wjones127 commented on code in PR #34133: URL: https://github.com/apache/arrow/pull/34133#discussion_r1142814166
########## csharp/src/Apache.Arrow/C/CArrowSchemaImporter.cs: ########## @@ -0,0 +1,285 @@ +// Licensed to the Apache Software Foundation (ASF) under one or more +// contributor license agreements. See the NOTICE file distributed with +// this work for additional information regarding copyright ownership. +// The ASF licenses this file to You under the Apache License, Version 2.0 +// (the "License"); you may not use this file except in compliance with +// the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + + +using System; +using System.Collections.Generic; +using System.IO; +using System.Linq; +using Apache.Arrow.Types; + +namespace Apache.Arrow.C +{ + public static class CArrowSchemaImporter + { + /// <summary> + /// Import C pointer as an <see cref="ArrowType"/>. + /// </summary> + /// <examples> + /// Typically, you will allocate a uninitialized CArrowSchema pointer, + /// pass that to external function, and then use this method to import + /// the result. + /// + /// <code> + /// CArrowSchema* importedPtr = CArrowSchema.New(); + /// foreign_export_function(importedPtr); + /// ArrowType importedType = CArrowSchemaImporter.ImportType(importedPtr); + /// CArrowSchema.Free(importedPtr); + /// </code> + /// </examples> + public static unsafe ArrowType ImportType(CArrowSchema* ptr) + { + using var importedType = new ImportedArrowSchema(ptr); + return importedType.GetAsType(); + } + + /// <summary> + /// Import C pointer as an <see cref="Field"/>. + /// </summary> + /// <examples> + /// Typically, you will allocate a uninitialized CArrowSchema pointer, + /// pass that to external function, and then use this method to import + /// the result. + /// + /// <code> + /// CArrowSchema* importedPtr = CArrowSchema.New(); + /// foreign_export_function(importedPtr); + /// Field importedField = CArrowSchemaImporter.ImportField(importedPtr); + /// CArrowSchema.Free(importedPtr); + /// </code> + /// </examples> + public static unsafe Field ImportField(CArrowSchema* ptr) + { + using var importedField = new ImportedArrowSchema(ptr); + return importedField.GetAsField(); + } + + /// <summary> + /// Import C pointer as an <see cref="Schema"/>. + /// </summary> + /// <examples> + /// Typically, you will allocate a uninitialized CArrowSchema pointer, + /// pass that to external function, and then use this method to import + /// the result. + /// + /// <code> + /// CArrowSchema* importedPtr = CArrowSchema.New(); + /// foreign_export_function(importedPtr); + /// Field importedSchema = CArrowSchemaImporter.ImportSchema(importedPtr); + /// CArrowSchema.Free(importedPtr); + /// </code> + /// </examples> + public static unsafe Schema ImportSchema(CArrowSchema* ptr) + { + using var importedSchema = new ImportedArrowSchema(ptr); + return importedSchema.GetAsSchema(); + } + + private sealed unsafe class ImportedArrowSchema : IDisposable + { + private readonly CArrowSchema* _data; + private readonly bool _isRoot; + + public ImportedArrowSchema(CArrowSchema* handle) + { + _data = (CArrowSchema*)handle; + if (_data->release == null) + { + throw new ArgumentException("Tried to import a schema that has already been released."); + } + _isRoot = true; + } + + public ImportedArrowSchema(CArrowSchema* handle, bool isRoot) : this(handle) + { + _isRoot = isRoot; + } + + public void Dispose() + { + // We only call release on a root-level schema, not child ones. + if (_isRoot && _data->release != null) + { + _data->release(_data); Review Comment: > I feel this is a bad pattern - releasing resources that you don't own/didn't create. To be clear: releasing doesn't necessarily mean freeing memory. It could also mean decrementing a reference count on a shared pointer. The release callback is just a way for the consumer to inform the producer that it is no longer using those resources. If the producer also isn't using those resources, it will free them. That's commonly the case with `ArrowSchema`, since it's all just small temporary values we created for exchange. But in other cases it might keep them around. That will be the case with `ArrowArray`, where the buffers will be behind reference counted pointers that the producer might still have references to. > It would be really surprising to me that I couldn't get a CArrowSchema* from somewhere, and I couldn't call CArrowSchemaImporter.ImportSchema multiple times on it. I think it's basically just a trade off. Either you can only import once, and the release callback is handled for you. Or you are allowed to import multiple times, but you are responsible for also calling the release callback (or else risking memory leaks). It seems the current design expects that most situations people only care to import something once, and that it's more important to prevent accidental memory leaks than it is support multiple imports. > Can you show me in the code where release is called? Indeed it's a little complicated: 1. When you import, you create a `SchemaImporter`: https://github.com/apache/arrow/blob/08fb86107b001e2e1903ae0e416ff51386d4239e/cpp/src/arrow/c/bridge.cc#L1212-L1213 2. When the schema importer is passed the `ArrowSchema`, it passes that to an `ExportGuard` https://github.com/apache/arrow/blob/08fb86107b001e2e1903ae0e416ff51386d4239e/cpp/src/arrow/c/bridge.cc#L831 3. When the `SchemaImporter` destructor is called, the `ExportGuard` destructor is called, and that calls the release callback https://github.com/apache/arrow/blob/08fb86107b001e2e1903ae0e416ff51386d4239e/cpp/src/arrow/c/util_internal.h#L61 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
