I'm pondering switching over to Anaconda from my vanilla Python in OSX. I know Anaconda brings its own NumPy. I was wondering if it was possible to make the GitHub version of NumPy the default version, or if Anaconda only works with its own version.
Anaconda installs python just like it already is on your system, only to a different location. It allows you to choose what packages you want to install. If you want to replace one you can go into the site-packages folder (Anaconda/lib/site-packages) and do so. In my experience, Anaconda was well worth the switch.
Related
I have python 3.8 installed in my windows 10 machine as system python.
I want to install anaconda. What precautions/steps should I take so that the system python does not interfere with the anaconda installation(which I am doing at a user level) while executing binaries or installing packages?
There are no specific precautions to take: a standard install of Anaconda won't replace the default Python installation of your system, unless you ask for it explicitly during setup.
Afterwards, you will be able to create independent conda environments, each of them having their own Python version and set of packages.
As #Woodford pointed to in eir comment, you will learn a lot by looking at the docs.
As I already have quite a lot of packages installed without installing Anaconda will I have to reinstall them again separately? Or are they accessible in Anaconda environment because I have them preinstalled? There is a similar question which has the answer suggesting installing only miniconda but in my case Anaconda is necessary.
No, anaconda uses completely different environments and whatever downgrading it does, it does it within it's own environments. So, if you are using anaconda environments within your applications, you don't need to reinstall anything but if you are using another environment (like pip only) you need to make sure you have installed all the necessary packages there as well.
Also, if a package is installed using pip and not installed in conda, it will automatically switch to pip so there is no problem there.
I just got a new computer, and I was installing some Python libraries. When I tried to install numpy, I got a message on the console saying numpy was already downloaded. I went into the library folder, and not only was numpy there, but scipy, matplotlib, and a bunch of other libraries as well. How is this possible, considering this computer is brand new? I had installed Python the previous evening, so does installing Python automatically install these libraries as well?
It depends on which Python package you install.
On any desktop oriented Linux distribution, installing the Python package will only install Python. And in practice, it is generally already installed.
On Windows (and perhaps OS X), you need to pick a Python bundle somewhere. Perhaps did you choose Anaconda, which AFAIK installs a lot of common Python libraries (as opposed to its light version Miniconda).
Python does not ship with these libraries unless you are using a pre-packaged distribution such as Anaconda.
If you copied your data from your previous computer to this one, you may have copied the python installation (and thereby the libraries you had installed before) in your appdata folder.
Another possibility is that you have install Anaconda, which is targeted especially at scientific things, and comes with numpy, scipy and some other things preinstalled.
Although this is not the place for these types of questions, yes, there is no need to install libraries, as most of the times when you download Python in a distribution, such as Anaconda, they are also included.
Is there any benefit of using Anaconda for PyCharm instead of the standard python distribution for PyCharm?
Using Anaconda instead of standard Python will benefit you as Anaconda comes with preinstalled packages. A lot of time will be saved there since installing various packages and making them run gets irritating sometimes.
Also, since it will have preinstalled packages, it can be heavy for your system. You can try miniconda as an alternative where you install packages when required. It is still better than having Python only. You can even install Anaconda by using conda install anaconda
Pip installation of Python packages sometimes may cause few problems to the user and you need to constantly update the pip as well before installing any other python packages via pip. So using Anaconda will be a benefit in this case.
I am using openSUSE Leap 42.1 and do some data analysis work in python. Most of the python packages I use are available in the standard openSUSE repositories (e.g. obs://build.opensuse.org/devel:languages:python); however sometimes they aren't, whereas they are available in Anaconda. I would like to replace all of the python packages installed on my computer with those available through Anaconda.
Is it possible to just install Anaconda in parallel with the normal openSUSE packages or should I manually delete the packages I've installed? I know python is used heavily throughout the operating system so I probably don't want to deep clean the system of python before going the Anaconda route.
Has anyone done this before? I was unable to find any info on this on the Anaconda site, and I'm curious if there is a clean way to do this.
I read the anaconda documentation, and there is no evidence of anaconda packages replacing your openSUSE packages. There isn't a reason for it to do so. If I got it right, then Conda is very similar to ruby's gem and similar tools, which definitely don't replace the installed packages. I think you can feel free to install it next to your current packages. Also, you can specify the python and python package version in the anaconda envinroments, which is another thing which it allows you to do, so you can decide what you will use there. Note, I'm not a conda user, this is how I understood the docs. Hope this helps.