[
https://issues.apache.org/jira/browse/CALCITE-6693?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17899260#comment-17899260
]
Julian Hyde commented on CALCITE-6693:
--------------------------------------
The test - as its name indicates - tests conversion from RelNode to SQL. That
is, the RelToSqlConverter class. We allow a SQL string because it is a more
concise way to create a relational algebra tree.
You seem to want to test translation from dialect X to dialect Y for various
pairs of dialects (X, Y). There are a few problems with that. First, there is a
huge number of combinations of X, Y. Second, this would no longer be a unit
test, because it would be checking parser, validator, sql-to-rel, and
rel-to-sql.
Can we figure out what are the goals here, and come up with a design that meets
those goals?
> Add Source SQL Dialect to RelToSqlConverterTest
> -----------------------------------------------
>
> Key: CALCITE-6693
> URL: https://issues.apache.org/jira/browse/CALCITE-6693
> Project: Calcite
> Issue Type: Improvement
> Components: core
> Affects Versions: 1.38.0
> Reporter: yanjing.wang
> Assignee: yanjing.wang
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.39.0
>
>
> Currently, {{RelToSqlConverterTest}} converts the original SQL to RelNode
> using the default validator config and the target dialect's type system. This
> is confusing because it's unclear whether the test is meant to verify the SQL
> conversion between different dialects or within the same dialect. I believe
> it should verify the conversion between source and target dialects.
> Therefore, we should clearly define the source and target dialects and
> provide a way to set them. The code would look like
> {code:java}
> @Test void testNullCollation() {
> final String query = "select * from \"product\" order by \"brand_name\"";
> final String expected = "SELECT *\n"
> + "FROM \"foodmart\".\"product\"\n"
> + "ORDER BY \"brand_name\"";
> final String sparkExpected = "SELECT *\n"
> + "FROM `foodmart`.`product`\n"
> + "ORDER BY `brand_name` NULLS LAST";
> sql(query)
> .sourceDialect(PrestoSqlDialect.DEFAULT)
> .withPresto().ok(expected)
> .withSpark().ok(sparkExpected);
> } {code}
> We need also set correct null collation config using source dialect due to
> the source and target dialect have different null collations.
> For the case that the source dialect equals the target dialect
> {code:java}
> @Test void testCastDecimalBigPrecision() {
> final String query = "select cast(\"product_id\" as decimal(60,2)) "
> + "from \"product\" ";
> final String expectedRedshift = "SELECT CAST(\"product_id\" AS DECIMAL(38,
> 2))\n"
> + "FROM \"foodmart\".\"product\"";
> sql(query)
> .withRedshift()
> .withSourceDialectEqualsTargetDialect()
> .ok(expectedRedshift);
> } {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
