spotify/scio

Reservoir sampling into a side output?

オープン

#1,321 opened on 2018/08/23

 (6 件のコメント) (0 件のリアクション) (0 人の担当者)Scala (529 件のフォーク)auto 404
enhancementgood first issue

Repository metrics

Stars
 (2,626 個のスター)
PR merge metrics
 (PR metrics pending)

説明

Right now we have SCollection#sample which uses a Beam transform and is implemented as a global combine operation. It'll be nice to have a sample into side output variation that passes main input through while randomly reservoir sample them into a side input. We should also benchmark this approach against the Beam implementation.

コントリビューターガイド