我的RedHat7机器上安装了python2.7和python3.4。我为这两个版本安装了tensorflow。
P2.7似乎工作正常：
Python 2.7.5 (default, Oct 11 2015, 17:47:16)
[GCC 4.8.3 20140911 (Red Hat 4.8.3-9)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import site; site.getsitepackages()
['/usr/lib64/python2.7/site-packages', '/usr/lib/python2.7/site-packages', '/usr/lib/site-python']
>>> import tensorflow
>>>
对于p3.4我得到了：
Python 3.4.3 (default, Jan 26 2016, 02:25:35)
[GCC 4.8.5 20150623 (Red Hat 4.8.5-4)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import site; site.getsitepackages()
['/usr/lib64/python3.4/site-packages', '/usr/lib/python3.4/site-packages', '/usr/lib/site-python']
>>> import tensorflow
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/python2.7/site-packages/tensorflow/__init__.py", line 23, in <module>
from tensorflow.python import *
File "/usr/lib/python2.7/site-packages/tensorflow/python/__init__.py", line 48, in <module>
from tensorflow.python import pywrap_tensorflow
File "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 28, in <module>
_pywrap_tensorflow = swig_import_helper()
File "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow', fp, pathname, description)
File "/usr/lib64/python3.4/imp.py", line 243, in load_module
return load_dynamic(name, filename, file)
ImportError: /usr/lib/python2.7/site-packages/tensorflow/python/_pywrap_tensorflow.so: undefined symbol: PyClass_Type
>>>
为什么p3试图抓住p2包？
更新＃1：
sys.path制作了['', '/usr/lib64/python34.zip', '/usr/lib64/python3.4', '/usr/lib64/python3.4/plat-linux', '/usr/lib64/python3.4/lib-dynload', '/usr/lib64/python3.4/site-packages', '/usr/lib/python3.4/site-packages'] PYTHONPATH未定义更新＃2 ：正如Martijn建议我跑
python3 -v -c 'import tensorflow' > ~/log.txt 2>&1
并看到以下块
...
import 'types' # <_frozen_importlib.SourceFileLoader object at 0x7fd5b3e7e8d0>
import 'site' # <_frozen_importlib.SourceFileLoader object at 0x7fd5b3f4bba8>
# code object from /usr/lib/python2.7/site-packages/tensorflow/__init__.py
# could not create '/usr/lib/python2.7/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc': PermissionError(13, 'Permission denied')
# wrote '/usr/lib/python2.7/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc'
# /usr/lib64/python3.4/__pycache__/__future__.cpython-34.pyc matches /usr/lib64/python3.4/__future__.py
# code object from '/usr/lib64/python3.4/__pycache__/__future__.cpython-34.pyc'
import '__future__' # <_frozen_importlib.SourceFileLoader object at 0x7fd5b3ea04e0>
# code object from /usr/lib/python2.7/site-packages/tensorflow/python/__init__.py
# could not create '/usr/lib/python2.7/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc': PermissionError(13, 'Permission denied')
# wrote '/usr/lib/python2.7/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc'
...
我运行了以下命令来消除这些错误：
sudo chmod -R ugo+rX /usr/lib/python2.7/site-packages/tensorflow/__pycache__
sudo chmod -R ugo+rX /usr/lib/python2.7/site-packages/tensorflow/python/__pycache__/
但我仍然得到以下内容：
import 'imp' # <_frozen_importlib.SourceFileLoader object at 0x7fb372a1f320>
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "<frozen importlib._bootstrap>", line 2237, in _find_and_load
File "<frozen importlib._bootstrap>", line 2226, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 1200, in _load_unlocked
File "<frozen importlib._bootstrap>", line 1129, in _exec
File "<frozen importlib._bootstrap>", line 1471, in exec_module
File "<frozen importlib._bootstrap>", line 321, in _call_with_frames_removed
File "/usr/lib/python2.7/site-packages/tensorflow/__init__.py", line 23, in <module>
from tensorflow.python import *
File "<frozen importlib._bootstrap>", line 2237, in _find_and_load
File "<frozen importlib._bootstrap>", line 2226, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 1200, in _load_unlocked
File "<frozen importlib._bootstrap>", line 1129, in _exec
File "<frozen importlib._bootstrap>", line 1471, in exec_module
File "<frozen importlib._bootstrap>", line 321, in _call_with_frames_removed
File "/usr/lib/python2.7/site-packages/tensorflow/python/__init__.py", line 48, in <module>
from tensorflow.python import pywrap_tensorflow
File "<frozen importlib._bootstrap>", line 2284, in _handle_fromlist
File "<frozen importlib._bootstrap>", line 321, in _call_with_frames_removed
File "<frozen importlib._bootstrap>", line 2237, in _find_and_load
File "<frozen importlib._bootstrap>", line 2226, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 1200, in _load_unlocked
File "<frozen importlib._bootstrap>", line 1129, in _exec
File "<frozen importlib._bootstrap>", line 1471, in exec_module
File "<frozen importlib._bootstrap>", line 321, in _call_with_frames_removed
File "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 28, in <module>
_pywrap_tensorflow = swig_import_helper()
File "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow', fp, pathname, description)
File "/usr/lib64/python3.4/imp.py", line 243, in load_module
return load_dynamic(name, filename, file)
ImportError: /usr/lib/python2.7/site-packages/tensorflow/python/_pywrap_tensorflow.so: undefined symbol: PyClass_Type
# clear builtins._
# clear sys.path
更新＃3：我要求我的系统管理员运行以安装TF for p3.4的命令是直接从TF安装页面提取的：
# Ubuntu/Linux 64-bit, CPU only, Python 3.4
export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.9.0-cp34-cp34m-linux_x86_64.whl
sudo pip3 install --upgrade $TF_BINARY_URL
pip3是否可能以某种方式失控并指向p2.7而不是p3.4？也就是说，我认为TF安装的一部分是安装numpy，并且在将其导入p3.4时没有出现任何错误，因此可能还有其他事情发生。
更新＃4：我已经问过我们的系统管理员（最初为我安装了TF）在他身边运行python3 -v -c 'import tensorflow'命令。他的结果与我的不同：python3拉出正确的库，TF加载很好。这是我日志的摘录：
# /usr/lib/python2.7/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc matches /usr/lib/python2.7/site-packages/tensorflow/__init__.py
# code object from '/usr/lib/python2.7/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc'
# /usr/lib/python2.7/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc matches /usr/lib/python2.7/site-packages/tensorflow/python/__init__.py
# code object from '/usr/lib/python2.7/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc'
这是来自他的：
# /usr/lib/python3.4/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc matches /usr/lib/python3.4/site-packages/tensorflow/__init__.py
# code object from '/usr/lib/python3.4/site-packages/tensorflow/__pycache__/__init__.cpython-34.pyc'
# /usr/lib/python3.4/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc matches /usr/lib/python3.4/site-packages/tensorflow/python/__init__.py
# code object from '/usr/lib/python3.4/site-packages/tensorflow/python/__pycache__/__init__.cpython-34.pyc'
PYTHONPATH既不是为他定义，也不是为我定义。那么为什么python对我们的工作方式不同呢？