test(transaction): Add round-trip tests for Properties & Location

[Kurtiscwright]

Adds three end-to-end tests that exercise Transaction::commit against an in-process MemoryCatalog, covering UpdatePropertiesAction, UpdateLocationAction, and a chained transaction that combines both. Each test asserts post-commit metadata state on both the returned Table and a fresh load_table from the catalog.

Also adds a make_v2_minimal_table_in_catalog helper in transaction::tests, parallel to the existing make_v3_minimal_table_in_catalog.

First PR for #1322 next PR will implement tests for (FastAppend beyond test_row_lineage, UpgradeFormatVersion, ReplaceSortOrder, UpdateStatistics) to follow in a separate PR once this PR is merged.

Which issue does this PR close?

• Closes #.

What changes are included in this PR?

Are these changes tested?

[dannycjones]

LGTM thank you for adding these, this looks like a good test pattern to build on

[github-actions]

This pull request has been marked as stale due to 30 days of inactivity. It will be closed in 1 week if no further activity occurs. If you think that’s incorrect or this pull request requires a review, please simply write any comment. If closed, you can revive the PR at any time and @mention a reviewer or discuss it on the dev@iceberg.apache.org list. Thank you for your contributions.

[andybradshaw]

I think the assertions here are insufficient to say that this happened in a "single commit" vs. two independent commits (i.e. this test but rewritten to commit on each tx action)... seems like not the most important behavior to test, but could add something like the following to assert it more directly:

assert_eq!(reloaded.metadata().metadata_log().len(), 1);

but it also feels like we should be testing the case where chained actions cause a single commit to abort. Something like:

#[tokio::test]
async fn test_failing_chained_actions_does_not_commit() {
    // A failed transaction with multiple actions should apply none of the changes.
	// ... same as other tests ...
    let tx = tx
        .update_schema()
        .delete_column("nonexistent".to_string())
        .apply(tx)
        .unwrap();

    // The commit fails and the catalog was not updated.
    assert!(tx.commit(&catalog).await.is_err());
    let reloaded = catalog.load_table(table.identifier()).await.unwrap();
    assert_eq!(reloaded.metadata().properties().get("owner"), None);
    assert_eq!(reloaded.metadata().location(), original_location);
    assert_eq!(reloaded.metadata().metadata_log().len(), 0);
}

Apologies for the drive-by review, trying to familiarize myself with the repo and contribute in some way.

[dannycjones]

Agree with Andy here, the suggestions here sound useful to add.

@andybradshaw the reviews are very welcome! The more reviewers come from contributors, the lesser the burden on maintainers.

[github-actions]

This pull request has been closed due to lack of activity. This is not a judgement on the merit of the PR in any way. It is just a way of keeping the PR queue manageable. If you think that is incorrect, or the pull request requires review, you can revive the PR at any time.

[dannycjones]

@Kurtiscwright I think this one would be really useful to get merged, especially considering the interest in adding new transactions like RowDelta and RewriteDataAction. Do you think we can get this over the line?

[dhruvarya-db]

Hi @dannycjones, since this looked stale I am attempting to revive it in #3133 . Could you please take a look when you get a chance?