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

Switch to Plain View
Drill, mail # dev - the interest list


+
Ted Dunning 2012-09-15, 03:49
+
Azuryy Yu 2012-09-15, 04:11
+
Ryan Rawson 2012-09-15, 04:45
+
Ian Holsman 2012-09-15, 04:54
+
Timothy Chen 2012-09-15, 06:14
+
Ryan Rawson 2012-09-15, 06:21
+
Michael Hausenblas 2012-09-15, 09:43
+
NAVEEN MAANJU 2012-09-15, 11:12
+
Azuryy Yu 2012-09-15, 11:49
+
NAVEEN MAANJU 2012-09-15, 12:28
+
Ted Dunning 2012-09-15, 12:06
+
NAVEEN MAANJU 2012-09-15, 12:30
+
Arun Yadav 2012-09-15, 17:48
+
Ted Dunning 2012-09-15, 21:19
Copy link to this message
-
Re: the interest list
Julian Hyde 2012-09-15, 22:00

On Sep 14, 2012, at 8:49 PM, Ted Dunning <[EMAIL PROTECTED]> wrote:

> Last night the following people signed up as being interested in working on
> different components of Drill:

<snip>

> SQL front-end / optimization
>
> - Nausher Cholavaram - [EMAIL PROTECTED]
>
> - Dan - [EMAIL PROTECTED]

I'll like to add my name to this list. (I got chatting after the meeting, and the sign-up sheets were gone before I noticed.)

 - Julian Hyde - [EMAIL PROTECTED]

I can see a quick way to get a "hello world" query working (JDBC/SQL, talking to a simple table scan operator, e.g. a CSV file reader). I'd use Optiq for JDBC/SQL, add one operator (CsvScanner) and one planner rule (that maps a logical table scan to a call to the CsvScanner). CsvScanner would "implement" itself by generating an LLVM string (can someone provide an example of such a string?) and an execution system could take that string an run it. Metadata is all supplied programmatically (catalog.addTable("EMP")). Next step could be a DrQL parser, also accessed via JDBC.

Julian
+
Ted Dunning 2012-09-16, 03:20