Battle Testing MaPS: Real world usage of the MaRDI Packaging System
The MaRDI Packaging System (MaPS) is a software tool created by MaRDI to address the problem of software preservation. This is achieved via predefined computer environments. Research software can be archived at the time it was written, together with the relevant academic paper. MaPS provides the confidence that the packaged software will still run at any later date, on computer systems with different configurations. MaPS achieves this by following a strategy similar to other sand boxing measures: bundle all application dependencies in a single package, called a ``runtime'', and deploy it in an environment isolated from any conflicting host libraries. A MaPS runtime is lighter than a full virtual machine, and less complicated to build than a flatpak, or a Docker image. It does not require sharing a disk image or hypervisor configuration, and does not require writing a build recipe in the form of flatpak manifests or dockerfiles. Building a runtime feels like logging into a clean Linux installation, manually setting up all software, and finally committing your changes. MaPS performs file level de-duplication across runtimes, to maximise disk usage efficiency. The only things required to deploy and run a MaPS runtime is MaPS itself, available via standard system package managers, and a unique ID pointing to the runtime. In a collaboration with LMFDB, Computation, and Number Theory (LuCANT) conference 2025, MaPS saw its first real world use. Out of 36 papers submitted to LuCANT, 23 were selected for software review. Out of these 23, 8 relied on software licensed as free and open source software (FOSS), 3 submissions did not include any code. 12 submissions included code in Magma, a proprietary Computer Algebra System, which we are not allowed to re distribute. We made runtimes for the 8 submissions relying on FOSS, and distributed them to the software reviewers for LuCANT. In this process, some final pieces of required functionality was added to MaPS. Of the 8 runtime packages created, 4 relied on Sagemath. Because the version of Sagemath required was not the same as available in the Debian software repositories, and compiling a lean version of Sagemath by hand is difficult, the Arch Linux archive was used to create a runtime based on Arch Linux, with versions pinned to the required versions of Sagemath. This was then successfully tested on a different computer running Linux Ubuntu. Since 1 runtime required the use of a web browser to display the computed output, the ability to run arbitrary graphical programs was added to MaPS. All in all, utilizing MaPS for the software reviewing for the LuCANT2025 conference was a success: neither did we encounter problems in creating the runtimes, nor did the reviewers report any problems using them. This shows that MaPS in its current state is user friendly and contains all essential functionality both for runtime creators, and runtime consumers. Thus, MaPS is ready for wider adoption.
Institutions
- University of Kaiserslautern (DE)
- University of Koblenz and Landau (DE)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2025-08-04
- DOI
- https://doi.org/10.5281/zenodo.16736153
- Primary Topic
- Trypanosoma species research and implications
- Type
- article
- Field-Weighted Citation Impact
- 0.00