From cb3759e44b4fb382d66aa3ee1ea1474be9537be2 Mon Sep 17 00:00:00 2001
From: eldenmoon <lihangyu@selectdb.com>
Date: Fri, 20 Dec 2024 10:04:17 +0800
Subject: [PATCH] [refactor](variant) remove duplicated and trivial sanitize

1. could improve serialization performance
2. check is incorrect, example int8 and int16 could be compatible
---
 be/src/vec/columns/column_object.cpp          | 26 -------------------
 be/src/vec/columns/column_object.h            |  3 ---
 .../serde/data_type_object_serde.cpp          |  1 -
 3 files changed, 30 deletions(-)

diff --git a/be/src/vec/columns/column_object.cpp b/be/src/vec/columns/column_object.cpp
index d67a70d2f630f03..958f99201d28578 100644
--- a/be/src/vec/columns/column_object.cpp
+++ b/be/src/vec/columns/column_object.cpp
@@ -1354,14 +1354,6 @@ Status find_and_set_leave_value(const IColumn* column, const PathInData& path,
                                 TypeIndex base_type_index, rapidjson::Value& root,
                                 rapidjson::Document::AllocatorType& allocator, Arena& mem_pool,
                                 size_t row) {
-#ifndef NDEBUG
-    // sanitize type and column
-    if (column->get_name() != type->create_column()->get_name()) {
-        return Status::InternalError(
-                "failed to set value for path {}, expected type {}, but got {} at row {}",
-                path.get_path(), type->get_name(), column->get_name(), row);
-    }
-#endif
     const auto* nullable = check_and_get_column<ColumnNullable>(column);
     if (skip_empty_json(nullable, type, base_type_index, row, path)) {
         return Status::OK();
@@ -1952,24 +1944,6 @@ std::string ColumnObject::debug_string() const {
     return res.str();
 }
 
-Status ColumnObject::sanitize() const {
-    RETURN_IF_CATCH_EXCEPTION(check_consistency());
-    for (const auto& subcolumn : subcolumns) {
-        if (subcolumn->data.is_finalized()) {
-            auto column = subcolumn->data.get_least_common_type()->create_column();
-            std::string original = subcolumn->data.get_finalized_column().get_name();
-            std::string expected = column->get_name();
-            if (original != expected) {
-                return Status::InternalError("Incompatible type between {} and {}, debug_info:",
-                                             original, expected, debug_string());
-            }
-        }
-    }
-
-    VLOG_DEBUG << "sanitized " << debug_string();
-    return Status::OK();
-}
-
 ColumnObject::Subcolumn ColumnObject::Subcolumn::cut(size_t start, size_t length) const {
     Subcolumn new_subcolumn(0, is_nullable);
     new_subcolumn.insert_range_from(*this, start, length);
diff --git a/be/src/vec/columns/column_object.h b/be/src/vec/columns/column_object.h
index e4127197a22b025..0f25a5e68d0c422 100644
--- a/be/src/vec/columns/column_object.h
+++ b/be/src/vec/columns/column_object.h
@@ -428,9 +428,6 @@ class ColumnObject final : public COWHelper<IColumn, ColumnObject> {
 
     bool empty() const;
 
-    // Check if all columns and types are aligned
-    Status sanitize() const;
-
     std::string debug_string() const;
 
     void update_hashes_with_value(uint64_t* __restrict hashes,
diff --git a/be/src/vec/data_types/serde/data_type_object_serde.cpp b/be/src/vec/data_types/serde/data_type_object_serde.cpp
index fc536d9ef0df7b6..c1e8645a6d2d8e4 100644
--- a/be/src/vec/data_types/serde/data_type_object_serde.cpp
+++ b/be/src/vec/data_types/serde/data_type_object_serde.cpp
@@ -51,7 +51,6 @@ Status DataTypeObjectSerDe::_write_column_to_mysql(const IColumn& column,
     if (!variant.is_finalized()) {
         const_cast<ColumnObject&>(variant).finalize();
     }
-    RETURN_IF_ERROR(variant.sanitize());
     if (variant.is_scalar_variant()) {
         // Serialize scalar types, like int, string, array, faster path
         const auto& root = variant.get_subcolumn({});