Re: Distributed DataFrame - ddf.io

Internally, DDF uses Flink Table APIs to process the SQL queries. I would say that DDF would be very useful to provide good virtualization when building application platform. - Henry On Thu, Dec 3, 2015 at 8:48 AM, Kostas Tzoumas wrote: > Hi Nam-Luc, > > I cc Rohit who implemented the DDF frame

[jira] [Created] (FLINK-3109) Join two streams with two different cache time

Wang Yangjun created FLINK-3109: --- Summary: Join two streams with two different cache time Key: FLINK-3109 URL: https://issues.apache.org/jira/browse/FLINK-3109 Project: Flink Issue Type: Improv

Re: Distributed DataFrame - ddf.io

Hi Nam-Luc, I cc Rohit who implemented the DDF framework. I would say that the main difference with the Table API is that DDF aims at portability (running the same code using Flink, Spark, or a database), whereas the Table API is meant to be part of Flink itself. Best, Kostas On Thu, Dec 3, 20

[jira] [Created] (FLINK-3108) JoinOperator's with() calls the wrong TypeExtractor method

Vasia Kalavri created FLINK-3108: Summary: JoinOperator's with() calls the wrong TypeExtractor method Key: FLINK-3108 URL: https://issues.apache.org/jira/browse/FLINK-3108 Project: Flink Issu

[jira] [Created] (FLINK-3107) ZooKeeperCheckpointIDCounter.start() can block JobManager actor

Ufuk Celebi created FLINK-3107: -- Summary: ZooKeeperCheckpointIDCounter.start() can block JobManager actor Key: FLINK-3107 URL: https://issues.apache.org/jira/browse/FLINK-3107 Project: Flink Is

[jira] [Created] (FLINK-3106) ChaosMonkeyITCase failure

Chesnay Schepler created FLINK-3106: --- Summary: ChaosMonkeyITCase failure Key: FLINK-3106 URL: https://issues.apache.org/jira/browse/FLINK-3106 Project: Flink Issue Type: Bug Compo

[jira] [Created] (FLINK-3105) Submission in per job YARN cluster mode reuses properties file of long-lived session

Ufuk Celebi created FLINK-3105: -- Summary: Submission in per job YARN cluster mode reuses properties file of long-lived session Key: FLINK-3105 URL: https://issues.apache.org/jira/browse/FLINK-3105 Projec

Distributed DataFrame - ddf.io

Hello Everyone, We came across the Distributed DataFrame project (http://ddf.io) that aims at implementing a dataframe representation targeting Spark and Flink. Has anybody already heard or played with this project? How would you position that with regards to Flink's Tables? Cheers, -- *Nam-L

RE: The null in Flink

Hi, Stephan Treat UNKOWN as FALSE may works if the Boolean expression is used in filter operation, but for other operations, such as select and groupBy, it does not make sense anymore, we should need UNKNOWN value(or unified as NULL) to distinguish with TRUE/FALSE . Thanks Chengxiang -Ori