TCM07: TEI Council Meeting, Oxford, 15-16 May 2003

Minutes of the TEI Council Meeting Held at Rewley House, Oxford, 15 and 16 May 2003

18 May 2003

Present: Syd Bauman (SB), Alexandro Bia (AB), David Birnbaum (DB), Lou Burnard (LB), Matthew Driscoll (MD), David Durand (DD), Tomaž; Erjavec (TE), Merrillee Proffitt (MP) [Friday only], Sebastian Rahtz (SR), Susan Schreibman (SS), John Unsworth (JU), Perry Willett (PW), Christian Wittern (CW). Apologies had been received from Laurent Romary (LR).

The previously circulated agenda was reviewed and the order modified; these minutes follow that revised order.
Actions from the previous meeting were reviewed; all have been completed, except those on CW, and on PW. The former to be addressed on today's agenda; the latter to be carried forward.

PW6 Sept to set up discussion list on the IMLS grant proposal, to include all Council members not explicitly opted out

CW presented documents CEW01 and CEW06 for review by the Council, noting that an additional working paper on language identification was yet to be folded in to CEW01, and that the version of CEW06 was only a preliminary draft. Council discussed these papers at some length.

It was agreed that CEW01 was close to completion: the WG was asked to consider adding more discussion of beginners' problems; it was felt that some illustrations would make it more accessible (examples were offered from the Menota project); some explicit discussion of typographic issues and in particular of scripts might also be useful. Maybe a section on output (as distinct from input and storage) of characters would provide an appropriate locus for these issues. Some explanation of character properties might also be helpful. Section 4.5 (on XML issues) might perhaps be better located in Chapter 2 of the Guidelines: this was a matter for the editors to resolve.

Turning to CEW06, a free-standing replacement for chapter 25 on the WSD, Council noted that this paper was less complete but endorsed its approach. CW said that it was intended to add discussion of how encoded documents using the proposed scheme would be migrated once non-Unicode characters had been standardized; and to review and expand ways of handling positional variant glyphs, and mapping rules for glyph rendering. There was some discussion of glyphs used to represent abbreviation; the scheme as proposed seemed adequate to the needs of medieval ms encoders in this regard. It was agreed that the proposed character description element might be placed in the Header, or as an external document linked to by a URI. It was noted that the reliance of this proposal on the availability of the c element suggested that the latter should be in the core tagset rather than in the segmentation tagset. The names and documentation used for the specific elements proposed needed to be brought in line with common TEI practice, notably to include more examples Editorsasap to provide sample tagdocs for charDesc elements to workgroup Combining characters such as ligatures should not be represented as nested c elements: the content of a c might be multiple codepoints. Council also requested more discussion of mapping issues in the document.

Council next discussed the CDATA attribute issue, and reviewed document EDW79. It was agreed that there was no viable alternative to the CE workgroup's proposal of converting attributes with textual content (which might therefore include c elements) to daughter elements: it was less clear whether such conversion should be mandatory or alternative. Reviewing the following examples:

buck duck (b)

duckbuck

Council expressed a strong preference for (a) above (b) as an alternative method of encoding (c) or its analog (d), but recognised that there might not always be an appropriate outer "wrapper" element. Council decided that such methods must be made available in P5, but did not decide whether or not the P4-style usage should continue to be available as an alternative, pending further examples.

EditorsJuly 2003 to revise and expand EDW79 to propose appropriate wrapper elements and to add more datatype information. Council thanked CW and the group for their work, and agreed that the group should continue for a further year. Council also agreed to recommend to the Board that funding should be made available for a further face to face meeting of the workgroup.

In the absence of LR, LB reported on the proposed timetable for the joint ISO/TEI Activity on feature structures. Council asked whether, if this work was successful, ISO would take on the task of maintenance. If so, it might be appropriate for future editions of the Guidelines to cite the ISO document rather than duplicate its content. It was suggested that the TEI should recommend to ISO that the document should be freely distributable.

Council was concerned that any changes necessary in the substance of the current proposals should be identified before production of the ISO document began: doubts were expressed as to the feasibility of keeping to the proposed timetable. TE, LR, Editors6 Sept to review chapters FS and FD and propose any needed changes to Council

LB asked whether there was any need to retain the current TEI chapter on terminology. Editors6 Sept to attempt to discover whether the terminology chapter is being used

AB gave a progress report from the Migration Work Group reviewing the working papers so far available for its next meeting, scheduled for mid June. Council members briefly discussed the problems posed by recommending a migration procedure dependent on a single piece of software (osx) which was not consistently available across platforms and suggested that the documentation should be expanded to include exactly what was involved in making osx work, and should also consider other solutions in more detail. Council noted that the various parts of MIW03 were yet to be integrated into a single document.

LB6 Sept to add osx binaries to website and investigate provision of an online conversion service

SS gave a brief report from the Nomination Committee. The following have so far agreed to stand for election to the Council (6 vacancies): Mavis Cournane, Matthew Driscoll, Tomaž Erjavec, Steve Ramsay, Laurent Romary, Chris Ruotolo, Natasha Smith, Edward Vanhoutte, John Walsh, Perry Willet, Christian Wittern. The following have so far agreed to stand for election to the Board (2 vacancies): Patrick Durusau, David Gants, Kurt Gartner, Harold Short, John Unsworth. Michael Beddow and Marilyn Deegan were to be approached, for Council and Board respectively. It was noted that there was still a requiremnent for a defined nomination protocol. Nominations close on 1st June; any further suggestions to SS.
AB presented his work on multilingual markup, also recently presented at XML Europe. In discussion, Council agreed that it would be desirable for TEI to provide canonical translations for TEI element names, attribute names, and default values into as many languages as possible, whether by means of a distinct mapping document (as AB had done) or by additional information in the ODDs. Such mappings would be best developed by user communities rather than centrally by the TEI, which might lead to consistency and conformance problems. (Council noted in passing that the chapter on conformance was also in need of revision, but did not assign any action). It was agreed to form a taskforce to look into ways of defining and maintaining a TEI term bank, with initial membership of CW, AB, SR.

CW, AB, SR6 Septto provide technical recommendations on definition and maintenance of TEI term bank

SR reported on work of the Meta workgroup. Initial priorities had been to define the ODD format in a way consistent with the rest of the Guidelines, and to remove any residual SGML dependencies. More specifically, it was proposed to change the way content models were presented in tagdocs, and to define datatypes for attribute values more abstractly, mapping them to W3C schema datatypes where possible. The work was well in hand, though there were some current problems in generating DTD fragments from relaxng.

In the P4 ODDs, element content models were expressed by embedded DTD language, and attribute list specifications by specific ODD elements. The WG was now proposing to replace the former by RelaxNG; it might also be appropriate to do the same for the latter. The decision to use RelaxNG for this purpose (rather than W3C Schema) did not preclude generation of W3C schemas. Council debated at some length which schema language should be used, and the extent to which it should be exposed to users of the Guidelines, before voting on the following propositions:

+ The maintenance form for the definition of the TEI schema should be RelaxNG + In the printed form of the Guidelines, formal definitions for elements should be given using the RelaxNG

compact syntax + In digital versions of the Guidelines, formal definitions should be available in one or more of the available schema languages (DTD, W3C schema, RelaxNG) as a user-configurable option + It should be possible for users to define extensions in any of the three schema languagesEach of the above motions was carried unanimously. Council recommended, however, that further comment on these fundamental changes in P5 should be sought from the wider TEI community. JU; Edsasapto report schema language proposals to TEI-L and invite comment

The WG had also identified a problem in the use of namespaces. Definition of a TEI namespace while superficially attractive would have the undesirable effect of causing all existing TEI software to fail, since namespaces cannot be defaulted by e.g. XSLT processors. It was also unclear whether there should be a single TEI name space, or multiple ones. Council felt that both questions should be further investigated before any recommendation could be made.

Meta WG6 SeptTo produce a working paper on the namespace problem

At SR's request, Council discussed the current licensing conditions specified in the Guidelines and associated DTD files. It was agreed that the present situation was not satisfactory, chiefly because it gave users of the scheme no rights to modify or even to use the DTD. Council agreed to recommend to the Board that the Guidelines should be released under the GNU Free Documentation Licence, and that the derived DTDs or schemas should include an MIT-style statement of acceptable use similar to that used by the Docbook DTD:

Such a licence would enable TEI materials to be built into Linux distributions and other Open Source products, which Council felt highly desirable. The Board was requested to review this issue at its conference call in June, and also to pursue the feasibility of getting sample TEI dtds and other materials built into commercial XML products.

JU reported that there had been no further report from Patrick Durusau about his proposed work on revising the Tress and Graphs chapter.
JU requested Council members to send him any recommendations concerning the next Chair of the Council by mid June. The new Chair would be appointed by the Board at its next meeting.

All15 June to make proposals for new Council chair to JU

JU noted that membership in the Consortium was still not increasing as rapidly as necessary.
JU reminded Council members that the 2003 Members Meeting would be held in Nancy, France, 6-7 Nov. The programme committee for the meeting, as agreed at the November Board meeting, comprised Peter Robinson, Laurent Romary, and Syd Bauman.
MD reviewed the origins of the two sets of proposals for manuscript description under review by the MS Description Taskforce. A major difference between the two sets of recommendations was that Master had been conceived of primarily for cataloguing de novo, whereas the TEI-MMSS had seen its role as being to facilitate retrospective conversion of existing print catalogues. MP noted that there was no pre-existing data-content standard for the area which Master addressed; MD that the intended user-community for Master including manuscript scholars at least as much as manuscript librarians; DB noted that the recommendations agreed substantially, with only small differences in naming and functionality.

Council agreed that the merged recommendations must support ways of encoding pre-existing descriptions as effectively as possible. After further discussion, focussed on ways of supporting both a structured normative view and an unstructured low-cost transitional view, Council suggested that the best way of reconciling the two sets of recommendations would be to define alternative elements, one for structured and one for unstructured elements, using identical low-level access points. This would be analogous to the existing biblStruct and

bibl, or entry and entryFree elements.

MD agreed to take this suggestion further with the taskforce. He noted also that it might be possible to generalize the model further to include non-manuscript material; it was agreed to postpone this work until a set of merged recommendations adequate to the handling of manuscripts in general (not simply Western medieval and renaissance materials) was available. Council re-iterated its concern that this work should be completed without further delay and agreed that a face to face meeting of the Taskforce within the next three months might help to accomplish this.

MD, MP, DB, Edsasapto fix date/place for MS ftf meeting

DD reported that the Standoff workgroup continued to meet regularly and was making slow progress; there were no major issues for Council to decide as yet. He briefly commented on the scope and status of the various working papers now available on the web site and clarified some questions about their intended coverage, for example that SOw04 was not intended to provide an exhaustive or authoritative list of recommended media formats, but merely to comment on the use of xpointer in combination with some commonly used formats. Council suggested that SOw05 on corpus applications needed to explicitly comment on the existing TEI recommendations for language corpora; and that SOw06 on standoff techniques similarly needed to review existing TEI proposals for accomplishing the same thing as xInclude. It was noted that (as mentioned above) the use of namespaces might be problematic. Council reviewed SOw08 on ways of supporting canonical reference schemes and agreed that these might be a generally useful alternative to the use of xpointer. Some method of supporting multiple schemes was necessary. Finally, DD reported that a paper on conversion between TEI extended pointer and Xpath was in production. Council approved the continued work of the group.
The outgoing Chair reiterated his opinion that the root element for TEI documents should be TEI rather than TEI.2, and that the use of camelCase identifiers should be reviewed. Council agreed to drop the .2 and to add a

version attribute to the new TEI element. The editors (and others) expressed strong opposition to the idea of dropping camel case identifiers.

Council recommended to the Board that funding should be provided for one face to face meeting for each of CE, MS and Meta and rechartered their respective activities for the next year.
Next meeting of the Council would probably be by conference call in September; no date was fixed. A face to face meeting might also be possible in conjunction with the Members Meeting in November.
A vote of thanks to the outgoing Chair was proposed, seconded and carried unanimously.