Hi Brendan,

On Thu, Nov 1, 2018 at 1:43 PM Brendan O'Brien <bren...@qri.io> wrote:

> Being pointed at the IPC protocol issue, and breaking it down is the exact
> conversation we're hoping to have.
>
> We're a few Go devs in Brooklyn, all *very* excited about Arrow, and would
> need IPC for our use case. We'd be happy to roll up our sleeves on this one
> in the off-hours.
>
> It seems Sebastien Binet & Stuart Carnie are doing great work on the Go
> implementation. Ideally we'd stay out of their way, and grab a a small Go
> sub-task to use as a working introduction and make sure our code practices
> align. If that task is on the roadmap for IPC support, all the better. In
> terms of dev bandwidth, we're limited now, but should be able to put fourth
> more sizable contributions in December. It'd be great if we're known
> entities by then.
>

I'll also need Arrow-IPC support for one of my other CERN-based
applications.
but this application isn't yet at the top of my TODO list (I'll probably
circle back to it beginning of December.)
so I'd be happy to review PRs related to adding IPC support in the meantime
:)

I haven't looked yet at what adding IPC support entails in details, thus
the rather generic and vague JIRA ticket.
breaking it into smaller items sounds like useful work indeed.

as for small-ish introductory sub-tasks, I can think of e.g. the
implementation of Time{32,64} and Date{32,64} arrays:
- https://issues.apache.org/jira/browse/ARROW-3672
- https://issues.apache.org/jira/browse/ARROW-3673
- https://issues.apache.org/jira/browse/ARROW-3674
- https://issues.apache.org/jira/browse/ARROW-3675

happy to see more gophers showing up :)

-s


> Apache arrow is an incredible project. Thanks to all for your
> contributions.
>
> On Thu, Nov 1, 2018 at 8:10 AM Wes McKinney <wesmck...@gmail.com> wrote:
>
> > On the Go point, it sounds like this means implementing the IPC protocol:
> >
> > https://issues.apache.org/jira/browse/ARROW-3679
> >
> > There's many layers to this, from metadata serialization to record
> > batch reconstruction, so may make sense to create some sub-tasks to
> > make the problem a bit less monolithic.
> >
> > Would also be great to get Go to participate in the integration test
> suite
> > On Wed, Oct 31, 2018 at 12:17 PM Jacques Nadeau <jacq...@apache.org>
> > wrote:
> > >
> > > Recap: Short one today.
> > >
> > > # Attendees
> > > Jacques
> > > Pearu
> > > Brendan
> > > Li
> > >
> > > # Topics
> > > ## Go flatbuffers support:
> > > Brendan and company are interested in contributing this. They want to
> > > discuss approach with existing Go developers. Recommendation was to
> > start a
> > > thread on mailing list and then create follow-up jiras as tasks are
> > > identified.
> > >
> > > ## Better Dictionary Support in Java
> > > Li has some ideas on this but they are complex. Plans to write up a
> > > proposal for the mailing list.
> > >
> > > On Wed, Oct 31, 2018 at 9:02 AM Jacques Nadeau <jacq...@apache.org>
> > wrote:
> > >
> > > > https://meet.google.com/vtm-teks-phx
> > > >
> >
>
>
> --
> --
> Brendan O'Brien
> caretaker, qri.io
> twitter.com/b_fiive
> meeting avail: https://calendly.com/b_five
>

Reply via email to