Ubuntu 18.04 project image upgrade

The CoCalc service provides and extensive collection of software packages. The underlying "image" for each project is based on Linux (Ubuntu).

For the year before August 2018, we provided a release of Ubuntu from 2016 – 16.04, named "xenial" – which we enhanced and maintained via many small incremental updates. All that became quite "dusty" and we decided to switched to Ubuntu 18.04, nickname "bionic", which is a much more recent release. That gave us the opportunity to start over with a clean slate. This is a major update, so despite our best efforts to offer the same overall set of solutions for everyone, there are a couple of subtle differences.

Overall, you can expect these updates of our more frequently used software packages:

• Python 3 in Ubuntu is currently at version 3.6 (instead of 3.5).
• Sage: the default version is 8.3
• R: 3.4.4 and we'll soon upgrade to 3.5
• PostgresQL: version 10 is installed, and we'll switch to it being the default soon
• Anaconda Python environment: start it via anaconda5 or the kernel with version 5.

---

User Control over the project image

Most importantly, we now offer a way for you to dial back this upgrade via the project settings page of each project. At the bottom of the project control box there is a new line labeled "Software Environment"; you can select the "old" image to work in the previous environment. This setting is just for that particular project. Keep in mind that we are only continuing to upgrade the newer "default" image. You can also opt in to use our most recent "experimental" image, which provides our newest updates a bit earlier.

---

Known Issues

• The GAP Jupyter Kernel. There are robustness issues with the newer version of GAP and its native kernel. As an alternative, you can use it in the Terminal via jupyter console --kernel=gap, by using the classical jupyter notebook (in project settings), or just use it directly on the command line. There is also a "GAP mode" in Sage Worksheets via %gap.
• Axiom: there seems to be an overall packaging issue in Ubuntu/Debian
• pandas_datareader in Python 3: that's an upstream issue, where it's broken with a newer version of pandas

Discontinued

There are some software packages which we're no longer supporting any more. We'll try to include them again if there is significant demand or the technical hurdles preventing us from offering them are resolved.

• Scala Jupyter Kernel: we were not able to get it to run in the newer environment. If someone is able to help us or knows a viable solution, please drop us a note at help@cocalc.com.
• Anaconda 3: we'll no longer offer updates and it will be removed
• Giac: UPDATE 2018-09-01: available again via a wrapper script for the one shipped inside of sage
• RethinkDB: as a mitigation, anyone can upload the binary release.
• PdfTK: it is no longer maintained and relies on a deprecated runtime which got removed from Ubuntu 18.04. Tickets: #1757314 and #1764450
  • Update 2018-09-01 installed Stapler as a more modern approach for the same task

Upcoming

• Julia 0.7/1.0: we're aware of this major release. Once we know how to setup the global repository with a set of libraries, we'll roll it out as the default.
• R 3.5 with a larger set of available libraries

Testing

Like before, we're committed to offer a well working and complete solution for everyone. If you're teaching using CoCalc and want to make sure that some specific functionalities are available, we're happy to add some tests to make sure they work fine.