TEI Council Meeting

Initials Used for People
• SBSyd Bauman
• ABAlejandro Bia
• DBDavid Birnbaum
• LBLou Burnard
• MDMatthew Driscoll
• DDDavid Durand
• JFJulia Flanders
• SRSebastian Rahtz
• LRLaurent Romary
• CRChristine Ruotolo
• SSSusan Schreibman
• NSNatasha Smith
• JUJohn Unsworth
• EVEdward Vanhoutte
• PWPerry Willett
• CWChristian Wittern

All times are local; i.e. CEST (Central European Summer Time), or +02.

Meeting started at 05-13 ~14:45 with SB, AB, DB, LB, MD, JF, SR, LR, SS, NS, EV, PW, and CW. Day 1 ended at 18:10. Resumed on day 2 (05-14) at ~09:14 with same people plus DD (who had flight problems and thus missed day 1), ending at 18:05.

Council extends warm gratitude to the Koninklijke Academie voor Nederlandse Taal- en Letterkunde not only for hosting our meeting in inspiring surroundings, but also for its generosity in providing refreshments & lunch.

Council reviewed the action items from the conference call of 2004-03-29. (See the notes from that call.)

• CW has contacted Sanskritters; current state of their work to be reviewed at this meeting.
• SB reports that MB has sent him mail, needs to follow up.
• DD not present.
• LB reports ODD for MS has been done; review of current version is on agenda for this meeting.
• SR reports that customization tests have been done but not promulgated.
• DB reports that he discussed collation sequences with JF face-to-face.
• JF reports proposed charge has been written, circulated, and is on the agenda for this meeting.

Discussed e-mail from CR to those who are on both the Council and the Migration Task Force (SB,AB,LB,SS,NS). Portions of this mail are reproduced below. As well as the points made there, it was noted that many case studies (MI W 06) still have outstanding comments from the copy editor, and Tobias Rischer’s checklist needs further work.

Council learned indirectly from Daniel Pitti that the grant period ended on 2004-04-30. Thus the final report is due to the NEH on 2004-07-30; this report will consist of a description of work completed and a financial report.

SS & NS propose (to Council & CR) that they (with AB & CR) take charge of contacting the authors of the various case studies and reminding them to finish up. They will all be at ALLC, so would wrap-up there.

It was pointed out that the section on P5 is (currently) time-dependant, and probably shouldn’t be in what is hopefully a static document. After very brief discussion it was agreed that these documents do need to be static, as there is no further funding to maintain them. Furthermore, tools pointed to by these documents need to be pointed to statically, i.e. on MI web sub-site.

There was some discussion of moving the reports (or copying them) into somewhere other than /Activities, but no resolution was reached.

Council learned that JF and DB had met in Providence and had concluded that the areas of overlap between MS and the proposed PB group were most significant in the physDesc and inline elements. They had agreed that it made most sense for MS group to continue to work on physDesc but to communicate first to avoid divergence; meanwhile the PB group would draw up a draft of inline elements to be discussed in consultation with MS taskforce.

Council learned that the PB group hopes that its output could be included in P5, whence the aggressive work-plan proposed. The group has purposefully assigned itself a very limited scope, but plans to identify issues they may wish to address later.

After some discussion, Council suggests that, because the changes they are likely to suggest are quite limited, that they de-couple their time-line from that of P5, and instead plan to submit one or more specific proposals for change.

LB presented the current state of P5 (reviewing edw81). Reviewed set of for checking chapters.

• Council considered, without definitive resolution, the idea of moving HD (ch. 5, teiHeader) to minor surgery. At least two Council members (LR & PW) anticipate submitting suggestions for HD.
• Council noted that PH (ch. 18, primary sources) will likely need minor surgery with input from the PB group.
• It was proposed, without definite resolution, to move ND (ch. 20, names & dates) to the minor surgery category for the addition of prosopographic tagging as discussed recently on TEI-L.
• Council agreed that the discussion of the relationship between teiHeader and other standards should be excised from IH (ch. 24, independent header) and moved to HD before IH is deleted.
• Council approved LR as the surgeon for TE (ch. 13, terminological databases).
• It was recommended that FT (ch. 22, tables, formulae, and graphics) be slated for major surgery, and that the SIG on presentation be the surgeon.

It was suggested that somewhere, if not in the

, the TEI-C needs to address issues of presentation. A now what? web page was suggested: the SIG on Presentation might be requested to make a first stab at this.

MD says Menota group has several changes (mostly to PH) to suggest. Discussion of call for proposals ensued.

SR demonstrated a bit of SourceForge. Council considered the question which of the various SourceForge features we want to use.

Council agreed to concept of using SourceForge artifacts for bug tracking, feature requests, etc. Discussion of what artifact types there should be deferred for further experimentation.

Council considered the question of whether the ODD source of P5 should be moved to the SourceForge CVS tree immediately, later on in the P5 development process, or after P5 has been released. After much discussion, Council recommends that after each chapter is checked (per edw81), it be put up on SourceForge.

SR asked for input from the Council on what, if anything, we wanted on the TEI Knoppix CD.

LB reports that the developer of the extensions that allow osx to build on Windows has received no response from OpenJade (the project that maintains OpenSP, of which osx is a part). Council agreed that whether or not an official version makes it through the OpenJade project, it should go on the MI/Tools page.

CW briefly summarized a report from John Smith ()

In the ensuing discussion it was pointed out that the problem raised and addressed in the report is a general linguistic problem, although particularly acute in Sanskrit. A morphosyntactic group of TC37 SC4 is also working on this and related problems. It was pointed out that the general problem is similar to that which is encoded by the mirror tags orig and reg: in general, one of the forms could not be automatically generated from the other, and therefore there was a need to encode both explicitly.

LB says that he expects P5 to suggest use of choice for this kind of problem, and in fact to use a Sandhi example.

Council thoughts are as follows.

• The problem is clearly stated and not restricted to Sanskrit: we should therefore seek a general solution
• The solution proposed uses mixed content in a way which seems inappropriate
• Sanskrit group should be made aware of related TC37/SC4 activities
• Our mechanisms (choice) are very general, and we’d appreciate their input in refining our documentation of it (as soon as we have some)

SB attempt to enlist the assistance of those on TEI-L most interested in choice to write a working paper on the issues surrounding use of it and of previous attribute mechanism. 2004-05-27

SR gave the Council a presentation on the current status of the ODD system. After the presentation the question of whether or not ODD format should rely more heavily on RelaxNG, in particular by using rng:pattern instead of classSpec, was raised.

After much discussion Council concluded by asking that the META work-group further discuss the issue.

It was noted that the proposed new modification system recommended the user to express modifications in ODD language rather than in either DTD or Schema. It was an open question as to whether the facility to plug in predefined sets of modifications written in DTD language would be retained.

SR presented the mechanism developed by META group in and after Paris meeting, using an additional ODD language patch which is applied as a customization when creating a schema.

A single dictionary file (teinames.xml) is used to map names. A style-sheet uses this to transform the ODDs themselves, inserting altIdent elements as necessary. A different style-sheet can translate instances; theoretically a third could translate (XML syntax) schemas. In discussion, it was noted that this dictionary-based approach did not extend to providing translations of the element-specific text such as the gloss or desc sub-elements.

This was felt to restrict the usefulness of the current system for the needs of localizers; some of these concerns might be addressed by a more conventional terminological structure.LR (as TE surgeon) look at current implementation and use it as a test case for his revision of the TE module.

Some discussion of how translations would actually take place occurred, but no definitive conclusions reached.

CW reports CE WG had now produced draft replacements for two chapters of P5:

(CH) and. (WD). The former now included a new section on lang identification, which might need some more work.

On the issue of xml:lang, LB requested a summary of counter arguments. CW raised issue of reliance on external standards.

It was suggested that we consider using ident (the ID half of the string match mechanism TD uses) instead of n on lang as the mechanism CE uses to associate a xml:lang value with a language element.

In the ensuing discussion it was noted that the association of the language attribute (be it lang or xml:lang) with a language element in the teiHeader is optional.

Turning to discussion of WD, council noted that this chapter was substantially updated from last year. CW presented the major mechanism (using the g element).

The CE WG feels its work is done. Council congratulates them.

LR brings up issue of object language vs. working language. SB suggests that ana is the appropriate mechanism to encode object language. LB suggests that MS’s new textLang is.

LR reported on the progress of the joint TEI-ISO FS work-group. At the Nancy meeting (minutes may be found here), the group (following ISO procedures) addressed comments that have been received so far. Currently the group has two goals:

• produce at least a portion of document for P5
• have FS (not FD) finalized by August

Some (but not all) members of the work-group will be in Lisbon (at LREC), and they plan to discuss and dispose of the remaining small, mostly editorial, issues, and also the larger issue of re-entrancy.

In August there will be a 1.5 day meeting: 1 half day on FS, and 2 half days on FSD. The goal for FSD will be the same as with FS: a combined TEI & ISO standard.

LB explains that the ISO standard format is not the same as TEI chapter format. Notes that a fair amount of work (on the ISO side) is left to perform the needed integration. There will, in the end, still need to be some minor differences (e.g., possibly references to the other; ISO version will want RelaxNG code as an appendix, etc.)

Question for Council is whether the break with P4 compatibility is OK. Quite a large portion of the world’s users of FS, and several implementors were on the WG, all think P5 version should go on without backward compatibility. Council has no objection to non-backward compatibility.

MD presented the current MS fascicule, which represents not only the changes the task force developed at their meeting (of which minutes are available) expressed in the new ODD format, but also includes some new changes. One of these changes was the subject of significant discussion: since the meeting, the msLooseDesc element had been dropped, and the p element is now permitted to intermingle at the top level as children of msDescription.

It was pointed out that this content model does not allow for fully structured records. Council did not resolve the details of this issue but rather concluded that the TF should do so as it prepares the proposed chapter for P5.

The chapter currently has a lot of material and topics that are addressed by other modules, and will soon need to be moved. editors determine what parts, if any, of MS need to be in CO

Council learned that the task force had only considered elements to an arbitrary depth in the hierarchy (due to lack of time & area-specific expertise), and that further down the hierarchy informal prose (in p elements) is permitted. The task force feels that there is a need to extend the specificity of specialized elements, i.e., to take the hierarchy further down. However, a chapter could be delivered for P5 without doing so first.

On the question of rechartering the group as a work-group (as opposed to a task force) Council agreed that the task force should finish up proposed chapter for P5 and then write a charge to be reconstituted as a WG for hammering out the lower level elements. MD Write examples 2004-06-04MD Write new charge for MS WG 2004-07-15MS TF Finish up MS chapter 2004-07-15

Council entertained report from DD on the progress of SO. In particular, on the immediate goal set forth in Council’s previous conference call to have decisions noted and written down so that P5 could proceed. Summary of suggestions:

• ID/IDREF no longer in Guidelines, replaced by URIs.
• drop entire extended pointer mechanism
• use XPointer scheme framework
• use xml:id instead of id
• Guidelines should recommend practice of using relative URIs, especially for large sets of files
• drop xptr and xref
• the two pointer schemes (bare name and element()) are missing a few methods of locating text that we dearly want (strings, ranges, regexps). In particular xpath(), xpath-2(), range(), string-point(), regexp().
• the W3C XPointer extension mechanism is very cumbersome, so it is recommended we create (register?) our own MIME type (xml/TEI), since the fragment identifier part of a URI is, according to IETF, interpreted based on the MIME-type
• For IDREFS (where the problem is that the attribute value would need to hold multiple URIs) the suggestion is to wrap each URI in curly braces

The question arose, with significant discussion, of whether or not we should include in the Guidelines any methods that have not yet been implemented.

The work group is soliciting input from Council on the following issues, which are listed here along with Council’s response.

• use xml:idCouncil approved
• drop ID/IDREF, but with separate customization module to use itCouncil discussed the question, and charged the work group to examine the issues, in particular those of IDREFS, more closely. Council wants the options for handling IDREFS (embedded pointing sub-elements or any kind of multiple pointers in single attribute value, in which case Council prefers space as a separator.
• Renaming target to hrefAlso referred back to WG
• Establishing a TEI MIME typeCouncil approves, instructs Board to register itJF register the MIME type
• pointer extensions of our own (if so, which ones)Council endorses keeping xpath1(), xpath2(), string-point(), and range(), but dropping regexp()

With respect to the IDREFS problem, concern was expressed that some processors in the world will process IDREFS, but none will process the proposed {} separated list of URIs solution.

Council identified three possible paths to follow:

• keep IDREFS
• use child elements
• separate URIs in a single attribute value

Although concern was expressed about using a blank to separate the multiple URIs in the third method, because escaping blanks in URIs is already an area where users typically make a lot of errors.

SR suggests continuous releases like OSS, rather than waiting.

CW voices preference that we measure P5 by milestones of accomplishment rather than by particular calendar dates.

The soon to be publicized call for participation, which could point out that now is the time for making suggestions for major changes, will have a deadline of 1 Sept for getting suggestions reviewed for inclusion in the Members’ Meeting presentation. Intention is to have a coherent set of ODDs on SourceForge by the end of June. Editors to make release snapshots at their own judgment.

We hope to issue a call for copy editing by the time of the Members’ Meeting.

LB suggests that we revisit mapping of modules to chapters, in particular CO should be broken up. The P5 update page needs to be updated before the call for suggestions, which is going out in 2 weeks, is made. JF Write draft updates to P5 update page 2004-05-27

LB,SR,CW,SB Produce a project plan inventory 2004-06-01LB,SR,CW,SB Produce complete project plan 2004-07-01

The issue of budget for WGs was briefly discussed. The possibility of a prosopography WG raised, but consensus was against.

Next conference call scheduled for Fri 02 Jul 04 at 13:00 UTC.

The following is an excerpt from e-mail which was distributed (in print form) at the meeting for discussion.