Ryan Blue

This can be done fairly easily by adding key-value properties when reading with Spark. We plan to do this to implement `AS OF SYSTEM TIME` SQL statements as well. You'd...

If it helps, I don't think this needs to be exact. We just want a rough estimate of the cost of a column for CBO.

I'm open to changing this, but I don't think that `spark.catalog.currentDatabase` is necessarily the right choice. That is the current database for the global catalog, but there's no guarantee that...

So it sounds like the problem was that those classes parse the table name instead of passing db and table separately? We'd definitely accept a PR that separated those options...

What do you mean by "schema mangling"?

Makes sense. For identity partitions, where the exact value is stored in the manifest file, we join to those values and then project to get the column order to match...

I think a refactor a while back fixed this. We still need to extend the tests for this in Spark to include ORC.

@omalley, I'm not sure if you saw my comment on ORC-189 but I don't think that's the right timestamp type. Hopefully it is early enough to influence what gets added.

@omervk, it would be great to have docs contributions! We just created a mkdocs site for the Iceberg incubator project, so you can add docs through pull requests to `site`...

Yes, I'll migrate the issues and then turn off issue tracking here.