|
 |
Paper: |
Gemini Data Reductions – Saving Legacy Software With The Cloud |
Volume: |
541, ADASS XXXIII |
Page: |
349 |
Authors: |
Wesley C. Fraser; JJ Kavelaars |
DOI: |
10.26624/PDXV8938 |
Abstract: |
The deprecation of software is a problem all too common in astronomical
research. Given the often slower pace of scientific research software evolution compared to that of the global software and hardware communities and the difficulty in securing funding for small software development projects, it is inevitable that every scientist will eventually be faced with having to install legacy software on modern platforms.
For example, the successful build of IRAF from scratch on a modern PC can be difficult on some platforms and impossible on others. Some technological developments,
however, can greatly alleviate this problem. With the advent of software containers,
many legacy software packages can be made to work without too much hassle. Moving
to containers can even enable new computing capabilities not previously available with
legacy systems. The Gemini Observatory is currently undergoing the herculean effort
of converting the most used IRAF reduction routines into Python stacks, though support
for some legacy instruments is not planned. In the interim, the popularity of the macOS
operating system and the move to Apple Silicon means that many Gemini users cannot
use the Gemini-provided IRAF-based tools (which themselves rely on modules in the
STSDAS Table package) to reduce their data. Here, we use the Gemini software stacks
as a good example to demonstrate the utility of the Docker container. After creating a
Dockerfile that installs those stacks inside a container, the container can be run on multiple platforms and launched by other users with a simple Docker command. Sharing
containers provides those compute environments locally, even on systems that would
otherwise not support said software, and with an installation that is straightforward and
follows well-known paradigms. Beyond the ability to launch otherwise incompatible
software, the resulting container has the notable advantage of being usable in the cloud.
The container we built for the Gemini stack is now available via the Canadian Advanced
Network for Astronomical Research (CANFAR) Science Portal, accessible to the entire
Canadian astronomical community. With a small amount of education, the burden of
maintenance does not solely fall on a small, underfunded development team, but can
largely be shouldered by the more abundant power users, like it is on CANFAR. |
|
 |
|
|