[Subtask] Replace ColumnDTO with ColumnImpl in test cases

[mchades]

Describe the subtask

Replace ColumnDTO with ColumnImpl in test cases

Parent issue

#1292

[zivali]

Hi @mchades,
I see in some test cases, the ColumnDTO object will be used to create TableDTO and TableCreateRequest.
e.g. in
https://github.com/datastrato/gravitino/blob/main/clients/client-java/src/test/java/com/datastrato/gravitino/client/TestRelationalTable.java#L82-L109

For TableDTO and TableCreateRequest, should they continue to use DTO object or switch to use ColumnImpl as well?
I see in ae0054f you stayed with ColumnDTO and added a toDTOs method. So I suppose all other test cases should do the same?

[mchades]

Hi @zivali , below is my supplement to this issue which may answer the question you mentioned above
The criterion for determining whether to use ColumnDTO is based on:

whether the constructed Column object is needed for Java client calls to the Gravitino API.

If it is required for this purpose, then ColumnImpl should be used; otherwise, ColumnDTO can be utilized.

In the examples you provided, Column objects are used to mock the server response containing ColumnDTO, which is acceptable.
https://github.com/datastrato/gravitino/blob/ec2fd57ce77d386e1eed0b7855ef6b16fef691ac/clients/client-java/src/test/java/com/datastrato/gravitino/client/TestRelationalTable.java#L82-L109

But the code immediately following it is faulty:
https://github.com/datastrato/gravitino/blob/ec2fd57ce77d386e1eed0b7855ef6b16fef691ac/clients/client-java/src/test/java/com/datastrato/gravitino/client/TestRelationalTable.java#L113-L123

As the client is invoking the createTable API of Gravitino, ColumnDTO should not be passed as a parameter. We can modify the code as follows to align with our standards:

    partitionedTable =
        catalog
            .asTableCatalog()
            .createTable(
                tableId,
                fromDTOs(columns), // client should use ColumnImpl to create Table, here use fromDTOs just for code reuse
                "comment",
                Collections.emptyMap(),
                partitioning,
                DistributionDTO.NONE,
                SortOrderDTO.EMPTY_SORT);

[zivali]

I see. So the key difference lies in whether the Column object is used to call Gravitino API.
Thank you for the detailed explanation! :) I will continue to finish those test cases, which I was uncertain about.