deps(core): upgrade DataFusion to 55.1

parquet 58 pins thrift ^0.17, which carries CVE-2026-43868 (fixed in
thrift 0.23). parquet 59 drops the thrift dependency, and DataFusion 55 is
the first release on arrow/parquet 59, so the fix requires this upgrade.

API migrations: TableReference moved out of datafusion::sql, as_any() is no
longer part of the UDF/TableProvider traits, Cast/TryCast carry a field
instead of a data type, and ContextProvider/Expr gained higher-order
function support.

Behaviour kept as before:
- IN lists keep the string-preferring coercion. DataFusion now coerces them
  numerically, which would emit `CAST(str_col AS BIGINT) IN (1, 2)` and fail
  in the target database on non-numeric values.
- UNION and CASE WHEN use the upstream split of comparison_coercion into
  type_union_coercion / get_coerce_type_for_case_when, matching their prior
  semantics.
- wren-core-py lists only scalar/aggregate/window functions, since
  information_schema.routines now also reports table functions.

The wasm release profile now uses fat LTO with one codegen unit; without it
the binary grows past the 15 MB gzip budget checked in CI.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

.claude/CLAUDE.md

@@ -4,7 +4,7 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 
 ## Project Overview
 
-Wren is an open-source semantic engine for MCP clients and AI agents. It translates SQL queries through a semantic layer (MDL — Modeling Definition Language) and executes them against 22+ data sources (PostgreSQL, BigQuery, Snowflake, Spark, etc.). The Rust engine is powered by Apache DataFusion (upstream, crates.io v53).
+Wren is an open-source semantic engine for MCP clients and AI agents. It translates SQL queries through a semantic layer (MDL — Modeling Definition Language) and executes them against 22+ data sources (PostgreSQL, BigQuery, Snowflake, Spark, etc.). The Rust engine is powered by Apache DataFusion (upstream, crates.io v55).
 
 The previous WrenAI services (`wren-ui/`, `wren-ai-service/`, `wren-launcher/`, `docker/`, `deployment/`) were moved to the `legacy/v1` branch (tag `v1-final`) as of the wren-engine import. Active development is focused on the Open Context Engine.
 
@@ -83,7 +83,7 @@ Uses `uv` (not Poetry). `pyproject.toml` uses `hatchling` as build backend. Opti
 SQL query
   → wren CLI / wren-core-py
   → wren-core (Rust): MDL analysis → logical plan → optimization
-  → DataFusion (query planning, upstream crates.io v53)
+  → DataFusion (query planning, upstream crates.io v55)
   → connector-specific SQL (Ibis / sqlglot)
   → native execution on the target data source
 ```
@@ -110,7 +110,7 @@ SQL query
 - **Commits**: Conventional commits (`feat:`, `fix:`, `chore:`, `refactor:`, `test:`, `docs:`, `perf:`, `deps:`). Releases are automated via release-please with independent release lines per module.
 - **Rust**: format with `cargo fmt`, lint with `clippy -D warnings`, TOML with `taplo`.
 - **Python**: format and lint with `ruff` (line-length 88, target Python 3.11). Both `core/wren-core-py` and `core/wren` use uv.
-- **DataFusion**: upstream `datafusion` v53 from crates.io (no longer the Canner fork).
+- **DataFusion**: upstream `datafusion` v55 from crates.io (no longer the Canner fork).
 - **Snapshot testing**: wren-core uses `insta` for Rust snapshot tests.
 - **CI**: Per-module path-filtered workflows trigger only on changes inside that module.
 

core/wren-core-py/Cargo.toml

@@ -15,7 +15,7 @@ crate-type = ["cdylib"]
 pyo3 = { version = "0.29.0", features = ["extension-module", "abi3-py311"] }
 wren-core = { version = "0.4.0", path = "../wren-core/core", package = "wren-semantic-core" }
 wren-core-base = { path = "../wren-core-base", features = ["python-binding"] }
-datafusion-common = "53.0.0"
+datafusion-common = "55.1.0"
 base64 = "0.22.1"
 futures = "0.3.31"
 serde_json = "1.0.117"

core/wren-core-py/src/context.rs

@@ -635,6 +635,8 @@ impl PySessionContext {
                 r.description
             FROM
                 information_schema.routines r
+            WHERE
+                r.function_type IN ('SCALAR', 'AGGREGATE', 'WINDOW')
         "#;
         let batches = ctx
             .sql(sql)
@@ -742,6 +744,7 @@ impl PySessionContext {
                 inputs i ON r.specific_name = i.specific_name
             WHERE
                 r.routine_name = '{}'
+                AND r.function_type IN ('SCALAR', 'AGGREGATE', 'WINDOW')
         "#,
             function_name
         );

core/wren-core-py/tests/test_modeling_core.py

@@ -144,7 +144,7 @@ def test_read_function_list():
     path = "tests/functions.csv"
     session_context = SessionContext(manifest_str, path)
     functions = session_context.get_available_functions()
-    assert len(functions) == 290
+    assert len(functions) == 291
 
     rewritten_sql = session_context.transform_sql(
         "SELECT add_two(c_custkey, c_custkey) FROM my_catalog.my_schema.customer"
@@ -156,7 +156,7 @@ def test_read_function_list():
 
     session_context = SessionContext(manifest_str, None)
     functions = session_context.get_available_functions()
-    assert len(functions) == 283
+    assert len(functions) == 284
 
 
 def test_get_available_functions():

core/wren-core-wasm/.claude/CLAUDE.md

@@ -4,7 +4,7 @@ Wren Engine compiled to WebAssembly for browser-native analytics. Runs SQL queri
 
 ## Why a Separate Crate?
 
-Uses **upstream DataFusion** (crates.io v53), not the Canner fork. The WASM version executes queries directly via DataFusion — no SQL unparser or dialect transpilation needed. Kept outside the `wren-core/` workspace to avoid dependency conflicts.
+Uses **upstream DataFusion** (crates.io v55), not the Canner fork. The WASM version executes queries directly via DataFusion — no SQL unparser or dialect transpilation needed. Kept outside the `wren-core/` workspace to avoid dependency conflicts.
 
 Shared code: `wren-core-base` (manifest types, no DataFusion dependency) and `wren-core` semantic layer (MDL analysis rules).
 
@@ -51,10 +51,10 @@ just clean           # Remove pkg/, dist/, target/
 
 ## Dependencies
 
-- **DataFusion v53** (upstream, crates.io) — query engine, `default-features = false` + selected features
-- **Arrow v58.1** — `json` feature for JSON reader
-- **Parquet v58.1** — `snap` + `lz4` only (no zstd — requires C library, can't compile to WASM)
-- **object_store v0.13.1** — `aws` + `http` features for URL mode (HttpStore)
+- **DataFusion v55** (upstream, crates.io) — query engine, `default-features = false` + selected features
+- **Arrow v59.2** — `json` feature for JSON reader
+- **Parquet v59.2** — `snap` + `lz4` only (no zstd — requires C library, can't compile to WASM)
+- **object_store v0.13.2** — `aws` + `http` features for URL mode (HttpStore)
 - **wren-core** (path: `../wren-core/core`) — semantic layer, `default-features = false`
 - **wren-core-base** (path: `../wren-core-base`) — shared manifest types
 - **wasm-bindgen / js-sys / web-sys** — WASM ↔ JS bindings
@@ -69,7 +69,7 @@ just clean           # Remove pkg/, dist/, target/
 - **No SystemTime**: chrono `wasmbind` feature required
 - **getrandom**: All three major versions (0.2, 0.3, 0.4) in the dep tree need explicit WASM JS backend
 - **macOS build**: Needs LLVM (`brew install llvm`) for C deps — justfile handles env vars automatically
-- **Binary size**: ~68 MB raw / ~14 MB gzip (target: < 15 MB gzip)
+- **Binary size**: ~57 MB raw / ~14 MB gzip (target: < 15 MB gzip); the release profile uses fat LTO + `codegen-units = 1` to stay under budget
 
 ## npm Package (wren-core-wasm)
 

core/wren-core-wasm/Cargo.toml

@@ -19,7 +19,7 @@ path = "src/lib.rs"
 # DataFusion WITHOUT parquet feature — parquet's default-features=true in
 # DataFusion enables zstd (C library) which cannot compile to WASM.
 # We read Parquet bytes → Arrow RecordBatch → MemTable ourselves.
-datafusion = { version = "53", default-features = false, features = [
+datafusion = { version = "55.1", default-features = false, features = [
   "nested_expressions",
   "crypto_expressions",
   "datetime_expressions",
@@ -30,16 +30,16 @@ datafusion = { version = "53", default-features = false, features = [
   "sql",
 ] }
 
-object_store = { version = "0.13.1", features = ["aws", "http"] }
+object_store = { version = "0.13.2", features = ["aws", "http"] }
 
 # --- Arrow / Parquet ---
-arrow = { version = "58.1", default-features = false, features = [
+arrow = { version = "59.2", default-features = false, features = [
   "json",
   "csv",
 ] }
 # No zstd: requires C library (zstd-sys) which cannot compile to WASM.
 # Snappy and LZ4 are pure Rust and WASM-compatible.
-parquet = { version = "58.1", default-features = false, features = [
+parquet = { version = "59.2", default-features = false, features = [
   "arrow",
   "snap",
   "lz4",
@@ -89,3 +89,9 @@ getrandom_04 = { package = "getrandom", version = "0.4", features = [
 
 [dev-dependencies]
 wasm-bindgen-test = "0.3"
+
+# Fat LTO with a single codegen unit keeps the binary under the 15 MB gzip
+# budget checked in CI (~14 MB vs ~16 MB without it on DataFusion 55).
+[profile.release]
+lto = true
+codegen-units = 1

core/wren-core/.claude/CLAUDE.md

@@ -1,6 +1,6 @@
 # wren-core
 
-Rust semantic engine for Wren Engine. Handles MDL analysis, query planning, logical plan optimization, and SQL generation via Apache DataFusion (upstream, crates.io v53). Published to crates.io as `wren-semantic-core` (lib name `wren_core`).
+Rust semantic engine for Wren Engine. Handles MDL analysis, query planning, logical plan optimization, and SQL generation via Apache DataFusion (upstream, crates.io v55). Published to crates.io as `wren-semantic-core` (lib name `wren_core`).
 
 ## Workspace Layout
 
@@ -38,7 +38,7 @@ taplo fmt                                               # Format Cargo.toml file
 
 ## Dependencies
 
-- **DataFusion**: upstream `datafusion` v53 from crates.io (no longer the Canner fork)
+- **DataFusion**: upstream `datafusion` v55 from crates.io (no longer the Canner fork)
 - **wren-core-base**: Shared manifest types from `../wren-core-base`
 
 ## Known Limitations

core/wren-core/Cargo.toml

@@ -9,12 +9,12 @@ homepage = "https://getwren.ai"
 license = "Apache-2.0"
 readme = "README.md"
 repository = "https://github.com/Canner/WrenAI"
-rust-version = "1.78"
+rust-version = "1.94"
 version = "0.4.0"
 
 [workspace.dependencies]
 async-trait = "0.1.89"
-datafusion = { version = "53", default-features = false, features = [
+datafusion = { version = "55.1", default-features = false, features = [
     "nested_expressions",
     "crypto_expressions",
     "datetime_expressions",
@@ -24,7 +24,7 @@ datafusion = { version = "53", default-features = false, features = [
     "parquet",
     "sql",
 ] }
-datafusion-session = { version = "53", default-features = false }
+datafusion-session = { version = "55.1", default-features = false }
 env_logger = "0.11.10"
 hashbrown = "0.17.0"
 insta = { version = "1.41.1" }

core/wren-core/core/src/logical_plan/analyze/access_control.rs

@@ -8,13 +8,13 @@ use datafusion::{
     arrow::datatypes::{DataType, SchemaRef},
     common::{
         config::ConfigOptions, file_options::file_type::FileType, plan_datafusion_err,
-        plan_err, Result, Spans,
+        plan_err, Result, Spans, TableReference,
     },
     error::DataFusionError,
     logical_expr::{
         builder::LogicalTableSource,
         planner::{ContextProvider, ExprPlanner, TypePlanner},
-        AggregateUDF, ScalarUDF, TableSource, WindowUDF,
+        AggregateUDF, HigherOrderUDF, ScalarUDF, TableSource, WindowUDF,
     },
     prelude::Expr,
     sql::{
@@ -27,7 +27,6 @@ use datafusion::{
             },
             dialect::GenericDialect,
         },
-        TableReference,
     },
 };
 use wren_core_base::mdl::RowLevelAccessControl;
@@ -395,6 +394,14 @@ impl ContextProvider for RlacContextProvider {
             .cloned()
     }
 
+    fn get_higher_order_meta(&self, name: &str) -> Option<Arc<HigherOrderUDF>> {
+        self.session_state
+            .read()
+            .higher_order_functions()
+            .get(name)
+            .cloned()
+    }
+
     fn get_aggregate_meta(&self, name: &str) -> Option<Arc<AggregateUDF>> {
         self.session_state
             .read()
@@ -428,6 +435,15 @@ impl ContextProvider for RlacContextProvider {
             .collect()
     }
 
+    fn higher_order_function_names(&self) -> Vec<String> {
+        self.session_state
+            .read()
+            .higher_order_functions()
+            .keys()
+            .cloned()
+            .collect()
+    }
+
     fn udaf_names(&self) -> Vec<String> {
         self.session_state
             .read()

core/wren-core/core/src/logical_plan/analyze/model_anlayze.rs

@@ -7,6 +7,7 @@ use crate::mdl::{AnalyzedWrenMDL, Dataset, SessionStateRef};
 use datafusion::common::tree_node::{
     Transformed, TransformedResult, TreeNode, TreeNodeRecursion, TreeNodeRewriter,
 };
+use datafusion::common::TableReference;
 use datafusion::common::{internal_err, plan_err, Column, DFSchemaRef, Result, Spans};
 use datafusion::config::ConfigOptions;
 use datafusion::error::DataFusionError;
@@ -17,7 +18,6 @@ use datafusion::logical_expr::{
     Window,
 };
 use datafusion::optimizer::AnalyzerRule;
-use datafusion::sql::TableReference;
 use std::cell::RefCell;
 use std::collections::HashSet;
 use std::fmt::Debug;

core/wren-core/core/src/logical_plan/analyze/model_generation.rs

@@ -17,12 +17,12 @@ use crate::DataFusionError;
 use datafusion::common::alias::AliasGenerator;
 use datafusion::common::config::ConfigOptions;
 use datafusion::common::tree_node::{Transformed, TransformedResult};
+use datafusion::common::TableReference;
 use datafusion::common::{plan_err, Result};
 use datafusion::logical_expr::{col, ident, Extension, UserDefinedLogicalNodeCore};
 use datafusion::logical_expr::{Expr, LogicalPlan, LogicalPlanBuilder};
 use datafusion::optimizer::analyzer::AnalyzerRule;
 use datafusion::physical_plan::internal_err;
-use datafusion::sql::TableReference;
 
 pub const SOURCE_ALIAS: &str = "__source";
 

core/wren-core/core/src/logical_plan/analyze/scope.rs

@@ -1,8 +1,8 @@
 use crate::mdl::Dataset;
 use datafusion::common::internal_err;
+use datafusion::common::TableReference;
 use datafusion::error::Result;
 use datafusion::prelude::Expr;
-use datafusion::sql::TableReference;
 use std::collections::{HashMap, HashSet, VecDeque};
 use std::fmt::{Debug, Display};
 

core/wren-core/core/src/logical_plan/optimize/simplify_timestamp.rs

@@ -79,7 +79,7 @@ impl TimestampSimplify {
         } else {
             Arc::new(DFSchema::empty())
         };
-        let info = SimplifyContext::default().with_schema(schema);
+        let info = SimplifyContext::builder().with_schema(schema).build();
 
         // Inputs have already been rewritten (due to bottom-up traversal handled by Optimizer)
         // Just need to rewrite our own expressions
@@ -123,13 +123,13 @@ impl TreeNodeRewriter for ExprRewriter<'_> {
         match &expr {
             // we only simplify the cast expression for the literal value
             Expr::Cast(Cast {
-                data_type,
+                field,
                 expr: sub_expr,
             })
             | Expr::TryCast(TryCast {
-                data_type,
+                field,
                 expr: sub_expr,
-            }) if is_timestamp(data_type)
+            }) if is_timestamp(field.data_type())
                 && matches!(sub_expr.as_ref(), Expr::Literal(_, _)) =>
             {
                 let original_name = self.name_preserver.save(&expr);

core/wren-core/core/src/logical_plan/optimize/type_coercion.rs

@@ -25,32 +25,35 @@ use datafusion::{
     common::{
         exec_err, internal_err, not_impl_err, plan_datafusion_err, plan_err,
         tree_node::{Transformed, TreeNode, TreeNodeRewriter},
-        Column, DFSchema, DFSchemaRef,
+        Column, DFSchema, DFSchemaRef, TableReference,
     },
     config::ConfigOptions,
     error::{DataFusionError, Result},
     logical_expr::{
-        binary::{comparison_coercion, like_coercion, BinaryTypeCoercer},
+        binary::{
+            comparison_coercion, like_coercion, type_union_coercion, BinaryTypeCoercer,
+        },
         expr::{
-            self, AggregateFunction, AggregateFunctionParams, Alias, Exists, InList,
-            InSubquery, ScalarFunction, SetComparison, WindowFunction,
-            WindowFunctionParams,
+            self, AggregateFunction, AggregateFunctionParams, Alias, Exists,
+            HigherOrderFunction, InList, InSubquery, ScalarFunction, SetComparison,
+            WindowFunction, WindowFunctionParams,
         },
         expr_rewriter::coerce_plan_expr_for_schema,
         expr_schema::cast_subquery,
         type_coercion::{
-            functions::fields_with_udf,
+            functions::{
+                fields_with_udf, value_fields_with_higher_order_udf_and_lambdas,
+            },
             is_datetime,
-            other::{get_coerce_type_for_case_expression, get_coerce_type_for_list},
+            other::{get_coerce_type_for_case_expression, get_coerce_type_for_case_when},
         },
         utils::merge_schema,
         Between, BinaryExpr, Case, ExprSchemable, Join, Like, Limit, LogicalPlan,
-        Operator, Projection, Subquery, Union, WindowFunctionDefinition,
+        Operator, Projection, Subquery, Union, ValueOrLambda, WindowFunctionDefinition,
     },
     optimizer::{utils::NamePreserver, AnalyzerRule},
     prelude::Expr,
     scalar::ScalarValue,
-    sql::TableReference,
 };
 
 use datafusion::logical_expr::{
@@ -535,8 +538,15 @@ impl TreeNodeRewriter for TypeCoercionRewriter<'_> {
                     .iter()
                     .map(|list_expr| list_expr.get_type(self.schema))
                     .collect::<Result<Vec<_>>>()?;
-                let result_type =
-                    get_coerce_type_for_list(&expr_data_type, &list_data_types);
+                // DataFusion coerces IN lists with the numeric-preferring
+                // `comparison_coercion`, which would cast a string column to a number
+                // (`CAST(s AS BIGINT) IN (1, 2)`) and fail in the target database on
+                // non-numeric values. Keep the string-preferring coercion instead.
+                let result_type = list_data_types
+                    .iter()
+                    .try_fold(expr_data_type.clone(), |left_type, right_type| {
+                        type_union_coercion(&left_type, right_type)
+                    });
                 match result_type {
                     None => plan_err!(
                         "Can not find compatible types to compare {expr_data_type:?} with {list_data_types:?}"
@@ -641,6 +651,41 @@ impl TreeNodeRewriter for TypeCoercionRewriter<'_> {
                     },
                 })))
             }
+            Expr::HigherOrderFunction(HigherOrderFunction { func, args }) => {
+                let current_fields = args
+                    .iter()
+                    .map(|arg| match arg {
+                        Expr::Lambda(lambda) => Ok(ValueOrLambda::Lambda(
+                            lambda.body.to_field(self.schema)?.1,
+                        )),
+                        _ => Ok(ValueOrLambda::Value(arg.to_field(self.schema)?.1)),
+                    })
+                    .collect::<Result<Vec<_>>>()?;
+
+                let new_fields = value_fields_with_higher_order_udf_and_lambdas(
+                    &current_fields,
+                    func.as_ref(),
+                )?;
+
+                let new_args = std::iter::zip(args, new_fields)
+                    .map(|(arg, new_field)| match (&arg, new_field) {
+                        (Expr::Lambda(_), ValueOrLambda::Lambda(_)) => Ok(arg),
+                        (Expr::Lambda(_), ValueOrLambda::Value(_)) => internal_err!(
+                            "value_fields_with_higher_order_udf returned a value for a lambda argument"
+                        ),
+                        (_, ValueOrLambda::Value(new_field)) => {
+                            arg.cast_to(new_field.data_type(), self.schema)
+                        }
+                        (_, ValueOrLambda::Lambda(_)) => internal_err!(
+                            "value_fields_with_higher_order_udf returned a lambda for a value argument"
+                        ),
+                    })
+                    .collect::<Result<_>>()?;
+
+                Ok(Transformed::yes(Expr::HigherOrderFunction(
+                    HigherOrderFunction::new(func, new_args),
+                )))
+            }
             // TODO: remove the next line after `Expr::Wildcard` is removed
             #[expect(deprecated)]
             Expr::Alias(_)
@@ -656,7 +701,9 @@ impl TreeNodeRewriter for TypeCoercionRewriter<'_> {
             | Expr::Wildcard { .. }
             | Expr::GroupingSet(_)
             | Expr::Placeholder(_)
-            | Expr::OuterReferenceColumn(_, _) => Ok(Transformed::no(expr)),
+            | Expr::OuterReferenceColumn(_, _)
+            | Expr::Lambda(_)
+            | Expr::LambdaVariable(_) => Ok(Transformed::no(expr)),
         }
     }
 }
@@ -967,8 +1014,7 @@ fn coerce_case_expression(case: Case, schema: &DFSchema) -> Result<Case> {
                 .iter()
                 .map(|(when, _then)| when.get_type(schema))
                 .collect::<Result<Vec<_>>>()?;
-            let coerced_type =
-                get_coerce_type_for_case_expression(&when_types, Some(case_type));
+            let coerced_type = get_coerce_type_for_case_when(&when_types, case_type);
             coerced_type.ok_or_else(|| {
                 plan_datafusion_err!(
                     "Failed to coerce case ({case_type:?}) and when ({when_types:?}) \
@@ -1072,7 +1118,7 @@ fn coerce_union_schema_with_schema(
             plan_schema.fields().iter()
         ) {
             let coerced_type =
-                comparison_coercion(union_datatype, plan_field.data_type()).ok_or_else(
+                type_union_coercion(union_datatype, plan_field.data_type()).ok_or_else(
                     || {
                         plan_datafusion_err!(
                             "Incompatible inputs for Union: Previous inputs were \

core/wren-core/core/src/logical_plan/utils.rs

@@ -13,6 +13,7 @@ use datafusion::common::types::{
     logical_binary, logical_boolean, logical_date, logical_float16, logical_float32,
     logical_float64, logical_string,
 };
+use datafusion::common::TableReference;
 use datafusion::common::{plan_err, DFSchema, DFSchemaRef};
 use datafusion::datasource::DefaultTableSource;
 use datafusion::error::Result;
@@ -22,7 +23,6 @@ use datafusion::logical_expr::{builder::LogicalTableSource, Expr, TableSource};
 use datafusion::logical_expr::{Coercion, TypeSignatureClass};
 use datafusion::sql::sqlparser::ast;
 use datafusion::sql::sqlparser::parser::Parser;
-use datafusion::sql::TableReference;
 use log::debug;
 use petgraph::dot::{Config, Dot};
 use petgraph::Graph;
@@ -406,7 +406,10 @@ pub fn expr_to_columns(
             | Expr::InSubquery(_)
             | Expr::ScalarSubquery(_)
             | Expr::Wildcard { .. }
-            | Expr::Placeholder(_) => {}
+            | Expr::Placeholder(_)
+            | Expr::HigherOrderFunction(_)
+            | Expr::Lambda(_)
+            | Expr::LambdaVariable(_) => {}
             Expr::SetComparison(_) => {}
         }
         Ok(TreeNodeRecursion::Continue)

core/wren-core/core/src/mdl/context.rs

@@ -1,4 +1,3 @@
-use std::any::Any;
 use std::collections::HashMap;
 use std::ops::Deref;
 use std::sync::Arc;
@@ -19,6 +18,7 @@ use datafusion::catalog::memory::{MemoryCatalogProvider, MemoryCatalogProviderLi
 use datafusion::catalog::CatalogProvider;
 use datafusion::catalog::CatalogProviderList;
 use datafusion::catalog::{MemorySchemaProvider, Session};
+use datafusion::common::TableReference;
 use datafusion::common::{internal_err, Result};
 use datafusion::datasource::{TableProvider, TableType, ViewTable};
 use datafusion::execution::session_state::SessionStateBuilder;
@@ -36,7 +36,6 @@ use datafusion::optimizer::{AnalyzerRule, OptimizerRule};
 use datafusion::physical_plan::ExecutionPlan;
 use datafusion::prelude::SessionContext;
 use datafusion::scalar::ScalarValue;
-use datafusion::sql::TableReference;
 use parking_lot::RwLock;
 
 pub type SessionPropertiesRef = Arc<HashMap<String, Option<String>>>;
@@ -429,10 +428,6 @@ impl WrenDataSource {
 
 #[async_trait]
 impl TableProvider for WrenDataSource {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
     fn schema(&self) -> SchemaRef {
         self.schema.clone()
     }

core/wren-core/core/src/mdl/function/remote_function.rs

@@ -10,7 +10,6 @@ use datafusion::logical_expr::{
     Signature, TypeSignature, Volatility, WindowUDFImpl,
 };
 use serde::{Deserialize, Serialize};
-use std::any::Any;
 use std::fmt::Display;
 use std::str::FromStr;
 use std::sync::Arc;
@@ -288,10 +287,6 @@ fn build_document(func: &RemoteFunction) -> Documentation {
 }
 
 impl ScalarUDFImpl for ByPassScalarUDF {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
     fn name(&self) -> &str {
         &self.name
     }
@@ -389,10 +384,6 @@ impl From<RemoteFunction> for ByPassAggregateUDF {
 }
 
 impl AggregateUDFImpl for ByPassAggregateUDF {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
     fn name(&self) -> &str {
         &self.name
     }
@@ -470,10 +461,6 @@ impl From<RemoteFunction> for ByPassWindowFunction {
 }
 
 impl WindowUDFImpl for ByPassWindowFunction {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
     fn name(&self) -> &str {
         &self.name
     }

core/wren-core/core/src/mdl/function/scalar/to_char.rs

@@ -1,5 +1,3 @@
-use std::any::Any;
-
 use datafusion::arrow::datatypes::DataType;
 use datafusion::common::not_impl_err;
 use datafusion::common::types::{
@@ -53,10 +51,6 @@ impl ToCharFunc {
 }
 
 impl ScalarUDFImpl for ToCharFunc {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
     fn name(&self) -> &str {
         "to_char"
     }

core/wren-core/core/src/mdl/lineage.rs

@@ -2,9 +2,9 @@ use std::collections::{HashMap, HashSet};
 use std::fmt::Display;
 use std::sync::Arc;
 
+use datafusion::common::TableReference;
 use datafusion::common::{internal_err, plan_err, Column};
 use datafusion::error::Result;
-use datafusion::sql::TableReference;
 use petgraph::Graph;
 
 use crate::logical_plan::utils::from_qualified_name;
@@ -388,9 +388,9 @@ fn get_dataset_link_revers_if_need(
 
 #[cfg(test)]
 mod test {
+    use datafusion::common::TableReference;
     use datafusion::common::{Column, Spans};
     use datafusion::error::Result;
-    use datafusion::sql::TableReference;
     use std::collections::{HashMap, HashSet};
     use std::sync::Arc;
 

core/wren-core/core/src/mdl/mod.rs

@@ -13,6 +13,7 @@ use crate::mdl::utils::{dequote_identifier, quoted, to_field};
 use crate::DataFusionError;
 use context::SessionPropertiesRef;
 use datafusion::arrow::datatypes::Field;
+use datafusion::common::TableReference;
 use datafusion::common::{internal_datafusion_err, plan_err};
 use datafusion::datasource::TableProvider;
 use datafusion::error::Result;
@@ -24,7 +25,6 @@ use datafusion::sql::parser::DFParser;
 use datafusion::sql::sqlparser::ast::{Expr, ExprWithAlias, Ident};
 use datafusion::sql::sqlparser::dialect::dialect_from_str;
 use datafusion::sql::unparser::Unparser;
-use datafusion::sql::TableReference;
 pub use dataset::Dataset;
 use dialect::WrenDialect;
 use log::{debug, info, warn};
@@ -944,6 +944,56 @@ mod test {
         Ok(())
     }
 
+    #[tokio::test]
+    async fn test_in_list_mixed_string_numeric_prefers_string() -> Result<()> {
+        let manifest = ManifestBuilder::new()
+            .catalog("wren")
+            .schema("test")
+            .model(
+                ModelBuilder::new("customer")
+                    .table_reference("customer")
+                    .column(ColumnBuilder::new("c_custkey", "int").build())
+                    .column(ColumnBuilder::new("c_name", "varchar").build())
+                    .build(),
+            )
+            .build();
+        let analyzed_mdl = Arc::new(AnalyzedWrenMDL::analyze(
+            manifest,
+            Arc::new(HashMap::default()),
+            Mode::Unparse,
+        )?);
+        let ctx = create_wren_ctx(None, analyzed_mdl.wren_mdl().data_source().as_ref());
+
+        let sql = "select c_custkey from customer where c_name in (1, 2)";
+        let actual = mdl::transform_sql_with_ctx(
+            &ctx,
+            Arc::clone(&analyzed_mdl),
+            &[],
+            Arc::new(HashMap::new()),
+            sql,
+        )
+        .await?;
+        assert_snapshot!(actual,
+            @"SELECT customer.c_custkey FROM (SELECT customer.c_custkey, customer.c_name FROM \
+            (SELECT __source.c_custkey AS c_custkey, __source.c_name AS c_name FROM customer AS __source) AS customer) AS customer \
+            WHERE customer.c_name IN (CAST(1 AS VARCHAR), CAST(2 AS VARCHAR))");
+
+        let sql = "select c_custkey from customer where c_custkey in ('1', '2')";
+        let actual = mdl::transform_sql_with_ctx(
+            &ctx,
+            Arc::clone(&analyzed_mdl),
+            &[],
+            Arc::new(HashMap::new()),
+            sql,
+        )
+        .await?;
+        assert_snapshot!(actual,
+            @"SELECT customer.c_custkey FROM (SELECT customer.c_custkey FROM \
+            (SELECT __source.c_custkey AS c_custkey FROM customer AS __source) AS customer) AS customer \
+            WHERE CAST(customer.c_custkey AS VARCHAR) IN ('1', '2')");
+        Ok(())
+    }
+
     #[tokio::test]
     async fn test_remote_function() -> Result<()> {
         env_logger::init();
@@ -1210,7 +1260,10 @@ mod test {
             .await.map_err(|e| {
                 assert_snapshot!(
                     e.to_string(),
-                    @"Schema error: No field named \"名字\". Valid fields are wren.test.artist.\"串接名字\"."
+                    @r#"
+                Schema error: No field named "名字". Did you mean 'wren.test.artist."串接名字"'?
+                Valid fields are wren.test.artist."串接名字".
+                "#
                 )
             });
         Ok(())
@@ -3076,7 +3129,10 @@ mod test {
                 panic!("whitelist_name is hidden, it should not be selected directly")
             }
             Err(e) => {
-                assert_snapshot!(e.to_string(), @"Schema error: No field named whitelist_name. Valid fields are customer.c_custkey, customer.mock_id.")
+                assert_snapshot!(e.to_string(), @r"
+                Schema error: No field named whitelist_name.
+                Valid fields are customer.c_custkey, customer.mock_id.
+                ")
             }
         }
 
@@ -3227,7 +3283,10 @@ mod test {
             Err(e) => {
                 assert_snapshot!(
                     e.to_string(),
-                    @"Schema error: No field named c_name. Valid fields are customer.c_custkey."
+                    @r"
+                Schema error: No field named c_name.
+                Valid fields are customer.c_custkey.
+                "
                 )
             }
         }
@@ -5235,7 +5294,7 @@ mod test {
         use datafusion::catalog::memory::MemoryCatalogProvider;
         use datafusion::catalog::CatalogProvider;
         use datafusion::catalog::SchemaProvider;
-        use datafusion::sql::TableReference;
+        use datafusion::common::TableReference;
         use std::sync::Barrier;
         use std::thread;
 

core/wren-core/sqllogictest/test_files/model.slt

@@ -72,7 +72,7 @@ select "Customer_id" from wrenai.public."Orders" where exists (select 1 from wre
 04eafb40a16989307464f27f1fed8907
 0732c0881c70ebcda536a4b14e9db106
 
-query RIITRTTT
+query RIITRTTT rowsort
 select * from "Order_items" where "Order_id" in ('03c83b31dbc387f83f1b5579b53182fb', '08cbb1d4cd574b126569b208fd4b26ea')
 ----
 14.68 1 1 03c83b31dbc387f83f1b5579b53182fb 119.8 a04087ab6a96ffa041f8a2701a72b616 2023/1/15 7:26 CA

core/wren-core/sqllogictest/test_files/tpch/q22.slt.part

@@ -16,7 +16,7 @@
 # specific language governing permissions and limitations
 # under the License.
 
-query ?IR
+query TIR
 select
     cntrycode,
     count(*) as numcust,