Re: [DISCUSS] Table name in table metadata

2025-02-10 Thread rdb...@gmail.com
I don't think it is a good idea to add the table name to metadata because it can easily get stale and would be misleading. Table name is a catalog concern and we typically try to keep catalog concerns out of the table space. Instead, I'd suggest updating the error that your users see so that the er

Re: [DISCUSS] Table name in table metadata

2025-02-10 Thread Yufei Gu
I see the table identifier (catalog.namespace.table), including the table name, as a catalog concept rather than a table property. You can register a table with the same location but different names, which makes sense from a catalog perspective. This also makes tables more portable when moving or

Re: [DISCUSS] Table name in table metadata

2025-02-10 Thread Gabor Kaszab
Hi Manu, I'm just brainstorming about how this addition could be problematic: Even though it's not recommended, it's feasible to register the same table in multiple catalogs. Different catalogs could use different names for the same underlying table, see register_table(name, metadata_location)

[DISCUSS] Table name in table metadata

2025-02-09 Thread Manu Zhang
Hi all, >From time to time, users ask me about the status of their Iceberg tables by sending me a *path*, which they've received in a file system alert email. Usually I look for the corresponding *table name *and query metadata tables through Spark SQL. However, it's not easy to find the table nam