This version of the TEI Guidelines is a major step forward, and thus merits incrementing the major portion of the version number from 2 to 3. The significant changes that warrant this change are twofold, as follows. The first is the implementation of Pure ODD, which replaces RELAX NG content models with TEI elements in the definitions of new elements and attributes. This means that apart from Schematron constraints, TEI is now defined entirely in TEI. The new specification elements and their use are described in Chapter 22. The TEI Stylesheets have been extensively modified to handle Pure ODD content models. The second is the introduction of the new Processing Model specification in section 22.5.5. TEI Processing Models permit the writer of an ODD to specify how TEI elements might be processed for different output formats.

Release 3.0.0 of the TEI Guidelines also introduces other new features and resolves a number of issues raised by the TEI community. As always, the majority of these changes and corrections are a consequence of feature requests or bugs reported by the TEI community using the GitHub tracking system. If you find something you think needs to change in the TEI Guidelines, schemas, tools, or website, please submit a feature request or bug issue at https://github.com/TEIC/TEI/issues for consideration. A full list of the issues resolved in the course of this release cycle may be found under the 3.0.0 milestone.

Some of the notable changes other than the Processing Model and Pure ODD in this release include: The new msFrag element was added to permit Manuscript Descriptions to contain virtual reconstructions of fragmented documents as well as to analyse their components. The new annotationBlock element was added to group together linguistic annotations. The new transcriptionDesc element was added to describe the set of transcription conventions used, particularly for spoken material. The oVar and pVar elements are no longer recommended; rather, use of oRef and pRef is recommended instead. The attributes of oRef and pRef have been adjusted to accommodate this usage. The type attribute of stage now allows multiple values. The TEI and table elements now claim membership in att.typed (and thus gets the type and subtype attributes). The hand attribute was removed from att.damaged, att.textCritical, and att.transcriptional, and added to a new att.written class, to which att.damaged, att.textCritical, att.transcriptional, ab, closer, div, fw, head, hi, label, line, note, opener, p, salute, seg, text, and zone belong. The seg element was added to the content of notatedMusic. The explanations and discussions of several features have been improved, including: the att.scoping attributes target and match the attributes of att.datable.w3c the attRef element the revisionDesc element The display of element documentation has been re-ordered so that notes and examples now precede the content models (given in both Pure ODD and RELAX NG). And, of course, many typos were corrected.

In addition, improvements have been made to the XSL stylesheets (which provide processing of TEI ODD files for Roma and OxGarage as well as other TEI conversions). The Stylesheets are maintained separately from the Guidelines and are at https://github.com/TEIC/Stylesheets.

This release is dedicated to the memory of our colleague Sebastian Rahtz (13 February 1955 – 15 March 2016). We are greatly diminished without his generosity, wisdom, and humor. Release 3.0.0, however, has a lot of him in it. He was the architect of the new Processing Model, and had implemented most of the support for Pure ODD in the Stylesheets. Unlike poets, the creators of software and living standards generally don't get to make a monumentum aere perennius. At best, we can hope that our work will be carried on, rewritten, extended, and refactored. That process has already begun for Sebastian's TEI Stylesheets, and they will continue to evolve to meet the community's needs going forward; but they, and we, are immeasurably better off for his contributions.