Quite a lot of time has passed since the last entry on this blog, and many things have happened with regard to EVT development. It seems to me a good idea to propose a quick recap of the latest events, starting with a bit of history first, though.
There have been several incarnations of EVT, right from the times when the idea of a Digital Vercelli Book was just taking shape and well before even an experimental encoding of the transcription had been attempted. As you can see from the images below, at the time (about 2002-03) I was thinking of a very simple viewer to show images and the corresponding transcription/commentary texts, with a few features to handle the former (magnifying lens, hot spots, graphic filters).
The reference model was that of the Electronic Beowulf edition by Kevin S. Kiern, which looked like an incredibly powerful software to me, something too complex to be effectively replicated with the available resources (at that time: no resources to speak of, actually).
After the transcription efforts were started for good, using analogue photos converted to digital images, I kept looking around and testing all digital editions I could land my hands on. I was quite impressed by the viewer created by B. Muir for the Junius Manuscript, the Electronic Exeter Book and other image-based editions: in spite of some problematic UI choices (see After the editing is done: Designing a Graphic User Interface for digital editions), it was a very effective tool, suitable for all the research activities (image analysis, text search, etc.) that I envisioned for the DVB. It came with a few limitations, though:
- very hardware and software specific: it required a specific version of a specific browser, which of course was the one available for the most widely used operating system (Internet Explorer 5.5 for Windows), but not for Linux or MacOS;
- everything was HTML-based: while a static conversion from TEI to HTML has never been a troubling prospect for me, giving up on all the semantic information slowly and carefully included in the encoded transcription was quite a big deal; note, however, that now HTML is much “smarter” and semantic-friendly (thanks to HTML5 and microdata);
- the viewer was to be licensed, being proprietary software, and while I’m sure I could have bargained a good deal about a VBD-specific customization, I’m a big supporter of the open source concept: doing something good and having its use and circulation restricted didn’t really sound right to me; in any case, have I mentioned I had few resources at the time?
Another interesting software that I had a chance to see in action at about the same time (at Kalamazoo 2005 if I remember correctly) was the Elwood viewer: it sported a very similar mix of good (fast, effective, powerful, multi-platform) and less good (proprietary, some UI issues, still in development) characteristics, so I had to rule it out, too. The good news is that development is finished and that the author is going to open source it after a final code clean up, hopefully sooner rather than later: it’s an amazing piece of software, and my EVT development leader can’t wait to have a look at the code😉
One project that caught my attention, and with which I collaborated for a while, was the EPPT software whose development was directed by Kevin S. Kiernan: it offered very comprehensive features for creating an image-based edition, and the plan was to extend its capabilities to the edition visualization part. Unfortunately, it never received the resources necessary fulfil such an ambitious plan, and it didn’t make all the necessary progress with regard to the presentation part. At the moment the web site doesn’t seem to be reachable any more: the lead developer informed me that this is due to a new server going to replace the old one, which crashed and was retired from service. The really good news is that this software has already been open sourced and it’s available on the SourceForge platform. This is excellent because its production component is mature and very sophisticated; among other features, it allowed to overcome the infamous multiple hierarchy problem which is XML’s greatest limitation. An updated version that could save in TEI XML format would be a great asset to the digital philology community.
What about the TEI community? There have been several projects in the past, some of which have unfortunately been abandoned (TEIViewer, TeiPublisher), while others were, and still are, unsuitable for diplomatic and image-based editions (TEI Boilerplate). Furthermore, almost always I had issues with the UI, or with some other critical aspect, such as supported platforms and OSes etc., of the software. What can I say, I am that fastidious …
So, in spite of my aversion to the NIH syndrome, possibly the worst problem plaguing the open source/free software community, I started looking into a light weight, web-based solution for a digital edition of the Vercelli Book manuscript. Starting from 2008 onwards I was asked to teach text encoding at the University of Pisa, in a Digital Humanities degree course, and I came in contact with a good number of smart and talented students. It was natural, then, to propose EVT as a PBL (Project-Based Learning) subject, with the realistic purpose of having a good experimentation playground and the more distant goal of actually producing something usable. In fact, the first EVT version was already useful for an edition, so both expectations were met.
But the student who created this first version, Francesca Fiorentini, wasn’t interested in continuing its development, so I assigned the task to another student, Francesca Capochiani. EVT v. 2 was born as a completely new project, written from scratch, and looked particularly feature-rich:
Well, perhaps even too much feature-rich: Francesca worked hard and developed her version in a very autonomous manner, so basically I was showed the end result of her efforts. Just while I was grumbling about the need of a full team of developers, one that would discuss and make decisions in a cooperative way, Francesca Capochiani decided to pursue other interests and dropped all EVT-related development. Which reminded me of another advantage of a development team: if you lose a single individual, the common knowledge acquired is shared and lives on, development can continue.
I was thinking about this kind of organizational problems when I met Raffaele Masotti, who worked on EVT first for a stage and then for his BA degree. Meeting Raffaele was crucial for the destiny of EVT in more than one way:
- first of all, Raffaele brought in a careful, meditated approach: see what is working in the current code base, decide which features to put aside (at least for the moment), examine the several problems which still had to be solved; for me, who feared another “let’s rewrite it from scratch!” jump in the dark, this was quite a relief;
- one of the problems that still had to be solved was quite a fundamental one: what is the best way to load your data in a web-based viewer? after quite some study and research, inspired by the recent TEI Boilerplate and another project adopting an XSLT-based strategy, Raffaele turned that question upside down in what is EVT’s approach: you don’t do that, you build the viewer around the data; that’s how EVT Builder was born;
- so we ended up rewriting EVT from scratch, after all, but it was for a very good reason, a complete re-design of its basic architecture that proved to be simple and effective;
- last but not least, Raffaele proved to be very capable in handling EVT development under several aspects: when a dev team was eventually created by accepting more students for EVT-related stages and dissertations, Raffaele took care of introducing his colleagues to the software’s inner workings, of assigning tasks, of merging inputs in the code tree and more.
This last point is particularly important because the dev team has grown quite a lot: first we had Julia Kenny implementing two fundamental features, the magnifying lens and the image-text linking; she also did an excellent job at modifying the XSLT build chain, in such a way that she can be considered a co-author of the software. After Julia we added Jacopo Pugliese and Chiara Di Pietro to the team, dealing with the search functionality and critical edition support, respectively. The four of them can be considered the current EVT core team, to which you could add Giancarlo Buomprisco, currently working on the Digital Lightbox functionality for the DigiPal project; hopefully we’ll see a “light” version of his work in EVT at some point during 2014.
The newly born EVT Builder was presented at the Easy Tools for Difficult Texts workshop (The Hague 18-19 April 2013). The presentation was well received, so I encouraged my valiant students to work on a poster for the TEI MM 2013 (Rome 2-5 October 2013). The TEI Conference was a great occasion to meet people and show our work, and again we got a fair bit of attention, but it was thanks to the support received by the Biblioteca Capitolare that we could push for a an acceleration in EVT development, which led to the beta release of the Digital Vercelli Book shortly before Christmas 2013. The weeks before it were quite hectic: we only met a few times in person, but all the team was constantly in contact through an assorted set of chatting tools, mails, shared docs and even, when necessary, old fashioned phone calls! The development repository on GitLab was hit with scores of commits every day, until the software was ready for its debut. The screenshot below shows the final version of the beta, uploaded on SourceForge soon after the release.
Thanks to the hard work of all the team EVT development continues, for the moment being aimed at fixing bugs and polishing the Digital Vercelli Book beta, after which more important features will be added. But that’s the subject for future posts