[GitHub] [spark] dongjoon-hyun commented on pull request #26804: [SPARK-26346][BUILD][SQL] Upgrade parquet to 1.11.0

2020-05-19 Thread GitBox
dongjoon-hyun commented on pull request #26804: URL: https://github.com/apache/spark/pull/26804#issuecomment-630997965 @h-vetinari . This is wrong, isn't it? Did someone (except you) say it's low priority here? We want that, but currently it looks infeasible technically. Do you think that

[GitHub] [spark] dongjoon-hyun commented on pull request #26804: [SPARK-26346][BUILD][SQL] Upgrade parquet to 1.11.0

2020-05-19 Thread GitBox
dongjoon-hyun commented on pull request #26804: URL: https://github.com/apache/spark/pull/26804#issuecomment-630904663 @h-vetinari . Parquet is a de-facto standard in Apache Spark and is related to all the other module. That's the reason why Parquet should not break anything in all the

[GitHub] [spark] dongjoon-hyun commented on pull request #26804: [SPARK-26346][BUILD][SQL] Upgrade parquet to 1.11.0

2020-05-18 Thread GitBox
dongjoon-hyun commented on pull request #26804: URL: https://github.com/apache/spark/pull/26804#issuecomment-630471244 @iemejia 's Avro PR (#27609) didn't pass Apache Spark UTs. And, according to his report, this Parquet PR seems to be blocked by Avro dependency upgrade. If we have a