Importing torch failed when lazyimported cv2 first

Question

I am using self-compiled python 3.6.3 (link) on Ubuntu 14.04. My torch is 1.0.1.post2, Tensorlayer 1.10.1. To reproduce my issue, try with

from tensorlayer.lazy_imports import LazyImport
cv2 = LazyImport("cv2")
import torch

It fails with,

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/local/lib/python3.6/site-packages/torch/__init__.py", line 84, in <module>
    from torch._C import *
ImportError: dlopen: cannot load any more object with static TLS

My torch is installed with pip install torch torchvision. Also before compiling Python, I did apt-get install tk8.6-dev in order to make python configured with tkinter. The reason why I compile Python on my own is that shipped apt-get install-ed Python is incompatible with tkinter somehow, even with packages installed, it errs at importing too.

I wish not to change the importing order at it surely works. I hope to find out the real cause of it. Many thanks in advance!

There are some old posts, where they change the order or import cv2 again after tensorflow. Have you tried that? — Garvita Tiwari, Apr 12 '19 at 12:10
@GarvitaTiwari sure, it won't work. I upgraded py3.6.3 to py3.6.4 and the error disappears. — MeadowMuffins, May 16 '19 at 06:52

Importing torch failed when lazyimported cv2 first

0 Answers0