Announcing Anaconda Distribution 2019.10

 

Anaconda Python scales

We are pleased to announce the release of Anaconda Distribution 2019.10!

As there were some significant changes in the previous Anaconda Distribution 2019.07 installers, this release focuses on polishing up rough edges in that release and bringing all the packages up to date with the latest available in repo.anaconda.com. This means many key packages are updated including Numpy, Scipy, Scikit-Learn, Matplotlib, Pandas, Jupyter Notebook, and many more. As many of the package updates have addressed Common Vulnerabilities and Exposures (CVEs), it is important to update to the latest. 

Another key change since the last release is that Apple released macOS version 10.15 – Catalina. Unfortunately, this was a breaking release for previous versions of Anaconda that used the pkg installer. The Anaconda Distribution 2019.10 installers address the issues and should install without trouble on macOS Catalina.  If you would rather repair your current Anaconda installation, please check out this blog post for tips.

In the previous Anaconda Distribution release, we added a new dependency, libarchive, to handle our new .conda file format. The Python wrapper we used, python-libarchive-c, used CFFI to load libarchive. Unfortunately, we had several users report issues with this library loading correctly. We have added a new Cython-based Python wrapper, and we have statically linked it against libarchive and the component compression libraries. This work is present in conda-package-handling 1.6.0. This should clear up the library loading problems.

The installer itself has undergone small changes to fix edge case installation scenarios, such as the tmp volume being mounted noexec or GNU Coreutils being old enough that the flags we were using to extract the payload shell scripts were not valid. We have reworked our extraction scripts to use older commands, so Anaconda should install more smoothly on older operating systems. This retains all previous improvements to the installer, including support for noarch packages and the new .conda package format.

There have also been some improvements in conda since the last Anaconda release. Conda has added support for recording environment files based on explicit specs in the history of an environment, rather than just all the packages that are present. There is now a configuration parameter to disable conflict finding, so you can save time and fail faster if you don’t want the conflict results anyway. There were also many bug fixes for functionality and performance. You can find the full release notes for conda here.

You can find the full release notes for Anaconda Distribution 2019.10 here.

Update to Anaconda 2019.10 now with conda install anaconda=2019.10, or download and install Anaconda 2019.10.


You May Also Like

For Practitioners
Anaconda Easy Button – Microsoft SQL Server and Python
Previously there were many twisty roads that you may have followed if you wanted to use Python on a client system to connect to a Microsoft SQL Server database, and not all of...
Read More
Enterprise Data Science
How PNC Financial Services Leveraged Anaconda to Enable Data Science and Machine Learning Capabilities Across the Company
As an AI software company passionate about the real-world practice of data science, machine learning, and predictive analytics, we take great pleasure in hearing about the ins...
Read More
For Practitioners
Using Anaconda and H2O to Supercharge your Machine Learning and Predictive Analytics
Anaconda integrates with many different providers and platforms to give you access to the data science libraries you love with the tools you use, including Amazon Web Services...
Read More