Skip to main content
v2026.11,986 entries · CC-BY 4.0

Fedora Repository: Open-Source Digital Object Repository Software Explained

Fedora Repository is open-source digital object software for libraries and archives, not Fedora Linux: governance, REST API and limits.

Written and maintained by CASRAI Editorial Board

Published

On this page

Fedora Repository is open-source software for managing and preserving digital content. It stores objects behind an HTTP interface defined by the Fedora API Specification, and it is aimed at digital libraries and archives that need long-term access and preservation. It is not the package repositories of the Fedora Linux operating system, which is a different project.

This guide explains what the Fedora repository system is, who governs it, how a team gets a local instance running, what the API specification adds beyond ordinary web storage, and where the software’s limits lie. The data table at the end of the guide maps each specification feature to what it does for a repository manager.

What Fedora Repository is

The Fedora project’s code repository describes the software as a “robust, modular, open source repository system” for the management and dissemination of digital content, especially suited to digital libraries and archives. Invest in Open Infrastructure’s Infra Finder profile uses the same wording and adds that the system is used for access and preservation. The code is licensed under Apache License 2.0 and the documentation under CC BY 4.0. The server is written in Java.

Infra Finder gives 2001 as the launch year and traces the architecture to a 1997 concept at Cornell University. Fedora joined DuraSpace in 2009, and DuraSpace merged with Lyrasis in 2019; Lyrasis is now the organizational home.

Fedora Repository is not Fedora Linux

The name is shared by two unrelated efforts. The Fedora Linux distribution has its own package repositories, which users add to a package manager. The repository system in this guide is a digital object store: libraries and archives use it to hold content and its metadata. If you arrived here looking for software packages for Linux, the Fedora Project’s own documentation covers that topic; nothing in this guide applies to it.

Who runs it and what it covers

Infra Finder describes Fedora as community-owned, supported and governed. A Fedora Governance Group leads strategy, bylaws are in place, and member institutions can take part in governance depending on their commitment level. The organization is listed as a non-profit (501(c)(3)) with between one and five full-time staff, and the profile says it is “funded entirely by membership contributions from institutions.”

On scope, the profile lists these standards as the ones Fedora follows: Activity Streams 2.0, JSON-LD, Linked Data design principles, the Memento framework, RDF and REST, with OCFL named for preservation. It describes the REST API as open, with public core code and documentation. Hosting is available through third-party vendors only, according to the same profile. The profile gives no installation or user counts, so this guide does not state any.

For engagement, the profile reports two new members added in 2024 to 2025, a second annual Virtual Fedora Showcase with more than 200 registered attendees, weekly open technical meetings, a community Slack, a monthly newsletter and a code of conduct. It also reports that Fedora 6.5.1 was released in January 2025. The profile record is dated 26 November 2025.

How to get started with the repository software

The project’s README on GitHub gives the build route for a local test instance. It is aimed at developers and system administrators; a production deployment follows the Deployment page in the technical documentation. The steps are:

  1. Install Java 21 and Maven 3.6.3, the versions the README lists as requirements. If Maven picks up the wrong Java, set the JAVA_HOME variable and check with mvn --version.
  2. Clone the fcrepo/fcrepo repository from GitHub.
  3. Run mvn install in the cloned directory.
  4. Find the built WAR file in ./fcrepo-webapp/target.
  5. To try it locally, change into fcrepo-webapp and run mvn jetty:run.
  6. Open the REST endpoint at http://localhost:8080/rest/. If the WAR is instead deployed to a servlet container as fcrepo.war, the endpoint is http://localhost:8080/fcrepo/rest.

The README points to four documentation sets: JavaDocs, the Fedora wiki, the technical documentation for the 6.x line, and a RESTful HTTP API page. It also lists the fedora-community and fedora-tech mailing lists and a Slack workspace for questions, and it says the weekly Zoom technical calls are open to anyone. Contributions can be testing, documentation or code, submitted as pull requests.

What the Fedora API Specification defines

The Fedora API Specification 1.0 is the part of the project that matters most to people choosing between platforms, because it states what any conforming server must do. It follows a Candidate Recommendation of 25 June 2018, with minor changes. Its stated interoperability goal is that clients written to the specification should work with any conforming Fedora server with little modification, even though servers need not behave identically.

A conforming server must be a conforming Linked Data Platform (LDP) 1.0 server that also follows the specification’s rules for resource management, resource versioning, resource authorization and notifications. The sections below summarise each area.

Containers, binaries and HTTP methods

Resources are LDP resources: RDF-described resources (LDP-RS) and non-RDF binary resources (LDP-NR), held in LDP containers. Basic containers are required; direct and indirect containers are optional. A client creates a binary by POSTing with a link header naming the NonRDFSource type, and the server then also creates an associated RDF description for it. GET, HEAD, OPTIONS, POST, PUT and PATCH are required for the resource types they apply to; PATCH on RDF resources must accept SPARQL Update. DELETE is optional, and if it is supported, atomicity is not guaranteed when it removes several resources at once.

Versioning with Memento

Versioning reconciles LDP with Memento (RFC 7089). A versioned resource (LDPRv) acts as both the original resource and a TimeGate; a request carrying an Accept-Datetime header is answered with a 302 redirect to the matching version resource (LDPRm). Version resources may be deleted but never modified. The version container (LDPCv) is a Memento TimeMap. The specification describes two patterns: server-managed versioning, where a PUT or PATCH creates a version automatically, and client-managed versioning, where a client POSTs to the version container to mint one.

Authorization with Web Access Control

Servers must follow Web Access Control (WAC), and access is denied when no authorization grants it. Every controlled resource must advertise its access control list through a link with rel="acl", even if that list does not exist. When no access control list exists anywhere in the hierarchy, the server must supply a default; the specification’s example grants public read on the root. The specification also defines how acl:Append behaves for each resource type, so that an agent allowed only to append can add triples or create child resources without being able to overwrite existing ones.

Notifications and fixity

Every resource whose state changes through an HTTP operation must produce a notification, expressed in ActivityStreams 2.0. Each event must carry the resource IRI and the event type, and consumers must not assume events arrive in order. For binary fixity, a client can send a Digest header on POST or PUT to detect transfer errors, and can use Want-Digest on GET or HEAD to retrieve the current checksum and compare it with earlier values. A Digest that does not match the content is rejected with 409.

External content and state tokens

A binary does not have to be stored inside the repository. An external-content link carries a handling attribute with three values: copy fetches the content and stores it, redirect records the location and answers with a redirect, and proxy records the location and serves the content by proxying. The specification warns that external content can expose sensitive data and must be access-controlled. Optional X-State-Token headers let a client detect a changed resource state, and a mismatched token on PUT or PATCH returns 412.

Preservation layout and the wider Fedora ecosystem

Infra Finder lists OCFL, the Oxford Common File Layout, as the preservation standard associated with Fedora. The OCFL specification (version 1.1 in the copy checked, with a 1.1.1 clarifications release dated 7 November 2024) is an application-independent approach to storing digital objects on disk. Each object carries an inventory.json file recording its history and current state, with a sidecar digest file written last, and changes are tracked as forward deltas so earlier versions can be rebuilt. The point for a repository manager is that stored objects stay readable by people and by other software even without the repository application that wrote them.

The fedora.info site also publishes ontologies for repository, WebAC, event, audit, indexing, OAI-PMH and object-state vocabularies, a page of API specification test results, and an Atomic Operations specification. The Fedora API Specification itself defines no transactional model.

Fedora is normally paired with a front end. Infra Finder lists integrations with the Samvera and Islandora front ends and the Camel Toolbox for Apache Camel. This is why the software appears in lists of repository platforms next to systems that ship their own public interface, such as EPrints.

Strengths and limits

Strengths documented by the sources

  • An openly published API contract. The specification states required, recommended and optional behaviour, so a client can be written against the specification rather than a single product.
  • Standards-based building blocks. LDP, Memento, WAC, ActivityStreams and RFC 3230 digests are all existing specifications that the Fedora API Specification refines.
  • Permissive licensing and open governance. Apache 2.0 code and a community governance group, per Infra Finder.
  • Versioning and fixity at the API level. Version history and checksum verification are part of the interface rather than add-ons.

Limits documented by the sources

  • Small core organization. Infra Finder reports one to five full-time staff and funding entirely from institutional memberships.
  • Third-party hosting only. The profile says hosting is available through third-party vendors only.
  • No multi-resource atomicity. The specification does not define atomic operations, and deletes that touch several resources are not atomic.
  • Developer-oriented setup. The documented install path is a Java and Maven build, so the work typically involves technical staff and a front end chosen separately.

For an institution weighing repository platforms for theses and articles, the question is often whether it wants an API-first object store or a repository that arrives with its own public interface. Our DSpace, EPrints, Digital Commons and Fedora comparison covers that choice, and the guide on what an institutional repository is explains the service layer that sits above any of these systems.

The table lists CASRAI pages that cover platforms or concepts a reader comparing Fedora with other repository options is likely to need. It describes the pages, not the products.

CASRAI page What it covers
DSpace vs EPrints vs Digital Commons vs Fedora Side-by-side comparison of four institutional repository platforms
EPrints Platform explainer for the EPrints repository software
Zenodo repository Guide to a hosted general-purpose research repository
What is an institutional repository The service concept that platforms such as Fedora can support
Trusted digital repository Definition of the term used for repositories committed to long-term preservation
DuraSpace Definition of the organization Fedora joined in 2009

Sources

Data: Fedora API Specification 1.0: features and what each does for a repository manager

Compiled from the Fedora API Specification 1.0 and the Infra Finder profile of Fedora. Each row names a feature of the specification, the standard it builds on, the mechanism the specification gives, and the practical effect for the people managing a repository.

Fedora API Specification 1.0: features and what each does for a repository manager
FeatureBuilds onMechanism in the specificationWhat it does for a repository manager
Resource containersLinked Data Platform (LDP) 1.0Basic containers required; direct and indirect containers optional; binaries created by POST with a NonRDFSource linkOrganizes objects and files in a containment hierarchy over plain HTTP
VersioningMemento (RFC 7089)Accept-Datetime on a versioned resource returns 302 to a version resource; version container acts as a TimeMapLets clients retrieve a resource as it was at a given date and time
Access controlWeb Access Control (WAC)rel=acl link on controlled resources; default ACL when none exists; acl:Append rules per resource typeSets who can read, write or only append, with inheritance down the containment hierarchy
NotificationsActivityStreams 2.0Every state change produces a notification with resource IRI and event typeLets indexing and other services react to changes in the repository
Fixity checkingHTTP Digest and Want-Digest (RFC 3230)Digest on POST/PUT checks transfer; Want-Digest on GET/HEAD checks the stored fileDetects damaged transfers and later corruption of stored binaries
External contentFedora-defined link relationhandling attribute with copy, redirect or proxyRecords files held elsewhere without always copying them into the repository
State tokensFedora-defined headersX-State-Token on GET/HEAD; X-If-State-Token on PUT/PATCH returns 412 on mismatchGuards against overwriting a resource that changed since it was read

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Fedora Repository: Open-Source Digital Object Repository Software Explained

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →