Notes: Attendees: Carl: Genome rich. interest: SparqSql. (ParquetRDD) Dan, Tongjie: Netflix. interest: PR to review. Frank Austin from berkeley on the Adam project. interest: Predicate pushdown Jacques: Drill. interest: DateTime PR, Mickael: Criteo. Ryan Blue: Cloudera. interest: project that integrates avro and Parquet
Agenda - How many active committers, better way to look at PR. - identify people with expertise per component. - non contentious => should just merge. - action: Julien, email the committers, identify expertise. - build a group of committers more engaged. - Predicate pushdown, backward incompatible changes? - PR #4: added new filter API that applies both to row groups and records - did not remove the old API. Plan to move towards unifying the two. The new API should be a superset of the old. - we should deprecate the old API. - action: Frank to propose an integration to wrap the old API in the new API. (given time) - Semantic versioning: enforced in the pom.xml. PARQUET-31 for discussion. - Renaming packages: PARQUET-23 - publish 1.6 in com.twitter parquet - rename. - publish 1.7 in org.apache parquet - publish 2.0 when new encodings become default. - Ryan: investigated options for renaming. recommends renaming with IntelliJ. volunteers to do the renaming once 1.6 is published. - DateTime PR (Jacques + Ryan) - Jacques: we need to merge it soon. - https://github.com/apache/incubator-parquet-format/pull/3 - Jacques will remove TIMESTAMP_MILLIS which is contentious - Extensions to memory management in reader and writer (Drill) - Jacques to open JIRAs for extensions on the writer path. On Jul 27, 2014, at 5:47 PM, Julien Le Dem <[email protected]> wrote: > This will happen on google hangout Monday 10:30 AM PST > Anybody interested in development of Parquet is welcome to attend. > https://plus.google.com/events/cd5k61u5erb1bqqeg2bv4lpu16c > Notes will be posted on the list afterwards. > sync ups usually happen every 3 weeks. >
