You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@drill.apache.org by Madhu Borkar <ba...@gmail.com> on 2014/02/07 20:06:56 UTC

Support for distinct, count

Hi,
I could create a parquet data file and query it finally. Thanks.
1. I could use count(*) to count, but count(_MAP[filed]) doesn't work?
2. Also, what about distinct and any other sql functionality? It would be
nice to know what is implemented.
3. I installed drill with binary. Do you recommend using source code to get
latest version?
4. Does drill support creating tables?
Please, let me know.

-Madhu

Re: Support for distinct, count

Posted by Timothy Chen <tn...@gmail.com>.
1, Not sure about that, can you share your exception?

2, I know Aman is working on Hash aggregates, so Distinct and similiar
functionality will be supported soon.

3, Yes since we're still early in the project our binary release is quite
dated. Latest source code contains all the latest updates and tip of master
should just work as well.

4, Not yet, but had discussions about it. I don't think we settled on a
direction around creating tables.

Tim




On Fri, Feb 7, 2014 at 11:06 AM, Madhu Borkar <ba...@gmail.com> wrote:

> Hi,
> I could create a parquet data file and query it finally. Thanks.
> 1. I could use count(*) to count, but count(_MAP[filed]) doesn't work?
> 2. Also, what about distinct and any other sql functionality? It would be
> nice to know what is implemented.
> 3. I installed drill with binary. Do you recommend using source code to get
> latest version?
> 4. Does drill support creating tables?
> Please, let me know.
>
> -Madhu
>

Re: Support for distinct, count

Posted by Jinfeng Ni <ji...@gmail.com>.
Hi Madhu,

1.  count(_MAP[filed]) should work, if you specify the correct field name.
If the field name does not exist in your parquet file, then it will
complain "NULL type is not supported.".  This is something we need fix.

2. distinct aggregation is currently not supported yet.

For other SQL functions, we are working on implementing a subset of common
SQL functions.

3. It would be better if you download the source code and get the latest
version.

4. AFAIK, drill currently does not support creating tables.





On Fri, Feb 7, 2014 at 11:06 AM, Madhu Borkar <ba...@gmail.com> wrote:

> Hi,
> I could create a parquet data file and query it finally. Thanks.
> 1. I could use count(*) to count, but count(_MAP[filed]) doesn't work?
> 2. Also, what about distinct and any other sql functionality? It would be
> nice to know what is implemented.
> 3. I installed drill with binary. Do you recommend using source code to get
> latest version?
> 4. Does drill support creating tables?
> Please, let me know.
>
> -Madhu
>