Develop Reference

Something wrong with my conda channel and leads to a tensorflow import error

I created a Conda environment with python 3.6 and venv and activated it. then installed tensorflow with
pip install tensorflow
However, there is something wrong with my python. Basically like when i run python i get:
Python 3.6.2 |Anaconda, Inc.| (default, Oct 5 2017, 07:59:26) [GCC 7.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'tensorflow'
but with python3.6 everything works fine:
Python 3.6.10 | packaged by conda-forge | (default, Apr 24 2020, 16:44:11) [GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>>
I basically want to type the "python" into my command line and to work fine with tensorFlow.

Something is really strange with the way you are doing this, you shouldn't need a venv on top of your environment and you should install tensorflow with a conda rather than a pip install (reason being conda is an environment manager and solves for all packages to work together, whereas pip may install or uninstall things in a way that messes up the whole environment solution; tldr, more likely to work). Try making a new environment and see how this works out
conda create -n tfenv tensorflow=2
if you instead want Python 3.6 (other one will use latest python that works with tensorflow, should be 3.7)
conda create -n tfenv python=3.6 tensorflow=2
then, as usual activate the environment with
conda activate tfenv

No module named 'nltk'

I successfully installed nltk and works fine. I have to run a file where nltk was imported and tensorflow too ,hence, i have to activate tensorflow.
When I activate tensorflow the .py file i want to run gives an error. I have read some solution but they didn't help.
HP-250-G5-Notebook-PC:~$ python
Python 3.6.3 |Anaconda custom (64-bit)| (default, Oct 13 2017, 12:02:49)
[GCC 7.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import nltk
>>>
this works fine but this does not
(tensorflow)HP-250-G5-Notebook-PC:~/AIG2018/Chatbot$ python
Python 3.5.2 (default, Nov 23 2017, 16:37:01)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import nltk
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
What is that I am doing worng?
ImportError: No module named 'nltk'

You're using two different versions of python, and you probably installed nltk in your root environment, but not your virtual environment. When you "activate" the environment called tensorflow, you are using a different virtual environment, in which you haven't installed nltk. Try activating tensorflow, then using pip install nltk, then starting python. Because you seem to be using anaconda, this would probably look like this:
# Do these first 2 steps in your terminal:
source activate tensorflow
# you're now in the virtual environment called tensorflow
pip install nltk
# you now have nltk in that virtual environment
# Now, you can start python
python
Python 3.5.2 (default, Nov 23 2017, 16:37:01)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import nltk
And you shouldn't have that error

If you look closely at your messages you will see that the successful import of nltk is on Python 3.6.3 and the failed import is on Python 3.5.2.
This indicates that you have two Python installations of different versions, and nltk is installed in one but not in the other.

Installing Scipy in Virtualenv

I've tried many ways to install scipy but without successes. I use virtualenv on Linux Mate.
After the workon command I tried
pip install scipy
or
easy_install scipy
end every time is a "failed with exit status 1"
I tried also
apt-get install python-scipy
and I've no errors but if I run python
python
import scipy
I'll have "ImportError: No module named scipy"
What is wrong?

I'd suggest using Miniconda to install scipy.
wget https://repo.continuum.io/miniconda/Miniconda-latest-Linux-x86_64.sh
bash Miniconda-latest-Linux-x86_64.sh -b
Then, create a Conda environment with scipy installed:
conda create -n scipy scipy
Now you'll have access to the activate and deactivate scripts, used to open and close Conda environments.
$ source activate scipy
discarding /Users/username/miniconda/bin from PATH
prepending /Users/username/miniconda/envs/scipy/bin to PATH
(scipy)$ python
Python 2.7.11 |Continuum Analytics, Inc.| (default, Dec 6 2015, 18:57:58)
[GCC 4.2.1 (Apple Inc. build 5577)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://anaconda.org
>>> import scipy
>>> scipy.version.version
'0.16.0'
[Edit] You indicated you want to use the system installed scipy package, that you have successfully installed via apt, in a virtualenv. You can instruct virtualenv to create a new environment that includes all the system Python packages like this:
$ virtualenv --system-site-packages scipy_env
New python executable in scipy_env/bin/python
Installing setuptools, pip...done.
$ source scipy_env/bin/activate
(scipy_env)$ python
Python 2.7.10 (default, Jul 14 2015, 19:46:27)
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.39)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import scipy

Wrong Django version returned in virtualenv

I'm on a Mac running OS X !0.10 Yosemite. Default versions of Python & Django are 2.7 & 1.5. I'm want to set up a virtualenv that has Django 1.8 so I'm doing the following:
$ virtualenv --no-site-packages django18env
New python executable in django18env/bin/python2.7
Also creating executable in django18env/bin/python
Installing setuptools, pip...done.
$ source django18env/bin/activate
(django18env)$
Then I'm installing Django 1.8
(django18env)$ sudo pip install django==1.8
Password:
Downloading/unpacking django==1.8
Downloading Django-1.8-py2.py3-none-any.whl (6.2MB): 6.2MB downloaded
Installing collected packages: django
Successfully installed django
Cleaning up...
(django18env)$
Once that has run I have Django installed under django18env/lib/python2.7/site-packages/django
If I look at the __init__.py file in that directory it shows:
from django.utils.version import get_version
VERSION = (1, 8, 0, 'final', 0)
So it certainly looks like the right version is installed in the virtualenv directory. However, if I use django-admin --version I get:
(django18env)$ django-admin version
1.5.4
I've also tried starting python in the virtual env and getting the django version that way:
(django18env)$ python
Python 2.7.8 (default, Jul 29 2014, 21:50:48)
[GCC 4.2.1 Compatible Apple LLVM 5.1 (clang-503.0.40)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import django
>>> django.get_version()
'1.5.4'
>>>
Any ideas on why it still seems to be pointing to 1.5 when 1.8 is installed in the vertualenv?
I've read various other threads on here but can't get the version to point to 1.8
Any help much appreciated
Thanks

don't use sudo on virtualenv. the point of vitualenv is, to install software not system wide, but verily for that enviroment. but no matter inside a virtualenv or outside it, if you use sudo, it will install software to your system globally.
ziya#ziya:~/Desktop/coursera/python/lorem$ virtualenv ipsum
New python executable in ipsum/bin/python2.6
Also creating executable in ipsum/bin/python
Installing setuptools, pip...done.
#created a virtualenv
ziya#ziya:~/Desktop/coursera/python/lorem$ cd ipsum/
ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ . bin/activate
# will now install package with sudo
(ipsum)ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ sudo pip install sudokulib # i don't know what it is, just installing.
[sudo] password for ziya:
.....
Collecting sudokulib
/usr/local/lib/python2.7/dist-packages # attention to this path!
...
Successfully installed sudokulib-0.6a0
(ipsum)ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ python
Python 2.6.5 (r265:79063, Apr 16 2010, 13:57:41)
[GCC 4.4.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import sudokulib # import the newly installed module
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: No module named sudokulib
>>> exit()
(ipsum)ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ deactivate
#deactivating virtualenv and starting default python
ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ python
Python 2.7.3 (default, Dec 18 2014, 19:10:20)
[GCC 4.6.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import sudokulib
>>> sudokulib.__version__
'0.6a' #here it is!
>>> exit()
ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ . bin/activate
(ipsum)ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ pip install sudokulib #now installing the same module without sudo
Collecting sudokulib
Downloading sudokulib-0.6a.tar.gz
/home/ziya/Desktop/coursera/python/lorem/ipsum/lib/python2.6/site-packages
....
Successfully installed sudokulib-0.6a0
(ipsum)ziya#ziya:~/Desktop/coursera/python/lorem/ipsum$ python
Python 2.6.5 (r265:79063, Apr 16 2010, 13:57:41)
[GCC 4.4.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import sudokulib
>>> sudokulib.__version__
'0.6a' #seems ok now :)
>>>

I found the issue. As I say above, Django 1.8 was being installed in the virtualenv OK but Python wasn't using it. In the vitualenv I started Python and then:
>>>import django
>>>django.__file__
This showed that Django had been imported from:
/usr/local/Cellar/python/2.7.8/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages
And when I looked in my .bash_profile there was a line:
export PYTHONPATH=$PYTHONPATH:/usr/local/Cellar/python/2.7.8/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages
Not sure how/when this got in there but I commented it out, restarted the shell and I now get Django1.8 in my virtualenv and the older (default) version 1.5 outside the virtualenv.
Thanks for the help and suggestions

After miniconda 3.3.0 (anaconda) installation, I get no module named conda.cli error

I installed latest version of miniconda on a linux 64-bit filesystem that does not allow hard links (openAFS).
These are installation steps:
run installer until it fails on hard link errors
then modify install.py specified in $CONDA_INSTALL adding two line at the top of _link function
if linktype==HARD_LINK:
linktype=SOFT_LINK
now execute $PYTHON $CONDA_INSTALL --prefix=$PREFIX --pkgs-dir=$PREFIX/pkgs --link-all and everything works
The problem is that even if python is correctly installed,
> miniconda/bin/python
Python 2.7.6 |Continuum Analytics, Inc.| (default, Jan 17 2014, 10:13:17)
[GCC 4.1.2 20080704 (Red Hat 4.1.2-54)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>>
I cannot run conda:
> miniconda/bin/conda
Traceback (most recent call last):
File "miniconda/bin/conda", line 3, in <module>
from conda.cli import main
ImportError: No module named conda.cli
I tried to set $PATH and/or $PYTHONPATH and/or $LD_LIBRARY_PATH but without success.
Is there a way to solve this issue? Is there a way to force soft link in place of hard link during installation?
Thanks,
Angelo