drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <ted.dunn...@gmail.com>
Subject Re: Drill Volunteer Developer.
Date Wed, 24 Oct 2012 13:06:55 GMT
Adding/modifying tests is a good place to work on the query parser.  Having
new eyes look at it and make it more clear where it is not clear would also
be good.

As far as execution, there is a beginning on a spec for an intermediate
language at

https://docs.google.com/document/d/1QTL8warUYS2KjldQrGUse7zp8eA72VKtLOHwfXy6c7I/edit

which is linked from

http://incubator.apache.org/drill/

There is a very simple and insufficient interpreter already checked in for
the physical plan.  This needs to be extended and made more readable.

Julian has suggested that we use Optiq for transforming between physical
plan and execution plan.  I wonder if it could be used to move from query
AST to physical plan as well.  Some analysis on that might be good.

On Tue, Oct 23, 2012 at 3:23 PM, David Alves <davidralves@gmail.com> wrote:

> Hi
>
>         I'd like to start helping too.
>         What is involved in whipping the query parser into shape? adding
> tests?
>         With regard to the execution engines, is there a concrete idea on
> where to start?
>
> Best
> David
>
> On Oct 23, 2012, at 4:18 PM, Ted Dunning <ted.dunning@gmail.com> wrote:
>
> > Absolutely.
> >
> > The current big issue is to whip the query parser into shape and to
> manage
> > to move from the query semantic form into the Drill physical plan.
> >
> > A pending issue that will be front and center is the conversion from
> > physical plan to an execution plan and the concurrent creation of more
> and
> > then better execution engines.
> >
> > What can you do?  What would you like to do?
> >
> > On Tue, Oct 23, 2012 at 12:57 PM, Sathiya <ausathya@gmail.com> wrote:
> >
> >> Folks, I am interested in contributing to Drill. Are you guys looking
> for
> >> Individual Contributors..
> >>
> >> http://www.linkedin.com/in/sathiyas
> >>
>
>

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