[ARROW-3048] [Python] Import pyarrow fails if scikit-learn is installed from conda (boost-cpp / libboost issue)

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.10.0
• Fix Version/s: 0.12.0
• Component/s: Python
• Environment: Ubuntu 16.04
• External issue URL: https://github.com/apache/arrow/issues/19407
• Assignee: Unassigned
• Reporter: Jarno Seppanen
• Created: 13/Aug/18 14:24
• Updated: 11/Jan/23 07:24
• Resolved: 23/Jan/19 12:55

Description

Hi, installing both pyarrow 0.10.0 and scikit-learn 0.19.2 causes pyarrow import to break.

Steps to reproduce

1. cat >environment.yml <<EOF

name: asdf
channels:
- defaults
- conda-forge
dependencies:
- python=3.6
- pyarrow=0.10.0
- scikit-learn=0.19.2

EOF

1. conda env create
2. source activate asdf
3. python -c 'import pyarrow'

Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/home/jarno/miniconda3/envs/asdf/lib/python3.6/site-packages/pyarrow/__init__.py", line 60, in <module>
from pyarrow.lib import cpu_count, set_cpu_count
ImportError: /home/jarno/miniconda3/envs/asdf/lib/python3.6/site-packages/pyarrow/../../../libparquet.so.1: undefined symbol: _ZN5boost13match_resultsIN9__gnu_cxx17__normal_iteratorIPKcSsEESaINS_9sub_matchIS5_EEEE12maybe_assignERKS9_