionality, but substituting the
> RecordBatch with a RecordBatchLoader. Note that both classes already
> implement the required getValueVectorId(SchemaPath) method.
>
> It seems there are several ways to make this work (generics, additional
> interface, additional materializer, etc.), but the most appropriate
> solution is not obvious to me. What's the best way to go about this?
>
> Thanks,
> Ben
>