Clarify or fixe create or replace statement for iceberg connector

[nikita-sheremet-java-developer]

Consider exmaple:

CREATE TABLE iceberg.temp.change_check
  WITH (
    format = 'PARQUET',
    format_version = 2,
    location = 's3://mybucket/change_check'
    )

as SELECT  *
   FROM    (
             VALUES
               ('1', 2),
               ('3', 4)
           ) AS q (uid, site_id);

CREATE OR REPLACE TABLE iceberg.temp.change_check
  WITH (
    format = 'PARQUET',
    format_version = 2,
    location = 's3://mybucket/change_check'
    )

as SELECT  *
   FROM    (
             VALUES
               (1, '2'),
               (3, '4')
           ) AS q (uid, site_id);

Expected reesult: Table created
Actual result: Error

org.apache.iceberg.exceptions.CommitStateUnknownException: The following columns have types incompatible with the existing columns in their respective positions :
uid
Cannot determine whether the commit was successful or not, the underlying data files may or may not be needed. Manual intervention via the Remove Orphan Files Action can remove these files when a connection to the Catalog can be re-established if the commit was actually unsuccessful.
Please check to see whether or not your commit was successful before retrying this commit. Retrying an already successful operation will result in duplicate records or unintentional modifications.
At this time no files will be deleted including possibly unused manifest lists.

In theory this should work - just new table replaces previous one. New commit/snpathos new metadata etc. But here we get another question how to make time travel queries for the table? With same schena I just query old version and this work. But when schema changed it is unclear how to manage it. I do not know much details but Trino uses hive metastore that has only one table version.

So this should be fixed or restrictions about create or replace should be clarified.

[nikita-sheremet-java-developer]

@mdesmet
I also appologies if I'm bothering you, but could you please as author of create or replace feature share your opinion wether this a bug or fundamental Trino restriction?

Thanks in advvance!

[nikita-sheremet-java-developer]

Hm, also there is a question about create or replace with partition column changing. In theory - just add new files in different partition :) For iceberg there is no matter where files are located. But this is just theoretical point of view.