feat(cubesql): Support join with type coercion

rust/cubesql/cubesql/Cargo.toml

@@ -10,7 +10,7 @@ homepage = "https://cube.dev"
 
 [dependencies]
 arc-swap = "1"
-datafusion = { git = 'https://github.com/cube-js/arrow-datafusion.git', rev = "400fa0d889a8a38ca69f36d5750dfb572fc6018e", default-features = false, features = ["regex_expressions", "unicode_expressions"] }
+datafusion = { git = 'https://github.com/cube-js/arrow-datafusion.git', rev = "dcf3e4aa26fd112043ef26fa4a78db5dbd443c86", default-features = false, features = ["regex_expressions", "unicode_expressions"] }
 anyhow = "1.0"
 thiserror = "1.0.50"
 cubeclient = { path = "../cubeclient" }

rust/cubesql/cubesql/src/compile/test/mod.rs

@@ -31,6 +31,8 @@ pub mod rewrite_engine;
 #[cfg(test)]
 pub mod test_bi_workarounds;
 #[cfg(test)]
+pub mod test_df_execution;
+#[cfg(test)]
 pub mod test_introspection;
 #[cfg(test)]
 pub mod test_udfs;

rust/cubesql/cubesql/src/compile/test/snapshots/cubesql__compile__test__test_df_execution__join_with_coercion.snap

@@ -0,0 +1,9 @@
+---
+source: cubesql/src/compile/test/test_df_execution.rs
+expression: "execute_query(r#\"\n                WITH\n                    t1 AS (\n                        SELECT 1::int2 AS i1\n                    ),\n                    t2 AS (\n                        SELECT 1::int4 AS i2\n                    )\n                    SELECT\n                        *\n                    FROM\n                        t1 LEFT JOIN t2 ON (t1.i1 = t2.i2)\n                \"#.to_string(),\nDatabaseProtocol::PostgreSQL,).await.unwrap()"
+---
++----+----+
+| i1 | i2 |
++----+----+
+| 1  | 1  |
++----+----+

rust/cubesql/cubesql/src/compile/test/snapshots/cubesql__compile__test__test_df_execution__triple_join_with_coercion.snap

@@ -0,0 +1,9 @@
+---
+source: cubesql/src/compile/test/test_df_execution.rs
+expression: "execute_query(r#\"\n                WITH\n                    t1 AS (\n                        SELECT 1::int2 AS i1\n                    ),\n                    t2 AS (\n                        SELECT 1::int4 AS i2\n                    ),\n                    t3 AS (\n                        SELECT 1::int8 AS i3\n                    )\n                    SELECT\n                        *\n                    FROM\n                        t1\n                            LEFT JOIN t2 ON (t1.i1 = t2.i2)\n                            LEFT JOIN t3 ON (t3.i3 = t2.i2)\n                \"#.to_string(),\nDatabaseProtocol::PostgreSQL,).await.unwrap()"
+---
++----+----+----+
+| i1 | i2 | i3 |
++----+----+----+
+| 1  | 1  | 1  |
++----+----+----+

rust/cubesql/cubesql/src/compile/test/test_df_execution.rs

@@ -0,0 +1,63 @@
+//! Tests that validate that complex but self-contained queries can be executed correctly by DF
+
+use crate::compile::{
+    test::{execute_query, init_testing_logger},
+    DatabaseProtocol,
+};
+
+#[tokio::test]
+async fn test_join_with_coercion() {
+    init_testing_logger();
+
+    insta::assert_snapshot!(execute_query(
+        // language=PostgreSQL
+        r#"
+                WITH
+                    t1 AS (
+                        SELECT 1::int2 AS i1
+                    ),
+                    t2 AS (
+                        SELECT 1::int4 AS i2
+                    )
+                    SELECT
+                        *
+                    FROM
+                        t1 LEFT JOIN t2 ON (t1.i1 = t2.i2)
+                "#
+        .to_string(),
+        DatabaseProtocol::PostgreSQL,
+    )
+    .await
+    .unwrap());
+}
+
+#[tokio::test]
+async fn test_triple_join_with_coercion() {
+    init_testing_logger();
+
+    insta::assert_snapshot!(execute_query(
+        // language=PostgreSQL
+        r#"
+                WITH
+                    t1 AS (
+                        SELECT 1::int2 AS i1
+                    ),
+                    t2 AS (
+                        SELECT 1::int4 AS i2
+                    ),
+                    t3 AS (
+                        SELECT 1::int8 AS i3
+                    )
+                    SELECT
+                        *
+                    FROM
+                        t1
+                            LEFT JOIN t2 ON (t1.i1 = t2.i2)
+                            LEFT JOIN t3 ON (t3.i3 = t2.i2)
+                "#
+        .to_string(),
+        DatabaseProtocol::PostgreSQL,
+    )
+    .await
+    .unwrap());
+}

rust/cubesqlplanner/cubesqlplanner/Cargo.toml

@@ -6,7 +6,7 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-datafusion = { git = 'https://github.com/cube-js/arrow-datafusion.git', rev = "400fa0d889a8a38ca69f36d5750dfb572fc6018e", default-features = false, features = ["regex_expressions", "unicode_expressions"] }
+datafusion = { git = 'https://github.com/cube-js/arrow-datafusion.git', rev = "dcf3e4aa26fd112043ef26fa4a78db5dbd443c86", default-features = false, features = ["regex_expressions", "unicode_expressions"] }
 tokio = { version = "^1.35", features = ["full", "rt", "tracing"] }
 itertools = "0.10.2"
 cubeclient = { path = "../../cubesql/cubeclient" }