Use Python buffer protocol when converting Python objects to .NET arrays #1838
Comments
|
If somebody wants to take on it, this test shows simple use of And this is the method that needs to be changed: Converter.ToArray |
|
This would not break API/ABI, so we can integrate it in a normal point release after 3.0, right? |
|
Yep. |
|
Copying data from C# array to Python list or Numpy array is also a common task. Is there a more efficient way? |
|
Agreed, converting a python array to a C# array is a very common task, and it is currently extremely slow. It would be fantastic to make it faster. Is it currently possibly with Python (and Python.NET) only, without any C# programming? |
|
Hi! I’m looking for some guidance on how to speed up copying from Python to C#. |
|
Hi, is someone currently working on this? If not I'd be happy to try. I've been using pythonnet at work successfully for years and would be glad to help. I'm familiar with the topic because I have already dealt with converting numpy arrays from/to C# arrays in an efficient manner. It'd be great to have this natively in pythonnet. |
|
@chickenservice To my knowledge, no one is currently actively working on this, so give it a shot! An implementation would be greatly appreciated! |
|
@filmor Perfect, is there anything besides the steps in CONTRIBUTING.md that I need to be aware of? |
Copying data from NumPy array to C# is a common task. Python.NET implements a simple sequence-to-array copying mechanism, which uses iterators. It is extremely inefficient for large arrays of primitive types.
Now we have a better way - using
PyBufferand buffer protocol. We could detect, that Python object implements the buffer protocol, use it to copy data when possible, and fallback to iteration otherwise.The text was updated successfully, but these errors were encountered: