On the project's website it tells me, that there is Python 3.4 support (Readme.md, which is shown as text on that page). However, I fail to see where I could get that Python 3.4 compatible version of SFrame from. It's not on anaconda and not on PyPI, where I only see Python 2.7 versions.
So I wonder, if they simply didn't update their Readme.md to note the fact, that SFrame is not available for Python 3.4 anymore, or if it is somewhere, where I didn't find it. Why should they note that there is no support for Python 3.5.x yet after all, if they didn't have it for 3.4.
I did install it for a Python 3.4.5 anaconda environment before and if I recall correctly, that was from PyPI, so I am thinking they took it down from there for some reason.
Does anyone know how to get a Python 3.4 compatible version of SFrame into an anaconda environment today?
Related
I have the 3.5 version of python. I want to install Anaconda, but it says on the Anaconda website the latest version of it is for Python 3.6. My question is could I still use the packages for Python 3.5, or should I install Python 3.6?
When you install Anaconda Python, it installs into its own area and wouldn't conflict with an existing Python installation. If you already have additional Python packages installed, you will need to reinstall them for the new Python installation, preferably using a Python virtual environment. You can't use a Python virtual environment from an existing Python installation and would need to create a new one against Anaconda Python if already using one.
If your own personal code works with Python 3.5, it likely will work with Python 3.6 no problems.
So with the above caveats on re-installing additional Python packages, there shouldn't be any reason why you couldn't use Anaconda Python 3.6.
I had the same problem where I couldn't run anaconda with python 3.6. What I did was to install anaconda with python 3.5. Go to https://repo.continuum.io/archive/ then download Anaconda3-4.2.0-Windows-x86_64.
I have installed python 3.5 and NLTK. Also, I have installed the nltk_contrib library from https://github.com/nltk/nltk_contrib which is not supporting to the latest version of python. The error is that the library written for older version. So, the library has print without parenthesis which will not support to the latest version of Python.
Is there any latest version nltk_contrib without down grade the python?
Thanks in advance.
nltk contrib has not been ported to python 3 yet.
See this answer for more details:
How to install nltk_contrib in Anaconda
You need to figure out which part of the code you need for python 3 and either port it yourself.
Or set up a virtualenv or and env for anaconda with python 2.7 and use it there.
http://conda.pydata.org/docs/using/envs.html
my school told me to use python 3.2.5 for compatibility reasons as that's the version that they use, but I've found that a few modules I would like, don't work with python 3.2
I was wondering if there's a way to have two versions of python installed or if any later versions are compatible with python 3.2.5?
You can certainly have multiple versions of python installed. In my case I have coworkers that prefer Python 2.7 and so I have a virtual environment set up that will use python 2.7 instead of 3.4 (which i prefer). There are lots of useful cases when environments will save you headaches so do learn about them earlier than later.
Personally, I like to use the Anaconda distribution of python. With anaconda you can use conda create --name env_name python=3.x package package1 this will create a python environment with python version 3.x named env_name and install packages: package and package1. If you want an empty environment simply omit packages. The documentation for this can be found here.
However, as Antti Haapala points out 3.2 is not necessarily the most stable or compatible of versions. In fact conda has dropped 3.2 for environment creation, running conda create --name env_name python=3.2 returns an error.
I'm sure your school has some reason they want to stick to 3.2, but in my opinion you might be best served by using 3.3 (or later) instead.
ps. referring to your previous question about pip; with anaconda there is a command conda which accomplishes the same thing. For example you can run conda install pandas to install pandas. If a package can't be found, searching the anaconda site will give you info on how to add packages from other owners.
I basically use python 3.4 and the numpy 1.3.0 had been installed and launched successfully on my windows machine.
Recently, I've just installed python 2.7, in order to use Brian2 library, but I've an error in traceback as Importing numpy failed: No module named numpy, whereas it has been installed and works well under aegis of python 3.4
Actually, when I switch the interpreter address to python 2.7, the problem will be happened, in which the numpy is totally hidden in view of the python 2.7...
Any help?!
Edit:
I even tried to choose python 2.7 (As it was detected automatically in registry), within the re-installation of numpy... but still there is nothing as the installed package in Python Interpreters -> Python 2.7
Python x.y (on Windows) generally expects to find 3rd party modules and packages in .../pythonxy/Lib/site-packages. If the module or package is pure python code that is compatible with multiple python versions (such as code that works with 2.6 and 2.7, or 3.3 and 3.4, or much more rarely 2.7 and 3.4), then it is possible to access one copy of the package from multiple site-packages directories. (This is done with x.pth files, which is another topic not relevant here.)
However, on Windows, anything compiled from C must be compiled for Windows and a particular version of Python, including 32 versus 64 bits. (This is currently a requirement of Microsoft C, which may change in the future.) To run with 3.4.x, you installed numpy-1.3.0-for-3.4-winxx. To run with 2.7, you must install numpy compiled for 2.7.
By the way, current numpy is 1.8, with 1.9 due soon. You might want to take the opportunity to upgrade. If you have trouble getting Windows binaries for python extensions, gohlke's site is excellent.
I am having a hard time trying to install the Shogun machine-learning toolbox via MacPorts. It seems that the only version MacPorts has declares Python 2.6 as a dependency and as a result MacPorts downloads and installs Python 2.6, numPy for Python 2.6 etc. The problem is I have Python 2.7 already installed and a host of other packages, including numPy, that I want to use with shogun already installed under Python 2.7.
Has anyone had success using shogun with Python 2.7. I couldn't find any documentation stating it only works with 2.6, so I assume 2.7 should work.
It looks like it needs to be converted to a python Portgroup port where the python version is implied from the current python see py-scipy for an example depending on py-numpy.
Also see the current threads on the macports mailing list for octave.