You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@parquet.apache.org by Julien Le Dem <ju...@twitter.com.INVALID> on 2014/09/29 19:10:58 UTC

Parquet sync-up

Today at 10:30 am PST on Google Hangout.
If you want to introduce yourself, ask a question or just listen in:
https://plus.google.com/events/csp455h8d0fm2sbkhouuu95buuo
Julien

Re: Parquet sync-up

Posted by Julien Le Dem <ju...@twitter.com.INVALID>.

Notes
Attendees and Agenda
MapR: Jason, Parth working on Apache Drill.
- Status of ByteBuffer on read and write paths PRs
- Discuss Idea of vectorized reader (Daniel and Parth)
Netflix: Daniel
- discuss ongoing patches
- PARQUET-100: _metadata
- PARQUET-99: OOM
Pratik listening in
Cloudera: Ryan
- update on parquet-format rc
- talk about standard Map and List mapping
Twitter: Julien, Tianshuo
- talk about release.

ByteBuffer PRs plan:
 - make a release first
 - review and merge

Vectorized APIs
 - Netflix and Mapr (Drill) interested in collaborating

PARQUET-106:
 - enable subclassing Splits

PARQUET-99:
 - problems when some huge records
 - write path could be improved
 - impala uses 256MB row groups.

PARQUET-100:
 - add split strategy to read metadata only on tasks

Release:
 - will get a parquet-format rc this week.

Standard Map and List representation in the builder:
 - enforce that LIST OriginalType annotation only on the standard .
 - extra fields are free if they are required => simplify and unify mapping




On Mon, Sep 29, 2014 at 10:10 AM, Julien Le Dem <ju...@twitter.com> wrote:

> Today at 10:30 am PST on Google Hangout.
> If you want to introduce yourself, ask a question or just listen in:
> https://plus.google.com/events/csp455h8d0fm2sbkhouuu95buuo
> Julien
>