Technologies Used in this Edition

The technologies used in this edition changed as the digital world changed. The Canterbury Tales Project underlying this edition began in the early days of personal computing, using mainframe computers for data creation and handline. It continues in the age of social media and smart phones. One may distinguish three stages in the project's changing technologies.

In the first stage, lasting roughly from 1989 to 2006, the project transitioned from the use of a mainframe computer housed at Oxford University Computing Services for data storage and manipulation to use of Macintosh personal computers for transcription. Transcription files were originally prepared on terminals to a VAX system. Robinson wrote a computer program in SNOBAL (then SPITBOL) to collate the transcriptions, to create a record of the collation and then to export that record into various forms. Two articles by Robinson describe this early work, on computer-assisted collation and analysis. As personal computers become more widespread and more capable transcriptions came to be done on Macintosh computers and stored on floppy and hard discs. Robinson wrote a computer-assisted collation program, Collate for the Macintosh, now best remembered as the predecessor of two far more capable programs, CollateX and the Collation Editor. Cambridge University Press, through Kevein Taylor and Andy Brown, enabled the first project publications (1996 and 2000), both on CD-ROM, both using the now-defunct DynaText publishing system.

As early as the mid-90s, as the World Wide Web arrived, it became clear that the models which had supported us though the first phase of the project would not scale to sustain the project in the long-term. Transcription creation and storage needed to move to the web away from stand-alone computers. Especially, we needed to replace the DynaText system used in our publications. By 1998 the defects in DynaText made it clear that it could not be used for our future publications, and Cambridge had decided not to renew their license for its use. Cambridge also decided to withdraw as our publisher. So we needed a new publisher and a new publication system. Robinson (later joined by Bordalejo) created the publisher: Scholarly Digital Editions (at www.sd-editions.com . In those years, from around 1998 through to 2006, Robinson and others developed the Anastasia publishing system to replace DynaText. The first publication to emerge from the combination of Scholarly Digital Editions and Anastasia was Estelle Stubbs' edition of the Hengwrt Digital Facsimile. More publications followed, of the Miller's Tale, Nun's Priest's Tale, Caxton's Canterbury Tales, and Dan Mosser's Catalogue. This combination of publisher and publication represents the second technology phase of our work, running from around 1999 through to 2020.

A key aim of this second phase was that every stage of our making our editions should be done online, through a web interface. Our original intention was that Anastasia would not be just a publication system, but would also be a "Virtual Research Environment", permitting project collaborators (numbering, eventually, more than 200 people) to carry out and check every phase of the project's work, and especially transcriptions and collatiohs. Our first plan, dating from 2005 when Bordalejo and Robinson joined David Parker and other New Testament scholars in the newly established Institute for Textual Scholarship and Electronic Editing (ITSEE) at the University of Birmingham, was to create a single research platform for digital editing which would serve all our needs, and those of many others. In the event that did not happen. Funding pressure led the project leaders -- the New Testament Scholars -- to concentrate on the needs of their users in what became, eventually, the "Virtual Manuscript Room.". Accordingly they went ahead without us, to create what is indeed a superb scholarly resource highly-tuned to their needs. Another factor is that by around 2008 it was clear that the members of the wider University of Birmingham community (not members of ITSEE itself) did not see the advantages to ourselves and to other scholars at Birmingham of our continued collaboration. Once more, we would need to develop ourselves what we needed: in this case, a virtual research environment to house all our work. In 2010 Robinson took up a post at the University of Saskatchewan. A reason for this was that the prospects for sufficient funding for such a complex piece of research infrastructure were higher in Canada than in England. Further, the concentration of digital humanists in Saskatchewan and elsewhere in Western Canada offered a fertile intellectual environment.

We were not disappointed in these hopes. Substantial funding from the Canada Foundation for Innovation, the Social Sciences and Humanities Research Council and the University of Saskatchewan enabled the making of the Textual Communities system, now the daily foundation of all our work. Among other matters, this funding permitted Bordalejo to move to Canada to continue with the project. We needed a system which would both contral access to our work and allow our collaborators to see, in real time, the results of their work. It must enable acquisition and management of page images (including IIIF formats); link transcripts ot images and permit real-time creation of transcripts; carry out computer-assisted collation; generate output in multiple formats and translate between multiple formats. Of these, the real-time requirement was the most demanding. First, around 2008, we attempted to use Django as a front-end managing interactions with users and Oracle XML-DB as a back-end database. Beside the problem that this is propietary software, we found problems with updates being slow. After moving to Canada in 2010 and receiving funding from 2011 on we moved away from both Django and Oracle XML-DB. At first, we attempted to use long-established relational database technology, in the form of MySQL, as our back-end system. However the complex data joins needed to move data between XML and relational tables proved slow and difficult to manage. The breakthrough came in 2014 when a talented student programmer, Xiaohan Zhang, suggested moving to the then-new JSON document database technology, in the form of Mongo-DB. The advantages of this move far outweighed the overhead conversions between JSON and XML. Xiaohan also moved the development and front-end software environment to NodeJS, a change which made it possible for one person alone, initially Xiaohan and later Robinson, to manage the whole system.

By 2018 Textual Communities was fully functional and providing all we needed for the making of our editions. This completed the second phase of technological development. Our initial plan was to use Textual Communities as our publication platform, in the same we had used Anastasia earlier. However, in 2020 Robinson joined a partnership with Lino Leonardi of SISMEL and Prue Shaw, a distinguished Dante editor with whom Robinson had long worked, to update the 2010 edition of the Commedia on which all three had worked, in time for the 2021 700th anniversary of Dante's death. Leonardi had one stringent requirement: the online published edition must not rely on a database, of any kind. Robinson made the edition, now at www.dantecommedia.it according this requirement and came to see that for any digital edition to survive it must not depend on database technology for its delivery. Robinson had already had an experience with the fragility of database technology, when in 2023 the University of Saskatchewan declared they could not maintain the NodeJS and MongoDB systems used by Textual Communities. For the work of the project to survive, the final project webpages -- the edition you are now looking at -- must be delivered without a database.

Indeed, the Endings Project had come to the same conclusion, and issued guidelines for digital projects to follow if they wanted to survive on the web. Our attempts, since 2020, to restructure our editions to give them the best chance we can to survive on the web, constitute the third phase of our technological development. Accordingly, this edition is constructed from thousands of static html files, managed by standard css presentation and javascript function tools. The search tool used is the staticSearch system developed by Martin Holmes and Joey Takeda of the Endings Project. The one divergence from Endings in our editions is that we use the JQuery software library extensively. We will progressively remove this dependency. A particular concern of ours is the availability of manuscript images. We have packaged with this edition images of all the manuscript and incunable pages transcribed and collated in the edition, all held in the "iiifimages" folder, all in iiif form. In many cases, superior images are provided by libraries direct from their servers. In all such cases, we provide a link to those images rather than to the form held in the iiifimages folder. However, where the library server images are not available (as is too frequently the case) we automatically redirect to the version held in the iiifimages folder. Note that the file "images.js" at the toot folder of this edition (thus, here) provides a means of updating the images in this edition as they become available.