Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

pyproj >= 2 segmentation fault

See original GitHub issue

This code cause Seg-fault on conda env (python 3.6.7)

import geopandas
df = geopandas.read_file(geopandas.datasets.get_path('nybb'))
d  = df.to_crs({'init': 'epsg:4326', 'no_defs': True})

#  python -i ./test.py
[1]    16967 segmentation fault (core dumped)  python -i ./test.py
# pip show geopandas
Name: geopandas
Version: 0.5.0
Summary: Geographic pandas extensions
Home-page: http://geopandas.org
Author: GeoPandas contributors
Author-email: kjordahl@alum.mit.edu
License: BSD
Location: /home/vlz/miniconda3/envs/ocli/lib/python3.6/site-packages
Requires: shapely, fiona, pandas, pyproj
Required-by: ocli
# pip show pyproj
Name: pyproj
Version: 2.2.1
Summary: Python interface to PROJ (cartographic projections and coordinate transformations library)
Home-page: https://github.com/pyproj4/pyproj
Author: Jeff Whitaker
Author-email: jeffrey.s.whitaker@noaa.gov
License: OSI Approved
Location: /home/vlz/miniconda3/envs/ocli/lib/python3.6/site-packages
Requires:
Required-by: OWSLib, geopandas

Issue Analytics

State:
Created 4 years ago
Comments:10 (5 by maintainers)

Top GitHub Comments

1reaction

Zaitsevcommented, Jul 4, 2019

Thanks. Seems my env really corrupted

0reactions

ocefpafcommented, Jul 4, 2019

Indeed the mix of channels and pip/conda sources there make it really hard to debug and highly unlikely to work. I don’t know the recommendation for bioconda these days but I recall that the preferred channel order, to ensure stability, was:

channels:
  - conda-forge
  - bioconda
  - defaults

Also, if you are not using bioconda I recommend removing that. In addition be sure that you are using latest conda (conda update conda) and activate strict channel option (conda config --set channel_priority strict). After that recreate your working env and avoid as much as you can installing things from pip. (You have two numpy version there b/c of this!)

BTW, I cannot reproduce your segfault with a clean env and all the package combinations possible in conda-forge, so I’m pretty much sure the problem is your env.