Re: [DISCUSS] Hive Support

2024-12-18 Thread Manu Zhang
Hi Peter, > I think we should make sure that the Iceberg Hive version is independent > from the version used by Spark I'm afraid that is not how it works currently. When Spark is deployed with hive libraries (I suppose this is common), iceberg-spark runtime must be compatible with them. Otherwis

Re: Should we release a new Apache Tez version?

2024-12-18 Thread Ayush Saxena
Would be good to have a new Tez release, We even had a pending Hadoop upgrade PR in Tez which actually blocks Hive's way to upgrade Hadoop, since not having the same version in Hive & Tez in my past experience always crashed. If Hive waits for sometime, we might be able to push a Tez release soon

Re: [DISCUSS] Hive Support

2024-12-18 Thread Péter Váry
@Manu: What will be the end result? Do we have to use the same Hive version in Iceberg as it is defined by Spark? I think we should make sure that the Iceberg Hive version is independent from the version used by Spark On Mon, Dec 16, 2024, 21:58 rdb...@gmail.com wrote: > > I'm not sure there's a

Re: Should we release a new Apache Tez version?

2024-12-18 Thread László Bodor
Hi! Given that TEZ-4577 fixes an issue, it would make sense to let Tez 0.10.5 happen sooner rather than later. Unfortunately, as Tez 0.10.4 introduced some api changes too (TEZ-4548), users cannot easily revert to 0.10.3. I'm wondering how much we can block a hive release because of this: making a

Re: Should we release a new Apache Tez version?

2024-12-18 Thread Sungwoo Park
Hi, TEZ-4577[2] fixes a serious performance bug, which was introduced in the latest release of Tez and can affect production queries. For example, TPC-DS benchmark is affected by this bug. So, I also agree with Butao that the next release of Hive should be packaged with a new version of Tez with T

Re: Re: [DISCUSS] Hive 4.1.x release

2024-12-18 Thread Ayush Saxena
Hi Folks, I strongly believe that the 4.1 release should include JDK-17 support, as it has been a key feature we’ve been planning for a while. JDK-17 is a compelling reason for users to upgrade, whereas other features, though significant, might not be enough to drive widespread adoption. If JDK-17

Re: Re: [DISCUSS] Hive 4.1.x release

2024-12-18 Thread Butao Zhang
Thanks all, JDK17 is a really big feature, and we should carefully evaluate whether this feature should be included in this 4.1.0 release. Hope to have other folks to share their advice. BTW, i want to cut the new branch 4.1.0 next week. If you want to include some tickets