Support undelete columns #14426
Description
Feature Request / Improvement
Users might accidentally delete a column, after which the column data cannot be recovered, except by reading from an old snapshot via time travel. However, the column cannot be re-added to the table's current schema since iceberg uses unique IDs to track columns.
I think we could support undelete columns to address this issue, meaning re-adding columns that exist in old schemas but have been dropped from the latest schema, to produce a new table schema.
That said, we need to handle compatibility issues properly. For example, required columns must be converted to optional before being added back, and for columns whose types have been changed, we need to select the latest promoted type to avoid compatibility issues.
cc @rdblue @RussellSpitzer @nastra @Fokko Hi, can you take a look at this? Do you think there are any risks with this? Thanks!
Query engine
None
Willingness to contribute
- I can contribute this improvement/feature independently
- I would be willing to contribute this improvement/feature with guidance from the Iceberg community
- I cannot contribute this improvement/feature at this time