[Tutor] Installing python

[Mats Wichmann]

On 11/2/18 10:01 PM, Roger B. Atkins wrote:
> Okay, but what about the virtual environments? Once the Windows 10
> path environment variable has a version of Python added to it, Windows
> will find it (I am using Win 10), but will it be found from within any
> given virtual environment (venv), regardless of which version created
> the environment? And if 3rd party packages have been installed within
> a venv, will they be found a) by Windows, and b) by the Python version
> within the venv, even if it was not used to install the packages?

(a) when you activate a virtualenv, it will always find that python (the
one the env was created from), that's what the activation does - fiddle
paths so you see that envirnoment and not any other that may be sitting
around.

(b) when you pip install with a virtualenv active the installation will
be private to that virtualenv.

which leads to... Anaconda uses virtualenvs (afiak, _always_, rather
than "when you ask for one" like many IDEs do), so, packages installed
though Anaconda are in that sense "private".  They're not secret - you
can see them all in the physical location where the virtualenv is set
up, if you figure out where that is.

> These questions confounded me until I finally deleted Anaconda, and
> Python with it, because when running programs in Windows vice the
> Spyder IDE, Windows could find Python, but not the 3rd party packages
> such as pyperclip, bs4, and requests. I suspect it had something to do
> with the Anaconda installation process establishing a venv, and the
> package locations not existing (or not being visible) in Windows.
I don't think developers of projects like Anaconda expect that their
target users will try to run Python lots of different ways, so getting
things all under their control is helpful and presents a consistent
front, which is really the value the project brings - you want to do
some (for example) scientific work in Python, and some notes tell you
you need to install these 17 packages, and if you're on Windows, a
couple of them maybe don't work because you don't have a C compiler
installed - Anaconda puts that all in once place and makes it easier.
Doesn't mean it isn't still occasionally confusing. If you need to run
more versions, you can always install pkgs a second time, "outside" the
Anaconda world. It should even be quick to do, pip caches packages it
has downloaded so it probably won't even need to re-download.


p.s. Python can tell you which Python it is running:

  $ python
  Python 2.7.15 (default, Oct 15 2018, 15:24:06)
  [GCC 8.1.1 20180712 (Red Hat 8.1.1-5)] on linux2
  Type "help", "copyright", "credits" or "license" for more information.
  >>> import sys
  >>> sys.executable
  '/usr/bin/python'
  >>>
  $ source ~/virtualenv/py2/bin/activate
  $ python
  Python 2.7.15 (default, Oct 15 2018, 15:24:06)
  [GCC 8.1.1 20180712 (Red Hat 8.1.1-5)] on linux2
  Type "help", "copyright", "credits" or "license" for more information.
  >>> import sys
  >>> sys.executable
  '/home/mats/virtualenv/py2/bin/python'
  >>>