Guildford Protocol

Written and maintained by Thomas Krichel

This is the version of 2016–03-10. We still have the year 2000 version available.

1: Introduction

This document is the Guildford protocol. It is named after the town where it was first written. The protocol provides a set of rules for the publication and exchange of documents on the Internet. It could be implemented in any group that wishes to distribute documents on the Internet.

The idea behind the protocol goes back to a statement by William L. Goffe. On 15 July 1995, he wrote on the (now defunct) NetEc-admin list:

What I would suggest is this: a distributed system with any number of sites, each mirroring each other. It would have extensive bibliographic functions (cross-referencing, etc.), and my favorite, digital timestamps for when the papers were put up. For archives outside it, papers could be listed, but no cross-referencing. But, such archives could “join” the system (say it was written in perl so could run on NT as well as Unix). Then you'd have the best of both worlds: distributed, anybody could join, extensive cross-referencing, the whole works. Such a system could easily grow with the profession's use of the net. Such a system would GREATLY benefit the profession.

The way to achieve this “global and local” archive is through a comprehensive distribution process that is based on a set of archives. An archive is based on a machine that makes data available. It is a place where original data enters the system. The data are then distributed to any number of sites. A site is a collection of archives on the same computer system. It usually consists of a local archive augmented by frequently updated copies of remote archives. The local archive is maintained on the local computer, whereas the remote archives are maintained on other computers. We call a frequently updated copy of one archive on a remote site a “mirror”. There is no need for every site will need to mirror every archive in the system. Some may only mirror bibliographic information rather than the papers to conserve disk space. Others may mirror all the files of an archive. Others will mirror only parts of a few archives.

All archives hold papers and metadata about papers, as well as software that is useful to maintain archives. Everything contained in an archive may be mirrored. For example, if the full text of a paper is in the archive, it may be mirrored. If the archive does not wish the full text to be mirrored, it can store the papers outside the archive, for example on a directory that does not belong to the archive.

The Guildford protocol aims to find a set of minimal restrictions on archives such that a global and local system will work. A second key aim of this document is to provide a set of rules such that if they are followed locally, require almost no central effort. However a small amount of work has to be provided by a central archive. This archive is called the core archive. It contains ReDIF templates that describe all known archives in the system.

A limitation of this document is that it will not deal with charging money for metadata. We assume that the description of documents is free. Limiting the access to the documents themselves is possible but remains outside of the scope of the document.

A second limitation is that the protocol does not deal with archiving and preservation issues. A key feature of the protocol is that each document has exactly one home site. If the home site withdraws the document it is withdrawn (after a short delay) on all sites that maintain copies of the document.

The last limitation is that the protocol is not concerned with providing end user services. For example the protocol does not provide any ideas on how to present documents on a web server, on how documents should be indexed etc. However one of the key features of the protocol is that software used to perform these task can be written by a community of contributors and distributed among sites for the benefits of everybody.

2: Definitions and conventions

The “authority” is a group of people that have come together to implement the Guildford protocol on a set of documents and metadata. Authorities are listed at the end of this document.

ReDIF (Research Document Information Format)” is a set of rules to encode information about papers, series, and copying rights. It is discipline independent and independent of an organisational structure that supports its creation and deployment.

A “series” is a collection of documents that are kept together.

An “archive” is a directory on a computer that is open to access by ftp, http, or https. It holds a collection of series of papers or a collection of data about papers held elsewhere.

A “RePEc archive” carries data formated in ReDIF that pertain to Economics and use RePEc as administrator. It may also carry the full text of the documents.

A “site” is a collection of (normally one) local archive plus any number of mirrored archives. For the purpose of identification, sites and archives are treated identical. Normally each site runs on archive and mirrors several others.

The “core archive” is a single machine on which a limited number of important files are kept. These include the core templates (see the ReDIF documentation) of all participating archives.

The “administrator” is the person who keeps the core files on the core site.

“mirroring” is a process by which copies of series of documents are made from one site to another, such that the contents of the archive is the same on all sites except for a short delay that is inevitable.

3. The archive

Every archive has an identifier. The convention is that all archive identifiers have three letters. Archives store document metadata and, optionally, documents. Each document belongs into a series. Series identifiers have 6 letters. The protocol uses some reserved words. They have four letters. Archive identifiers are awarded by the administrator. The series identifiers are fixed by the archives in consultation with the administrator. The reserved words are those mentioned in the protocol.

An archive contains a set of files. All names of files are case-insensitive. All files ending with the extension .redif are files that contain ReDIF templates. These files are assumed to be in the UTF-8 encoding of Unicode, unless they contain a byte order mark indicating a UTF-16 encoding. ReDIF templates may also be stored in files with the legacy extension .rdf. These files are assumed to be in the Windows-1252 encoding unless they contain a byte order mark indicating a UTF-8 or UTF-16 encoding. Any file ending in .redif or .rdf is assumed to contain ReDIF templates.

Let archive_identifier be the indentifier of an archive. Let the . (dot) denote the location accessible at the URL of the archive. Then the structure of an archive is as follows.

./archive_identifierarch.redif a file describing the archive using a single ReDIF archive template. This file may also be called archive_identifierarch.rdf. This file is mandatory.

When an archive is moved from one location to another, it is sufficient to change the archive location in the archive_identifierarch.redif file on both the old and new location. If access to the old location is impossible, then the administrator needs to be informed to perform the change on the core site.

./archive_identifierseri.redif a file describing all the series in the archive using a sequence of ReDIF series templates. All series in the archive are described in this file, one template for each series. This file may also be called archive_identifierseri.rdf. This file is mandatory.

./series_identifier/ a directory where papers and metadata for the series that is identified by series_identifier are stored. All files that that pertain to the series series_identifier must be stored in that directory. Files that contain ReDIF information are called ReDIF files. Their names must end in .redif or—respecting legacy—.rdf. A series directory must not contain further subdirectories. Otherwise the structure of the series directory is free. You may put all templates for all papers in the series in one file or you may put each paper template in a different file, just do as you please.

./inst/ a place to store ReDIF files that contain institution templates. See the ReDIF draft for further information about that template.

./soft/ is for software that is written locally. For example, an archive may wish to write a specific procedure by which its ReDIF-Paper data is translated into html.

./pers/ is for person templates.

./conf/ is the location of configuration files for software. It does not matter whether the software is supplied by the local archive or a remote archive.

For any archive, the collection of the contents of ./archive_identifierarch.redif and ./archive_identifierseri.redif are called its core templates. They are mirrored on the core site.

The archive directory and the directories must be indexable, or indexing must be simulated.

Each site may mirror a number of series from any number of archives. If any site mirrors any series for an archive, it may mirror the complete subdirectories of the series or all ReDIF files (ending with .redif or—respecting legacy—with .rdf) of the series. If an archive does not wish the papers to be mirrored, then it will store them outside of this hierarchy.

4. Authority-specific details

4.1 RePEc

The RePEc core archive has the handle “all”. It operates at RePEc archives maintainers are free to choose the series identifiers that they want, but series identifiers must have exactly 6 letters or numbers.