[ARROW-2264] [Python] Efficiently serialize numpy arrays with dtype of unicode fixed length string

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.8.0
• Fix Version/s: 0.10.0
• Component/s: Python
• External issue URL: https://github.com/apache/arrow/issues/18222
• Assignee: Robert Nishihara
• Reporter: Mitar
• Created: 05/Mar/18 19:54
• Updated: 11/Jan/23 07:19
• Resolved: 21/Apr/18 22:56

Description

Looking at the numpy array serialization code it seems that if I have a dtype like "<U3" this will go through custom ndarray serializer and not through an efficient one.

Example:>>> np.array(['aaa', 'bbb'])
array(['aaa', 'bbb'], dtype='<U3')

This should be able to work, no? It has fixed offsets and memory layout.