Although researchers routinely deposit data in repositories, much research
software continues to live only on platforms like GitHub, GitLab,
BitBucket, and SourceForge. Our research has shown that the number of
GitHub URLS in preprints has risen exponentially in the past decade, while
fewer than half of researchers place their research software in an archive
that guarantees preservation. At the same time, even if software is
deposited into a repository, curators currently struggle to collect the
important documentation that often accompanies scholarly code. Such
supplementary information–including issue and pull request threads,
external documentation, and wikis–are key for future researchers who hope
to reuse or understand the software. We are creating an ecosystem to
preserve both software and its Web-based documentation as a part of
Collaborating
on Software Archiving For Institutions <https://cosai.gitlab.io/> (CoSAI,
funded by Sloan Foundation). We have linked together a suite of open source
tools to allow curators to package the scholarly code hosted on Git hosting
platforms alongside the software’s contextual documentation. We will
present this workflow, which includes the innovative targeted Web archiving
tool Memento Tracer, the software preservation system OCCAM, and an
Invenio-RDM repository. Our approach can operate at scale, incorporates
metadata creation, and allows federated sharing of workflows. Our interest
in this topic emerges from the desire to make reproducibility
longer-lasting. This project builds on the foundation of a previously
funded project, Investigating and Archiving the Scholarly Git Experience
<https://investigating-archiving-git.gitlab.io/>, which assessed the
usability of Git and Git hosting platforms are for researchers, and
conducted a scan of approaches to archiving not just code itself, but the
scholarly ephemera around it that contextualizes its development, history,
and usage. By providing a shareable workflow and ecosystem for preserving
Web-hosted code and its context, we hope to provide a pathway for
preserving these necessary pieces of the reproducibility bundle.
