Incomplete license header for `src/libImaging/Arrow.h`

[stefan6419846]

What did you do?

Reviewed distributed/bundled files.

What did you expect to happen?

License headers of third-party code are applied correctly. Source most likely is https://github.com/apache/arrow/blob/main/cpp/src/arrow/c/abi.h

What actually happened?

Only a minimal header is included, as well as omitting the corresponding copyright data from the third-party licenses:

Pillow/src/libImaging/Arrow.h

Lines 4 to 6 in 9c1097c

	// Apache License 2.0.
	// Source apache arrow project
	// https://arrow.apache.org/docs/format/CDataInterface.html

As the arrow project has a NOTICE.txt file, this further violates the terms of the Apache-2.0 license: https://github.com/apache/arrow/blob/main/NOTICE.txt

Additionally, executed changes have to be specified explicitly - doing a short diff of the file I found versus the file shipped by Pillow clearly show omissions and formatting changes.

Relevant sections from the license text: https://github.com/apache/arrow/blob/cb7383c71b7c87d45946eec1ba967472592e036b/LICENSE.txt#L98-L122

What are your OS, Python and Pillow versions?

• OS: irrelevant
• Python: irrelevant
• Pillow: main

[wiredfool]

No, source is exactly here: https://arrow.apache.org/docs/format/CDataInterface.html#structure-definitions

Structure definitions

The following free-standing definitions are enough to support the Arrow C data interface in your project. Like the rest of the Arrow project, they are available under the Apache License 2.0.

#ifndef ARROW_C_DATA_INTERFACE
#define ARROW_C_DATA_INTERFACE

#define ARROW_FLAG_DICTIONARY_ORDERED 1
#define ARROW_FLAG_NULLABLE 2
#define ARROW_FLAG_MAP_KEYS_SORTED 4

struct ArrowSchema {
  // Array type description
  const char* format;
  const char* name;
  const char* metadata;
  int64_t flags;
  int64_t n_children;
  struct ArrowSchema** children;
  struct ArrowSchema* dictionary;

  // Release callback
  void (*release)(struct ArrowSchema*);
  // Opaque producer-specific data
  void* private_data;
};

struct ArrowArray {
  // Array data description
  int64_t length;
  int64_t null_count;
  int64_t offset;
  int64_t n_buffers;
  int64_t n_children;
  const void** buffers;
  struct ArrowArray** children;
  struct ArrowArray* dictionary;

  // Release callback
  void (*release)(struct ArrowArray*);
  // Opaque producer-specific data
  void* private_data;
};

#endif  // ARROW_C_DATA_INTERFACE

Note

The canonical guard ARROW_C_DATA_INTERFACE is meant to avoid duplicate definitions if two projects copy the C data interface definitions in their own headers, and a third-party project includes from these two projects. It is therefore important that this guard is kept exactly as-is when these definitions are copied.

This is all that we are including from the arrow project, so none of the third party libraries or sources are included. We do use them for testing purposes, but they are not used for the arrow support.

[stefan6419846]

Thanks for the hint about the source - while there has been a link, I must have somehow missed the actual code snippet.

Apart from this, I still consider the current attribution insufficient for the Apache-2.0 licensed code from the Arrow project. (Maybe they should provide an appropriate header in the snippet upstream as well, but this is a slightly different topic. Pillow reproduces this header and should follow the requirements of the Apache-2.0 license here.)