Remove deprecated methods from JdbcTableHandle

[tangjiangling]

Related issues, pull requests, and links

Fixes #6797

Documentation

(x) No documentation is needed.

Release notes

(x) No release notes entries required.

[ebyhr]

Please check CI failures.

[tangjiangling]

Squashed commits and fixed CI.

[hashhar]

How did you decide on when to use asPlainTable vs getRequiredNamedRelation?

[tangjiangling]

How did you decide on when to use asPlainTable vs getRequiredNamedRelation?

This is a really good question, and I'm probably not the best person to answer it (CC: @findepi).

But I can share my understanding:

• If a table is not synthetic, like in JdbcClient#getTableHandle, JdbcClient#renameTable,JdbcClient#addColumn,JdbcClient#dropColumn and other such method calls, it is obvious that "Table" will simply mean which table under which database, and will not contain other information (such as filter push-down, limit push-down, etc.), I would use JdbcTableHandle#asPlainTable (as in the naming description, it just describes which table under which database)
• If a table is synthetic (such as the optimizer's various push-downs: filter push-down, limit push-down, etc., which carry additional information than just a plain table), I would use JdbcTableHandle#getRequiredNamedRelation

To summarize, you need to know what state your table is in at the time of use (synthetic or non-synthetic) before you can choose whether to use getRequiredNamedRelation or asPlainTable.

[hashhar]

LGTM % ci.

[hashhar]

I agree with your understanding of the usages @tangjiangling . However to me this seems like the API around JdbcTableHandle should be revisited at some point. The table handle today is basically a single class which can represent two different types of objects - actual tables as they exist in remote database or something represented by a PreparedQuery - maybe they should each be their own class and that way we can have the type-system guide us what kind of handle is expected where.

We'll discuss with Piotr once he's back and follow-up accordingly.

This PR looks good to me.

[tangjiangling]

(Fixed CI)

[tangjiangling]

CI hit #13362

[tangjiangling]

(Cherry-pick to resolve conflicts)

[tangjiangling]

(Cherry-pick to resolve conflicts again)

[hashhar]

Thanks. I did check that all the deprecations have been there for at-least 2 years now (almost since the Trino rename).