You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by George Chow <ge...@overcoil.com> on 2014/02/04 18:56:38 UTC

Suggestions for ZK and Netty interfaces from C++...

Hi,

For the ODBC driver, we need C++ interfaces to Drill.

I've done some searches for Zookeeper and Netty with no conclusive answer.

On today's Hangout, Steven suggested posing this to the list and you,
Jacques.

Any tips/suggestions on how to proceed here?

George

-- 
--
"Not everything that can be counted counts, and not everything that counts
can be counted." Albert Einstein

Re: Suggestions for ZK and Netty interfaces from C++...

Posted by George Chow <ge...@overcoil.com>.
Xiao and I are looking into Boost ASIO to implement the C++ RPC that we
need for the ODBC driver.

George


On Tue, Feb 4, 2014 at 11:26 AM, Jacques Nadeau <ja...@apache.org> wrote:

> For ZK, you can use the zookeeper c client api:
>
> http://zookeeper.apache.org/doc/r3.4.1/zookeeperProgrammers.html#ZooKeeper+C+client+API
>
> Netty is irrelevant to implement a c++ RPC.  Netty is just a framework
> we use to help manage the network interaction.  The Drill protocol
> doesn't do anything special that leverages netty for the wire bytes.
> All the wire format stuff is protobuf + valuevectors.
>
> On Tue, Feb 4, 2014 at 11:17 AM, Timothy Chen <tn...@gmail.com> wrote:
> > I think the C++ driver just needs to communicate to Drill via protocol
> > buffers?
> >
> > With submit plan, responses, etc.
> >
> > Tim
> >
> >
> > On Tue, Feb 4, 2014 at 9:56 AM, George Chow <ge...@overcoil.com> wrote:
> >
> >> Hi,
> >>
> >> For the ODBC driver, we need C++ interfaces to Drill.
> >>
> >> I've done some searches for Zookeeper and Netty with no conclusive
> answer.
> >>
> >> On today's Hangout, Steven suggested posing this to the list and you,
> >> Jacques.
> >>
> >> Any tips/suggestions on how to proceed here?
> >>
> >> George
> >>
> >> --
> >> --
> >> "Not everything that can be counted counts, and not everything that
> counts
> >> can be counted." Albert Einstein
> >>
>



-- 
--
"Not everything that can be counted counts, and not everything that counts
can be counted." Albert Einstein

Re: Suggestions for ZK and Netty interfaces from C++...

Posted by Jacques Nadeau <ja...@apache.org>.
For ZK, you can use the zookeeper c client api:
http://zookeeper.apache.org/doc/r3.4.1/zookeeperProgrammers.html#ZooKeeper+C+client+API

Netty is irrelevant to implement a c++ RPC.  Netty is just a framework
we use to help manage the network interaction.  The Drill protocol
doesn't do anything special that leverages netty for the wire bytes.
All the wire format stuff is protobuf + valuevectors.

On Tue, Feb 4, 2014 at 11:17 AM, Timothy Chen <tn...@gmail.com> wrote:
> I think the C++ driver just needs to communicate to Drill via protocol
> buffers?
>
> With submit plan, responses, etc.
>
> Tim
>
>
> On Tue, Feb 4, 2014 at 9:56 AM, George Chow <ge...@overcoil.com> wrote:
>
>> Hi,
>>
>> For the ODBC driver, we need C++ interfaces to Drill.
>>
>> I've done some searches for Zookeeper and Netty with no conclusive answer.
>>
>> On today's Hangout, Steven suggested posing this to the list and you,
>> Jacques.
>>
>> Any tips/suggestions on how to proceed here?
>>
>> George
>>
>> --
>> --
>> "Not everything that can be counted counts, and not everything that counts
>> can be counted." Albert Einstein
>>

Re: Suggestions for ZK and Netty interfaces from C++...

Posted by Timothy Chen <tn...@gmail.com>.
I think the C++ driver just needs to communicate to Drill via protocol
buffers?

With submit plan, responses, etc.

Tim


On Tue, Feb 4, 2014 at 9:56 AM, George Chow <ge...@overcoil.com> wrote:

> Hi,
>
> For the ODBC driver, we need C++ interfaces to Drill.
>
> I've done some searches for Zookeeper and Netty with no conclusive answer.
>
> On today's Hangout, Steven suggested posing this to the list and you,
> Jacques.
>
> Any tips/suggestions on how to proceed here?
>
> George
>
> --
> --
> "Not everything that can be counted counts, and not everything that counts
> can be counted." Albert Einstein
>