We Keep Coding

How to downgrade ubuntu to python 3.6?

I'm trying to install an older version of tensorflow and it needs python3.6 to support the whl file of installation.
I'm now running Ubuntu 20.04 with python 3.8.5, I've already done this :
sudo add-apt-repository ppa:deadsnakes/ppa
followed by :
sudo apt-get update
sudo apt-get install python3.6
By doing that it installs python 3.6 but when I see the version of python installed it's still 3.8.5. Should I do something to remove python3.8.5 ? maybe apt-get purge ?
P.S: I'm installing the wheel file through pip3 should I downgrade it too ?

You shouldn't hack the system installation of Python I'd say.
Instead you should use something that let's you manage multiple versions of it, something like pyenv.
It is a well-known and widely accepted utility and according to its readme:
pyenv lets you easily switch between multiple versions of Python. It's simple, unobtrusive, and follows the UNIX tradition of single-purpose tools that do one thing well.
You can find the installation instructions here: https://github.com/pyenv/pyenv#basic-github-checkout.
Once you installed it you can install your preferred Python distribution which will live besides your system one and all the others you want later.
E.g. if you want to install 3.6.7 and then use it globally you can do:
pyenv install 3.6.7
pyenv global 3.6.7
Then test it like:
python -V
And that's only to scratch the surface, you can do many more things with it, check out the documentation for more.

after the commands you've run, you should have a python3.6 binary installed on the path
that said, I'd recommend using virtualenvs instead of system installations
for deadsnakes, you can install the venv module by (sudo) apt install python3.6-venv (debian decided to split venv into a separate module, so the deadsnakes packaging follows that)
from there you can create and activate a virtualenv:
python3.6 -m venv venv # create the environment
. venv/bin/activate # activate the environment
pip install ... # install things to your isolated environment
inside this virtualenv the python command will refer to your own isolated python installation
disclaimer: I'm the maintainer of deadsnakes

Should I use pip or pip3? [duplicate]

This question already has answers here:
pip or pip3 to install packages for Python 3?
(10 answers)
Closed 2 years ago.
Eventually, every single time I install a new Linux distribution I do sudo apt-get install python3.
However, once installed I always get confused. python is Python 2.7 and python3 is Python 3.x. But also it appears that pip is for Python 2 and pip3 for Python 3. That said most tutorials I see on Internet always use the traditional pip install even though it is about Python 3.
How should I deal with this? Should I simply continue to put this annoying 3 every time I use Python (pip3, ipython3, python3...)? In most of my lectures I read that creating a symlink python->python3 is a bad practice. Is that correct?

Use python3 -m pip or python -m pip. That will use the correct pip for the python version you want. This method is mentioned in the pip documentation:
python -m pip executes pip using the Python interpreter you specified as python. So /usr/bin/python3.7 -m pip means you are executing pip for your interpreter located at /usr/bin/python3.7.
Symlinking python->python3 is a bad idea because some programs might rely on python being python 2. Though, I have seen some Dockerfiles symlink python->python3, like TensorFlow's CPU dockerfile (it's less of an issue in a Docker image). Coincidentally, that same Dockerfile uses the python3 -m pip install syntax that I recommend.

creating a symlink python->python3 is a bad practice. Is that correct?
Sometimes. Some OSs (looking at you, macOS) deeply rely on python pointing to a Python 2 interpreter for internal tools and tasks. Deleting the shipped Python 2 interpreter (or aliasing python to a Python 3 interpreter) will break stuff. How to uninstall Python 2.7 on a Mac OS X 10.6.4?

Whether the correct command for Python 3 is pip or pip3 or (say) gaschplutzga depends on a number of factors.
If you only have Python 3, and you have a command named pip, that's probably safe to use. Going forward, this will be the simple, obvious, safe answer in more and more places.
If you have both, and there is a command called pip3 installed on your system, probably that's the correct one to use.
More generally, you can go through your PATH and look for commands with suitable names. On Unix-like systems with a POSIX-compatible shell, try the commands command -V pip3 and command -V pip. (On Windows systems, maybe try where pip3 and where pip, or pray to whatever dark deity informed your choice of operating system.)
If you receive output like
/opt/random/nonstandard/whoa/pip
/usr/local/bin/pip
/usr/bin/pip
you can try each of these in turn with the full path and adding the --version option to have them identify themselves. When you specify the full path, you are bypassing the system's PATH mechanism entirely. For example,
/opt/random/nonstandard/whoa/pip --version
might identify itself as belonging to Python version 3.2.1. If that's the one you want, and it's at the top of your PATH, you can simply rely on the PATH to give you this version when you type just pip. If not, perhaps you can shuffle your PATH (but understand that this changes the resolution order for all commands in the directory whose position you change) or create a simple alias or wrapper which bypasses the PATH for this particular command in your personal account. On Unix-like systems with a POSIX-compatible shell, this might look like
alias pip=/opt/random/nonstandard/whoa/pip
(to persist this across sessions, you'd add this to your .profile or similar - for Bash, try .bash_profile if it exists; for Zsh, try .zshrc. The full scoop for each shell is more complicated than I can squeeze into these narrow parentheses); on Windows, you might be able to control this by setting the environment variable PY_PYTHON, but there's a huge can of worms behind that "might".
Some sites and OSes / distros have additional wrappers or conventions which introduce additional options; if you use a specific package manager, perhaps also study its documentation. (One common example is Anaconda, though I don't believe it affects the naming or location of pip specifically.)

Use virtual environments, then pip would be associated with the python used to create that virtual environment. Whether you use pip or pip3, it will be equivalent to python3 -m pip as mentioned in jakub's answer. Also, given that Python 2.7 is already EOL (which means you will most likely work with Python 3) and that pip install-ing things onto the system packages should be avoided, then a virtual environment would be helpful here.
For example, using pipenv:
$ pipenv --python=/usr/local/opt/python#3.8/bin/python3
$ pipenv shell
Launching subshell in virtual environment...
(TEMP) $ pip --version
pip 20.2.3 from /Users/me/.venvs/temp2-SbXvZiFd/lib/python3.8/site-packages/pip (python 3.8)
(TEMP) $ pip3 --version
pip 20.2.3 from /Users/me/.venvs/temp2-SbXvZiFd/lib/python3.8/site-packages/pip (python 3.8)
For example, using venv:
$ python3.8 -m venv .venv
$ source .venv/bin/activate
(.venv) $ pip --version
pip 20.2.3 from /Users/me/temp2/.venv/lib/python3.8/site-packages/pip (python 3.8)
(.venv) $ pip3 --version
pip 20.2.3 from /Users/me/temp2/.venv/lib/python3.8/site-packages/pip (python 3.8)
The virtual environment takes care of making sure pip or pip3 in this env refers to the pip from the correct Python version. You can then happily follow tutorials that still use pip install something (unless of course that tutorial refers to a Python 2.7 or a system-wide installation).

You can install pip through pip3 and this should resolve this issue.
$ pip --version
pip 19.0.3 from /usr/local/lib/python2.7/dist-packages/pip (python 2.7)
Notice that pip here is of Python 2.7 (in this example).
You can then force pip3 of Python 3.X to install pip under itself.
$ sudo pip3 install pip --upgrade
Installing collected packages: pip
Found existing installation: pip 8.1.1
Not uninstalling pip at /usr/lib/python3/dist-packages, outside environment /usr
Successfully installed pip-19.0.3
Once you check this again, it should reference Python 3.X so you don't have to deal with
what is what.
$ pip --version
pip 19.0.3 from /usr/local/lib/python3.5/dist-packages/pip (python 3.5)
I doubt you'll want to use Python 2 after this, but if you do happen to work with Python 2 code, you can create a virtual environment to access those commands again. Otherwise, you won't have to worry about the pip or pip3 distinction after this.
Not really a duplicate of this question, but this helped me suggest this answer: Can pip (python2) and pip3 (python3) coexist?

Pip is for python version less than 3. and pip3 is used when you want to install packages for python version 3 or higher.

Upgraded Python; Do I have to reinstall all site-packages manually?

I have upraded my Linux distro recently. Python 3.5 was replaced by Python 3.6.
All site packages I have installed with pip3 are still in the /usr/lib/python3.5/site-packages directory and Python does not find them there now, because it looks in .../python3.6/site-packages obviously.
I see the directory contents and I could manually install them again, but that does not look to me like the right way to do it. I could move the contents to the new directory, but again, this seems to me incorrect either.
How am I supposed to handle it properly?
Should I have prepared a pip3 freeze list before the upgrade?
I tried to search, but the keywords are probably too general and got many unrelated answers.

Python 3.5 was replaced by Python 3.6. But you still have the backup option of using python 3.5.
If you want to use python 3.6 you will have to reinstall all pip packages again for python 3.6. And it makes sense.
Say you were changing from 2.7 to 3.5. You would want to preserve both the environments separately. Hence 3.6 environment is different from 3.5.
A quick way to do this would be to pip freeze for 3.5 and then install those dependencies for 3.6.
pip freeze > reqs.txt
upgrade
pip install -r reqs.txt
Since you don't have this option anymore, first try and list all packages in your python3.5
for that you can install pip3.5 as answered by #kabanus.
sudo apt-get install python3=3.5.1*
sudo python3.5 easy_install.py pip
Also it's advised to use virtual environment per project so you can maintain separate environments for each of them.

I just hit the same problem upgrading from Python 3.6 to Python 3.7, I forgot to run pip freeze before I upgraded to Python 3.7. The solution that worked is to specify the --path option as the old site-packages/ directory (which was not deleted):
pip3 freeze --path /usr/local/lib/python3.6/site-packages/ > python3.6_requirements.txt
pip3 install -r python3.6_requirements.txt

This would have made things simpler for you to reinstall. Checkout the description. Using freeze you could have done something like:
$ env1/bin/pip3 freeze > requirements.txt
$ env2/bin/pip3 install -r requirements.txt
Generally the recommended method is you use a virtualenv for site packages, so you don't litter your installation areas, but TBH it never broke something for me. Another option is to check if the linux distribution has the package available for proper retrieval, as in:
sudo apt-get install python3-<somemodule>
This is what I prefer - and could have been upgraded with the distro. As for what to do now, If you really don't want to re-install everything properly you could try to cp /usr/lib/python3.5/site-packages/* /usr/lib/python3.6/site-packages. The differences between versions are not so great such that I believe most packages would work off the bat. You may have to sed to replace python3.5 with python3.6 in all files there though. Forgot delete all pyc files if you do this.
Python modules are self contained enough that if something is broken it can be handled per package, and the site packages are self contained completely, so you could always just remove everything and re-install.
A final note - you can try and install pyton3.5/pip3.5 for your linux, and then do the freeze thing. If there is no package you could install manually (whl or such) or compile a stand alone and configure the site path properly. If you want to keep things on a global site package directory or migrate to virtualenv this may be the safest option.

Problems with installing matplotlib in python 3.6

I'm trying to teach myself python, and I feel out of my depth. To start, I am working on a mac which already comes with python 2.7 installed.
I installed python 3.6 recently and have been using it to teach myself the basics. I'd like to eventually learn how to produce mathematical plots in python, and I know I will need the matplotlib package to do that.
Following some advice online, I was told that python3 already comes with pip installed, which is what I thought I should use to install matplotlib. The advice said I should type the following into the mac terminal:
python3.6 -m pip install matplotlib
I typed this, and it seemed like the package was installing, but I ended up getting some sort of error code that said:
Command "python setup.py egg_info" failed with error code 1 in [folder].
I tried opening IDLE and typing "import matplotlib", but I got the error: "no module named matplotlib". I also tried typing "import matplotlib.pyplot as plt", but I got the same error.
Based on further research and this youtube video, I've decided to just install miniconda in order to have access to the matplotlib package.
The problem is, I'm not sure if I should somehow be uninstalling whatever was installed when I ran the code above to install matplotlib. I've actually run that line of code 3 or 4 times. Should I remove anything before installing miniconda? Also, I am running python 3.6, while miniconda is listed on the website as being for python 3.5. Does this mean it won't work for my version of python?

Running pip like that would install packages system-wide. I'm guessing it's failing because you're not running as root (i.e. the administrator user). But wait! Don't try again as root! Instead of installing packages, do it in a virtual environment. First create it:
virtualenv myenv
This creates a directory called myenv with a bunch of stuff in it (so make note of where you run this command). Whenever you want to use the virtual environment (like straight away!) you first need to activate it:
. myenv/bin/activate
Don't miss out that dot (followed by a space) at the beginning! As the other answer says, the first thing you should do in it is upgrade pip:
pip install --upgrade pip
Now you're ready install whatever else you like:
pip install matplotlib
One last note: The virtual environment is tied to a particular Python version. By default it uses the system's Python 2.7 installation, so to use a different one you need to specify it when you create the virtual environment, like this (if that Python version is installed system-wide):
virtualenv -p python3.5 myenv
Or like this (if that Python version is not installed system-wide):
virtualenv -p /path/to/my/installation/of/python3.5 myenv
While the virtual environment is activated, you don't need to specify the particular path/version of Python. Just run it like this:
python

I also encountered many problems during my installation.
It seems that version 2 of matplotlib is not compatible with Python version 3.
Finally, I succeeded by specifying version 3 of matplotlib as follows with the following command:
sudo apt-get install python3-matplotlib
Reference from the Matplotlib website:
https://matplotlib.org/users/installing.html#building-on-linux

Try upgrade setup tools
--upgrade setuptools
or
easy_install -U setuptools
or upgrade pip
pip install --upgrade pip

I ended up downloading anaconda and using the python interpreter that comes with it, as anaconda comes with matplotlib and many other python packages of interest.

the pip command typically is for the Python 2. use pip3 instead to install the libraries in the python 3.X path
This should work
pip3 install matplotlib

The solution that work for me in python 3.6 is the following
py -m pip install matplotlib

Matplotlib files are downloaded in ~/.local/lib/python3.6/site-packages/ and not in /usr/lib/python3.6/ .
Try the command:
sudo cp -r ~/.local/lib/python3.6/site-packages/* /usr/lib/python3.6/

How do we separately use, maintain & install libraries for python 2.7 and python 3.5 on the same Ubuntu OS?

I need to work with Both Python 2.7.12 and python 3.5.2 simultaneously on my Ubuntu 16.04.1 LTS. Python 3 came pre-installed so I've no idea where it sits, in terms of path to the directory, while python 2 sits in /usr/local/lib/python2.7/.
I found lots of questions on SO and on askubuntu about how to install but nothing about how to use them separately, installing different libraries, and what should I avoid or be careful of, if I maintain this dual python thing for the long term? For example, I usually run pip install to install a library and I can check that its installed in my python2 directory but how do I install the same package for my python3 without conflicts? Something like: python3 pip install <package> ?? Where is the default python3 installed? And how do I call python3 for paths where python is not part of the command for example: pip freeze, sudo-apt get, etc.?
PS: I've not officially worked with Virtualenv but I've been informed that is usually good for isolating projects within a python language version, rather than isolating two different language versions from each other.
Please let me know.
Thanks

This is absolutely no problem, as Python does that for you. You don't need a virtualenv at all.
If you use Ubuntu packages, make sure you use the python3- versions for Python 3, and the normal python- versions for Python 2.
For example, python3-numpy and python-numpy.
If you use pip to install extra packages, you an either use the pip script with the version number appended: pip2.7 or pip3.5, or, my preferred method, call pip as a module for the respective Python executable:
python2.7 -m pip install <whatever>
and
python3.5 -m pip install <whatever>
Other than that, there should not be any issue: Python stores the packages in completely separate directories, and each Python executable only uses its respective directive.
Do not fiddle around with PYTHONPATH, unless you really know what you're doing. This has the danger of setting your PYTHONPATH to a directory with Python 2.7 modules and then using Python 3.5 to run things.
If you start from scratch, you may need to install pip first.
For the system Python(s), use the relevant package:
sudo apt install python-pip
sudo apt install python3-pip
For your locally installed Python(s), use the built-in bootstrapper module:
pythonx.y -m ensurepip
Note on the OS-installed Python executables:
Python 3.5 lives at /usr/bin/python3.5, Python 2.7 (the OS one) at /usr/bin/python2.7.
You could even use the OS 2.7 one next to your locally installed /usr/local/bin/python2.7 (and confuse yourself when a package can't be found because you used the wrong one).
Or install Python 3.6 next to Python 3.5 (provided you've used make altinstall, so python3 doesn't get overwritten).
This is also why you don't really want to run pip (or even pip2.7) as is: pip2.7 may get you the system one, instead of the one in /usr/local/bin/pip2.7, depending on your PATH.
(The same goes for the python2.7 executable, so if you need to specify the full path /usr/local/bin/python2.7 to run that one (or have an alias), the same holds for pip2.7. If, on the other hand, /usr/local/bin is first on your PATH, you should in principle never run into the same pip and python executables.)