+1 on the direction. It's great that Spark has standardized the error code so that Iceberg didn't have to rely on error messages.
Yufei On Thu, Dec 19, 2024 at 8:47 AM rdb...@gmail.com <rdb...@gmail.com> wrote: > This looks like a good improvement to me. Thanks, Huaxin! > > On Wed, Dec 18, 2024 at 11:37 PM huaxin gao <huaxin.ga...@gmail.com> > wrote: > >> Hi everyone, >> >> While working on integrating Spark 4.0 with Iceberg, I noticed that error >> conditions in the Spark module are primarily validated through the content >> of error messages. I need to revise some of the validation because the >> error messages have changed in Spark 4.0. Spark has standardized error >> handling by introducing error classes and SQLSTATE codes since 3.1. I would >> like to align the error handling in the Iceberg Spark module with Spark's >> standard error handling framework, specifically by shifting from validating >> error message content to validating error classes and SQLSTATE codes. I >> have prepared a quick write-up >> <https://docs.google.com/document/d/11qHUiCcKMJ-xAyfL__Yv7B1b5N-80-GIwE8AV96A2Ac/edit?usp=sharing> >> for background information and an example. Please let me know what you >> think. If there are no objections to this proposal, I will begin updating >> the error handling in the Spark module. >> >> Thanks, >> Huaxin >> >> >
