Anaconda 4.3.0 was released on January 31, 2017, so I installed it immediately.
It immediately supports Python 3.6, which was just released on December 23, 2016. Note that the root environment has been changed to Python 3.6.
Some packages have been removed as shown in the Changelog below. It's unclear which package applies, but some features are missing.
--Jupyter Notebook does not have Presentation mode. (Nbpresent?) --Cannot select kernel in Jupyter Notebook. (Nb_conda_kernels?) --When installing with conda, the list of downloaded packages is not displayed. conda no longer showing download summary · Issue #4348 · conda/conda --The installed channel is not output to the list of packages output by conda env export. (Maybe specification change)
As of 2/3, the above nbpresent and nb_conda_kernels cannot even be additionally installed with conda.
When reading a Shift-JIS CSV file that includes Japanese in the path with pandas.read_csv, an error occurs. I made a Python 3.4 environment and tried it, but no error occurred, so it's definitely not a mistake. It seems that the specification change of Python 3.6 is related. Is that a matter of compatibility between Python 3.6 and pandas rather than Anaconda?
PEP 528 -- Change Windows console encoding to UTF-8 | Python.org PEP 529 -- Change Windows filesystem encoding to UTF-8 | Python.org
[Anaconda changelog | Continuum Analytics: Documentation] (https://docs.continuum.io/anaconda/changelog) Highlights: The Anaconda3 installers are based on Python 3.6. Anaconda 4.3 supports Python 2.7, 3.4, 3.5 and 3.6. Anaconda 4.3 will be the last release which supports Python 3.4. We will discontinue regular Python 3.4 package updates in the next release. The Intel Math Kernel Library (MKL) is updated from 11.3.3 to 2017.0.1. Over 90 packages are updated. seaborn is now installed by default. Removed (from installer only):
Recommended Posts