From a9ed71399b5b0c973fe0a476486137512964d714 Mon Sep 17 00:00:00 2001
From: tfenne <tfenne@tfenne.com>
Date: Mon, 5 Oct 2026 16:28:31 -0600
Subject: [PATCH] Read negative compression levels correctly where char is
 unsigned

The array schema stored each compression level in a char. char is
unsigned on Linux aarch64 and other non-x86 Linux targets, so there a
level of -1, GZIP's default, read back as 255 and GZIP compression
failed with Z_STREAM_ERROR. The levels are now stored and read as int8_t:
the bytes on disk are unchanged, and they read the same on every
platform. Adds a test that GZIP's default level reads back as written.
---
 core/src/array/array_schema.cc          | 11 ++++++-----
 test/src/c_api/test_array_schema_api.cc | 14 ++++++++++++++
 2 files changed, 20 insertions(+), 5 deletions(-)

diff --git a/core/src/array/array_schema.cc b/core/src/array/array_schema.cc
index c5d9eb5..4012e3a 100644
--- a/core/src/array/array_schema.cc
+++ b/core/src/array/array_schema.cc
@@ -773,8 +773,8 @@ int ArraySchema::serialize(
     memcpy(buffer + offset, &compression, sizeof(char));
     offset += sizeof(char);
   }
-  // Copy compression_level_
-  char compression_level;
+  // Copy compression_level_. Levels can be negative, and char is unsigned on some platforms
+  int8_t compression_level;
   for(int i=0; i<=attribute_num_; ++i) {
     compression_level = compression_level_[i];
     assert(offset + sizeof(char) <= buffer_size);
@@ -790,7 +790,7 @@ int ArraySchema::serialize(
     offset += sizeof(char);
   }
   // Copy offsets compression_level_
-  char offsets_compression_level;
+  int8_t offsets_compression_level;
   for(int i=0; i<attribute_num_; ++i) {
     offsets_compression_level = offsets_compression_level_[i];
     assert(offset + sizeof(char) <= buffer_size);
@@ -1198,7 +1198,8 @@ int ArraySchema::deserialize(
   }
   // Load compression_level_. Support added in array schema version 1L
   if (get_version() >= 1L) {
-    char compression_level;
+    // Levels can be negative, and char is unsigned on some platforms
+    int8_t compression_level;
     for(int i=0; i<=attribute_num_; ++i) {
       assert(offset + sizeof(char) <= buffer_size);
       memcpy(&compression_level, buffer + offset, sizeof(char));
@@ -1218,7 +1219,7 @@ int ArraySchema::deserialize(
   }
   // Load offsets_compression_level_. Support added in array schema version 2L
   if (get_version() >= 2L) {
-    char offsets_compression_level;
+    int8_t offsets_compression_level;
     for(int i=0; i<attribute_num_; ++i) {
       assert(offset + sizeof(char) <= buffer_size);
       memcpy(&offsets_compression_level, buffer + offset, sizeof(char));
diff --git a/test/src/c_api/test_array_schema_api.cc b/test/src/c_api/test_array_schema_api.cc
index a1085bb..701e6d2 100644
--- a/test/src/c_api/test_array_schema_api.cc
+++ b/test/src/c_api/test_array_schema_api.cc
@@ -221,6 +221,20 @@ TEST_CASE_METHOD(ArraySchemaTestFixture, "Test Array Schema char", "[array_schem
   check_dense_array(array_name);
 }
 
+TEST_CASE_METHOD(ArraySchemaTestFixture, "A negative compression level is read back as negative",
+                 "[array_schema_compression_level]") {
+  std::string array_name = array_name_ + "gzip_default_level";
+
+  // With no compression levels given, GZIP gets its default, which is negative
+  REQUIRE(TILEDB_COMPRESSION_LEVEL_GZIP < 0);
+  REQUIRE(create_dense_array(array_name, TILEDB_INT32, TILEDB_GZIP) == TILEDB_OK);
+
+  TileDB_ArraySchema array_schema_disk;
+  REQUIRE(tiledb_array_load_schema(tiledb_ctx_, array_name.c_str(), &array_schema_disk) == TILEDB_OK);
+  CHECK(array_schema_disk.compression_level_[0] == TILEDB_COMPRESSION_LEVEL_GZIP);
+  REQUIRE(tiledb_array_free_schema(&array_schema_disk) == TILEDB_OK);
+}
+
 TEST_CASE_METHOD(ArraySchemaTestFixture, "Test Array Schema int8", "[array_schema_int8]") {
   std::string array_name = array_name_ + "int8";