Download websites that use EPrints

About EPrints

A platform first released in 2000 that still returns live fingerprints is describing its operators more than its technology. Software survives that long in one situation, which is when the institution running it treats the content as permanent and the interface as secondary.

EPrints came out of University of Southampton and is distributed under GPL v3, and the combination is unusual. The maintainer is a university rather than a company, so there is no sales pipeline, no upsell path and no commercial reason for anyone to keep an accurate deployment count. Adoption travelled through academic networks and funder mandates instead of procurement, which is why the installed base looks nothing like that of a commercial Miscellaneous.

As of October 2026 we detect the platform on 65 sites, and we can attribute 5 distinct releases across them, with 3.4.6 the one we see recorded more often than any other. A count of that size is small in web terms and large in institutional terms, because each install usually stands for an organisation, a deposit policy and a body of records rather than a single publisher's website. The value carried by each domain in the list is correspondingly higher than the total suggests.

The implementation language is part of the story. EPrints is written in Perl, and a working Perl application still in daily operation carries a specific maintenance profile. Fewer engineers want to touch it, local customisation accumulates in ways nobody documented, and the migration decision gets deferred every budget cycle because a repository cannot simply go offline while somebody rebuilds it. Presence on a domain implies stability and pressure at the same time, which is a combination worth knowing about if you sell replacements, hosting or long term preservation services.

The undercount deserves naming. Repositories usually sit on subdomains, sometimes behind institutional authentication, and are often linked only from library pages, so whether we see one depends on how the parent domain was crawled rather than on whether the software is running. Version detection depends on metadata an administrator can remove, so the release breakdown describes the installs that publish their version rather than all of them.

What the extract answers that a lookup cannot is a question about a sector rather than a site. Scholarly infrastructure vendors use it to size the population still running self hosted repository software before deciding whether that market is worth entering. Consultancies use it to find the estates approaching a migration they have already been putting off. Open access researchers use it to map where institutional repositories actually sit, which cannot be assembled one domain at a time when the domains are the thing being discovered.