TEI Technical Council Meeting,

<date notbefore="2012-04-16" notafter="2012-04-18">16–18 April 2012</date>TEI WebsiteTC M 50

Converted from google doc minutes taken during the meeting.

Added note about inclusion vs. exclusion version of Lite.CleanupFixed encoding errors reported by KSH on 2012-05-23

Piotr Banski (PB), Brett Barney (BB), Lou Burnard (LB), James Cummings (JC, chair), Kevin Hawkins (KH), Martin Holmes (MH), Sebastian Rahtz (SR), Paul Schaffner (PS), and Rebecca Welzenbach (RW)

Gabriel Bodard (GB) (via Skype), Stuart Yeates (SY) (via Skype)

Monday April 16th 2012

1.1 Morning

1.1.1 TEI – Google Books

  • KH and JC reported on progress with Google Books output of TEI.
  • After the last meeting, KH emailed Ranjith (the Google contact) with some questions, but hasn’t got responses yet. Ranjith also asked questions about TEI which would help him to market the service internally to Google, and JC created a Google doc to help answer those. All help with that appreciated – email JC for the URL. JC and KH may also be able to add metadata from the original MARC files from the public domain ECCO-TCP files which would make it easier for Google to process those files.
  • LB suggested developing some tutorial material showing how you can use Google Books-generated TEI as the starting-point for an encoding project.
  • There was some discussion about the level of error and accuracy, both in encoding and in OCR, to be expected from Google. SR wondered Why should we care whether Google gets TEI right or not? Just use whatever the best they can produce.

: Get more examples from Ranjith of the current state of the output so that we can calculate some statistics on accuracy, and encourage Google towards making this output format public.

1.1.2 TEI – EEBO / ECCO -TCP SR:

There are three specific problems remaining in the conversion from EEBO-TCP to P5, two of which are in the spreadsheet from MM:

  • signed : Is this a phrase-level element for only a signature, or is this a block-level element that is really a signature block? If the latter, then more structure is required within it. If there are multiple signatures, EEBO-TCP encodes them as a list, which TEI doesn’t allow. The proposed action is to loosen up the content model of signed.

    LB:

    Once we loosen it up, we open the door to it having a lot of extra structure.

    JC:

    The Guidelines are very inconsistent on this; they show examples of signed inside closer preceded by salute, as well as signed containing the salutation within it.

    PS:

    In many examples, there is considerable overlap between the signature and the name.;

    MH:

    There are two distinct issues: the Guidelines problem (bug 3439980), and the issue of converting the EEBO-TCP encoding to TEI.

    JC:

    We should decide one way or the other, for example we could choose the block-level signed option.

    LB:

    In the case of the “RICHARD CUFF” example, how would we tag it if signed were a block element?

    JC:

    Just as a name

    LB:

    The purpose of salute is to mark a formulaic phrase which is some sort of expression of politeness, whereas signed (in this new view) is the whole block of stuff at the end of the document that attests that someone has appended their name to it.

    RW:

    Is a name required for a signed?; LB: No: “Yours etc.”.

    BB:

    This raises the issue of the difference between signed and closer.

    SR:

    We could extend signed and not worry about anything else, so older documents won’t be broken, but “soft-deprecate” other usages by removing examples of it from the Guidelines.

    MH and JC:

    We should look at the content model of all of these things.

    : Create a council working group to look at this issue, and come back with a clear recommendation before next council face2face.

  • Lists inside labels: SR:

    We can’t remember the context of that, so we’ll have to come back to it.

    PS:

    It’s a minor issue that we can ignore it; it was an aberration in EEB-TCP

    : No action needed at this time.

  • byline inside figure: Illustrations are signed or attributed all the time, so we need a way to encode that; in EEBO-TCP this is done with byline.

    JC:

    figure can contain signed.

    PS:

    signed is a performative claim of authorship, whereas byline is an attribution.

    MH:

    Are we trying to solve the EEBO-TCP import issue, or address broader concerns about the content model of figure?

    : SR will convert the EEBO-TCP byline to signed, which solves the issue of the EEBO-TCP import; the broader issue of figure will be addressed in the discussion of floatingDiv later.

  • trailer/@type is used in EEBO-TCP, by analogy with head.

    JC:

    If something is repeatable, then it needs @type. All agree and the record should show this is a general principle when deciding whether something should get att.typed.

    : SR will implement this.

1.1.3 TEI Bugs and FR: (Green)

  • 3480609: JC:

    Someone needs to volunteer to check that every place where licensing is mentioned are correct.

    : RW will do this, and will report back about any inconsistencies. JC will close the bug, and a new one will be opened for any issues which emerge.

  • 3472477: JC:

    We agreed on the telco to add a tei_ prefix to all ids.

    LB:

    We could change all the ids in the actual source.

    JC:

    This would be a major task, wouldn’t it?

    : MH will undertake to change the stylesheets to implement this.

  • 3471119:

    : MH will implement the consensus (no spaces around em dashes used parenthetically). RW will create the styleguide document with help from KH who knows the CMS system already. The text style guide will be deleted.

  • 3395448:

    : JC will try to discover whether Virginia is technically still a partner or not. When it’s resolved, KH will implement any changes to the website.

  • 3511134:

    General consensus that it would be a good idea.

    JC:

    Should we include something to distinguish atts from elements as in MH’s suggestion on the ticket?

    GB:

    We could use a prefix such as #tei_att_type or #tei_el_type.

    JC:

    Another obscure format for people not to understand? Maybe simpler solution?

    : JC and MH will conspire to implement this or something like this in the stylesheets.

  • 3437509:

    The proposal has already been accepted to soft-deprecate @key, by suggesting that people switch to @ref. LB initially objected strongly to the deprecation of @key, but was reluctantly convinced that gentle encouragement to use @ref instead (with a private URI scheme) might be tolerable.

    KH:

    There is no clear consensus on exactly what format magic tokens should take in @ref, but we seem to be tending towards private URI schemes. General discussion reveals that uses of @key in the Guidelines and in examples reveal some of the disadvantages of it: it’s not clear in many contexts what @key values refer to.

    : KH will review the whole thread and make a recommendation about the form of @ref, and make the soft deprecation changes to the Guidelines; these will be reviewed by LB.

  • 3432520:

    SR and PB have almost finished implementing this.

    : SR and PB to finish implementing this.

  • 3046288: JC:

    This has been implemented, but is awaiting some further examples for the Guidelines.

    : PB will contact LR and ask him to provide further examples.

  • 3041605: KH:

    This is mostly done, but there are some historical details or old meetings which need to be added, and it’s not clear where this could be placed. There’s no place to mention the Kingston meeting, so it won’t be mentioned.

    : KH will close the ticket.

1.1.4 Amber Bugs/FRs:

These were divided into topical groups as suggested by email to council list as of 2012-03-25

  • Group A: LB, JC, GB, MH.
    • 3497369 MH:

      The original typo is fixed, but the ticket is the only one relating to the larger discussion of TEI Pointers.

      LB and JC suggest dealing with the topic on this ticket right now.

      LB’s suggestion on the ticket is “I think the specific use of XPath in the context of the location-referenced method for critical apparatus needs to be considered and (if it’s applicable) exemplified in the Guidelines.” Expanding on this, what it really means is that the @from and @to attributes, which are xsd:anyUri, could take xpath(), and we should suggest and exemplify this in the Guidelines. Use of @from and @to is possibly better than @loc (which is data.word+).

      : (JC) Strike the sentence that’s under discussion, because then the paragraph is less confusing. We should also add a new section (12.2.4), which explains that you don’t have to use ids in @from and @to; you can also use any URIs and XPointers with TEI XPointer schemes.

    • 3497356

      The suggestion is that app should be able to be organized into a listApp element. The group consensus was that there is no organizational principle by which different sets of app would likely be listed separately, so there’s no need for a special element.

      : MH will ask the submitter (and the TEI-L, pointing to the ticket) for any examples of the use of div which suggest that listApp might be useful (for instance, organization of apps into multiple divs with @type).

    • 3496998

      : LB will add an additional example of how to combine lacunaEnd inside a rdg. Ticket will be closed.

  • Group C: BB, KH, RW.
    • 3475748

      The group agreed that in this ticket, Laurent’s proposal is consistent with an example given in the reference page for usg, which explicitly suggests using usg type=”style” to indicate usage such as “literal”, so we recommend accepting this change. This should be implemented.

      : KH will implement this change in the guidelines offers to do this.

    • 3440771

      While the prose of the Guidelines says “All uses of geo within a document are required to use the same coordinate system,” prompting this ticket, which wants to be able to have more than one, we found it actually is already legal to have multiple geo tags in a location, as well as to have multiple geoDecl tags in the header.

      KH:

      the discussion comes down to: can you have multiple geos within location, or if you want to use multiple geos, do you actually need to use more than one location element to contain them

      Council wants to better understand whether all geographic coordinate systems are really capable of being converted from one to another (i.e., From any one coordinate system can you truly derive the place’s value in another, making it truly redundant to encode a location according to more than one system?)

      PFS pointed out that in a single document you might need to use more than one system (e.g. American Grid and British Grid)

      Council agreed that it might be necessary to use more than one coordinate system within a document, and since geoDecl is repeatable, it is already possible to declare more than one coordinate system

      Council recommends one location containing more than one geo if the location is really the same place in the universe, but different systems are being used to refer to it. If there is really lack of consensus about the location (for example, Camelot), use more than one location, each with its own geo

      : SR to coordinate, KH and GB to assist:

      • document how to point from a geo to the geoDecl in order to indicate which coordinate system is being used.
      • Remove from the Guidelines the statement that “All uses of geo within a document are required to use the same coordinate system.”
      • edit the definition of location to say something like “defines the location of a place as *one or more* sets of geographical coordinates…” to indicate that more than one system may be used at once
      • clarify relationship between location and geo in the prose of the Guidelines
    • 3458222

      add pc to either pPart.edit or pPart.editorial (both were suggested in the Sourceforge comments), but we aren’t sure which one

      In sourceforge, Lou noted some text in the guidelines that would also need to be revised—the implementor should take this into account

      : LB will add pc to pPart.edit and revise the prose accordingly

    • 3439980

      There is inconsistency in how signed and salute are used.

      One proposal is that signed and salute are siblings, you decide what goes in each, and they are both contained by closer;

      There certainly needs to be more distinctive definition of signed and salute, which right now are nearly identical. In particular: if we agree that signed depends on the presence of a name or the stand-in for a name, this should be made clear in the definition of the element

      The “thine to command” example should be adjusted so that “thine to command” is within salute, rather than signed. The same for “thy repentant husband for his diloyaltie”

      It was agreed that there is ambiguity between signed and salute, but accept that there will probably always be ambiguity here. For example, “thy repentant husband” might arguably go in either place

      When it comes to block of signatures: either

      • expand closer to include list, and enclose each signed inside an item in the list or
      • Expand signed to allow a list, in which each name would be an item

      SR wants to know why we insist on sibling salute and signed, since we all acknowledge its ambiguous/subjective. “What does it buy us?”

      A new sigBlock element was proposed, which would be a child of closer.; A more general action would be to allow list within closer. If instead we create sigBlock, we don’t need it; We seem well-disposed to the idea of sigBlock, but BB asks us to consider the name (and implications of ‘block’) to make sure its rational across the other elements.; (MH and some others like the name)

      : LB to chair an ad hoc council working group to report back on this, KH, LB, and PS to be members, others as necessary. (confirmed later in minutes)

    • 3500088

      Original ticket pointed out an error in an example, which has already been corrected

      In the comments, discussion turned to whether one might define more than one metrical syntax by including more than one metDecl in the header.

      Council agreed that this could be done (as with geoDecl above). In order to point from an instance of meter to the appropriate metDecl, use the @decls attribute. If this is not already stated in the Guidelines, it should be added. the guidelines if it’s not already there. Need to assign someone to do this.

      In the SourceForge comments, MH suggested that the metDecl should also specify what flavor of regex is being used, but KH pointed out that this is already defined sufficiently in the definition of @pattern, so nothing needs to be done here.

      : MH to re-visit to determine what’s unclear and correct Guidelines.

  • Group E: PS, SR, PB.
    • 3423686

      : SR will nag the Ontology group as on the ticket.

    • 3288293

      : PB will add a new attribute to app.pointing, called @targetLang.

    • 2724997 and 2507305: PB:

      timeline works for audio, and facsimile works for static images; what’s required for video is some clever combination of the two.

      Consensus is that we need to create a working group to make recommendations.

      KH:

      We should make an open call on the TEI list for people who could participate in the group.

      : JC will write to TEI-L asking for people who are interested in this area to get in touch. He will run the message by Council first. These people will join a Council working group to come back with a consistent proposal to be discussed later at this meeting.

1.2 Afternoon

1.2.1 TEI and Schematron Tests

SR provided an ad-hoc lecture with slides in an effort to educate MH and others on how Schematron fits into the TEI build process.

  • elementSpecs have constraintSpecs. constraintSpec has a scheme (normally isoschematron). Inside constraintSpec, constraintss have assert elements, which have @test attributes, which are xpath; if the xpath tests false, the assertion will be fired, and its contents will appear on the console when you build or validate. There is also a report element which is similar, but fires when true. In Roma, you will also get a schematron schema which you can also use to test your document against. This document is essentially a compilation in Schematron of all the TEI constraints.
  • P5 has a Test directory in which there are test instance documents for specific features, and ODD files which can be used to generate schematron which will be run against the test document. There are also “detest” documents which are designed to fail. When these fail, the results are compared against a number of known error messages, to confirm that the expected errors occur. The error messages expected are stored in “detest.log”. They have to be in the correct order, so the recommended way to do this is to run the process locally, and then copy the resulting log back into the detest.log. These are not well documented, and scrappily implemented; we need to be more methodical about how to do this, and in making sure that when we add or change elements, classes and attributes, we make an effort to imagine possible errors and write tests for them. There are also hand-crafted XSLT testing routines in e.g. Utilities/validator.xsl. Also remember that you don’t need to create tests for things which can be covered by your examples; the better and more extensive your examples are, the more problems will be caught initially anyway, because the examples themselves are validated at compile time.

: MH will write a new section for “How to edit the Guidelines” which explains all this in more detail; this will then be checked by SR and LB.

1.2.2 TEI Exemplars (e.g. Tite) JC:

Who owns the existing exemplars, and who should be keeping them up to date?

KH:

With a TEI example like Tite, there’s no mechanism by which a change in the TEI which should precipitate a change in Tite will in fact do so. There probably should be. If your define your customization by inclusion, you won’t get new elements, but if by exclusion, you will inherit more changes. This makes no difference, though, to the prose in the customizations, which ought also to be updated.

JC:

Only Lite and Tite have prose text associated with them. They are also the only examples (at present) which are maintained by the TEI.

JC and LB:

We undertake to try not to break any of the other exemplars, and they are tested during the build process.

KH:

The contradiction is: even if we think we’re not going to maintain the prose of Tite and Lite, we are updating the ODDs themselves, so they can get out of sync.

SR:

We should also remember that e.g. Tite might add its own element.

LB:

Lite is definitely a document of its time. It could be orphaned.

KH:

Tite has some bugs and frs on it, but is not actively under development.

MH:

Can we make a decision about the future of Lite now?

LB:

We should mothball Lite, and add a note to the effect that it’s old and no longer actively maintained, and that while he schemas will continue to change, the prose will not keep up with them. There is a version of Lite which is defined by inclusion, but the exclusion version is currently used to generate the schemas Actually, according to tei-council discussion in Julhy 2012, we have been using the inclusion version for quite a while., so the schemas will keep changing with the addition of new elements, taking it out of sync with the prose (e.g. pc and postscript).

MH:

We should make sure the inclusion odd is up to date, regenerate the schemas, then carefully update the prose to match. At that point, we can package it and archive it, as a stable and unchanging resource for those who want to use it.

: LB and MH to make sure a new-style inclusion ODD is up to date, regenerate schemas and update prose to match. The result should be packaged and archived as a stable and unchanging resource.

KH:

We should do the same thing with Tite as above, because there’s no more funding for its development. There were different versions of Tite—the vendor Apex made some customizations—but the process of reconciling them is nearly complete. I would like to revisit Tite in future, but it should be archived first.

: KH will collaborate with LB and MH to carry out the same process of freezing with Tite.

1.2.3 Best Practices for TEI in Libraries KH:

The SIG on libraries produced this document to emphasize interoperability, make more specific recommendations than the main Guidelines, and encourage TEI’s use in libraries. Now the question is whether the Council should be involved, and if so, how much; should we adopt the document and continue to maintain it, or should it be the responsibility of the original community?

SR:

We could look for external funding to find ways of integrating the document into existing TEI practices.

LB:

We could treat it as a case study, just as we did with the Genetic Editing group. But there are aspects of the document (such as the concept of levels) which may not have general application in the guidelines.

SR:

It wouldn’t be unusual to have material in the Guidelines which is specific to one community. There are things in the customization that are more specific than could ever go in the Guidelines.

KH:

Yes, the ODD is very restrictive. But that is useful in itself, as an exemplar. We could make the best practices ODDs, which are very restrictive, available from the TEI website list of community customizations, but not maintained by the TEI. Meanwhile, the Best Practices document could be examined by Council and all use possible made of it in moving forward with P5.

GB:

The case of EpiDoc is parallel, in that it’s (now) a conformant customization of TEI which is highly prescriptive and restrictive.

JC:

EpiDoc has a community willing to keep supporting it, but the libraries document seems to lack that community support.

GB:

There is a large community of people using EpiDoc, and I produce a new EpiDoc schema every time there’s a new TEI schema.

: KH will make the customization available through the TEI website (we will not incorporate it into our build process because it’s too restrictive); the document itself will be mined for good ideas, and turned into a set of recommendations for changes to the Guidelines.

: KH to submit tickets for items to be added to the main Guidelines, since most of the items are likely to be independent and can be assessed separately.

1.2.4 TEI Bugs and FR

  • 3496958

    GB added a note to the ticket saying that the availability of a single rdg is intentional and required.

    : (MH) We leave the content model alone (it’s horrible), but we use this as an example for Council to collaborate on the creation of one or more Schematron constraints which will enforce the presence of at least one rdg, to everyone’s edification.

  • 3496954

    : None required, ticket closed.

  • 3496951

    The only reason we can imagine for the use of @wit on rdgGrp is as a shorthand way to list all the witnesses who show up in a rdgGrp (although this information can easily be derived from parsing the content rdgs).

    LB:

    The availability of @wit on rdgGrp is an accidental byproduct of the class system. We should modify att.textcritical so that it no longer supplies @wit; then we add @wit explicitly to lem and rdg.

    JC:

    We should actually define a new class att.witnessed which has @wit, and add lem and rdg to it.

    MH:

    This breaks backward compatibility, so we should send a message to TEI-L first, to check that no-one is using it.

    : LB will carry this out.

  • 3496949

    : GB will review the particular example mentioned on the ticket, and revise and expand it: firstly, use a straightforward witness; secondly add an example for the “unattested” case.

Tuesday April 17th 2012

2.1 Morning

2.1.1 Report from GB

GB re-opened a ticket closed yesterday: he believes that an app with no rdg at all is acceptable and deliberate. It might contain a ptr or note or lem, and maybe also ref. He would like the Schematron rule to not require a single rdg

MH:

if GB can supply what the minimal requirements for app should be, then MH can write the Shematron rule to test for *that*, instead of specifically for rdg

2.1.2 TEI Namespaces and Conformance KH:

found guidelines confusing: in what cases, when you customize, should you put an element or attribute into a new namespace? Identified a few errors in text, at least one of which has already been corrected. Is it true that *any* customization that is *not* a pure P5 subset requires a new namespace?

JC:

Yes, because you’re modifying the elements that are there, which have been agreed upon by the council. It’s no longer the TEI element; LB and SR agree with JC

MH:

Say you added an attribute (like type) to an attribute class that didn’t have it already. Can you do this in a new namespace? SR says you can; JC agrees.

JC:

benefits of long-term clarity outweigh the annoyance of having to put namespaces everywhere

2.1.3 Chapter/module fragmentation and consolidation (e.g. should things like ‘dates’ become a chapter of its own)

JC suggests Group E people report back if they discussed this.

SR:

the group didn’t address this in break out because we planned to discuss as a group

JC:

realized that “dates” is only a small piece of the Names/Places chapter. Instead, should “dates” information throughout the guidlines be pulled into its own chapter?

PB:

points there are other places where this this is an issue

LB:

Guidelines are full of cross references for this very reason

JC:

If separation this could de-couple the relationship between chapters and modules

MH:

if it’s fine to de-couple the guidelines from the modular structure of the schemas, we can instead think of them in terms of user-friendliness. Chapter 11 (Representation of Primary Sources), e.g., which was proofread by the council in November is just too long

LB:

the Guidlines aren’t meant to be read through (though SR suggests that chapter structure suggests reading is suggested). This would be a major job—not a small/minor change; JC agrees.

SR:

we can’t just do this piecemeal, by splitting dates and times into its own chapter. We would need to investigate the whole structure of the guidlines. Where would we begin?

JC:

the original suggestion was indeed to create a module just for dates, which would correspond to the new chapter. THis would break people’s ODDS/not backwards compatible. Such reorganisation maybe should wait until major structural changes are being suggested.

GB and SY both recommend a more piecemeal approach. GB says we can go forward with dates and places without re-assessing the whole thing

SR, JC, LB recommend instead an overall analysis to establish principles/approach.

JC:

main concern is for the html output of the guidelines to change—but from his point of view this could then lead to a change in modules

SY:

If tags are used together, they should be explained together. If documents that use date also use place, then it makes sense to document them in the same place.

JC:

Almost all combinations of elements could be used with each other, sometimes semantic groupings make sense, other times element usage cuts across genres and types of encoding.

LB:

the Guidelines are deliberately top-down, because each element/attribute can be used in many different circumstances. The Guidelines explicitly are *not* meant to be derived from specfic use cases.

JC:

statistical analysis of documents or schemas might be interesting, SR agrees it should be part of the big picture

LB:

if you go along with the assertion that the Guidelines are organized in a top-down matter, then a good reason to break off a section is that it doesn’t fit where it is

JC:

This was exactly the motivation for the discussion of breaking off ‘dates’.

MH:

two models of organization: semantic areas, and levels of complexity. These compete.

LB:

some usage of date and time is used in all documents—this should be in core (but core chapter is too big). Could do a whole chapter on “common features”

MH:

break core into a group of smaller modules like names, dates. Core could then pull these in

LB:

core should get you simple bibliography, dates and times, etc. (something like TEI Lite). Then you could pull in more sophisticated/advanced modules as needed (for dates only, for example)

JC:

if you have “basic dates” and “advanced dates” that’s even more confusing. Instead, wouldn’t be good to go to one page and get everything about dates and time

PB:

LB’s suggestion (above) might be an interesting way of maintaining TEI Lite

LB:

dating information is very relevant to name and place issues (e.g. birth, death…)

SR:

ODD dependency: if you’ve got places, you’ve also got dates (an old problem never solved)

LB:

if you pull out dates, why wouldn’t you also separate names, persons, places?

: JC will close the ticket with “will not fix” for now, and keep this in mind of P6

: JC will start a wiki page or ticket to discuss P6

: SY will send an email to tei-l to encourage people to add their docs to the samples page to keep open the statistical approach in the long term.

2.1.4 TEI Examples JC:

this refers to our use of examples, and gaining more consistency in how we use them, and doing a better job to really typify the element they are meant to exemplify

SR:

want a way of filtering examples that truly exemplify the element you are interested in, and those that just happen to have a particular element in them; We want a way of indexing when an example is (or happens to be) a good example of a particular element

MH:

examples could be sorted according to rank/usefulness

JC:

when people use the guidelines, the first thing they do is look at an example and just try to find one that applies to their situation and copy it

SR:

is it a better use of time to assess/rank our existing examples, or come up with new ones?

JC:

if each element had several examples, including some more complex ones, that would be good—if they don’t get too long

SR:

should there be a way of pointing to longer examples? Consensus is that fully worked examples would be beneficial.

MH:

examples are down at the bottom, so is length really an issue?

KH:

At some point do we decide that the Guidlines are not suitable for print/typesetting in PDF format

SR:

there’s already a mismatch between web and PDF: the “see all examples,” available on the web, no such thing in PDF

Main conflict here seems to be that to do this well, need TEI experts to assess—but this will be a long, monotonous task. Is it a good use of time?

JC:

There’s a disconnect between examples in chapters and examples in element spec

KH:

actually, the examples in the chapters tend to be complex edge cases, rather than good examples for teaching

LB:

principle of beginning with a simple example, and either adding more complex examples or adding complexity to the initial example

JC:

One possibility would be to keep all examples separate in a corpus, and then include them into the guidelines or reference page

LB:

this doens’t work where the example is part of the exposition of the guidelines

JC:

It could still work, just duplication of examples.

MH:

need a way of identifying and prioritizing the “canonical” examples—whether in the element spec or prose

SR:

start with the ones in the spec, which are by definition canonical, and then add ones from the text when appropriate

SR:

change the algorith that generates the “all examples” page to distinguish between canonical examples and “incidental mentions”

JC:

But what about cases that might be “canonical” for more than one element

LB:

how are we defining canonical?

MH:

Maybe it’s canonical if it is complete—that is, the last stage of a tutorial example, that has all of the complexity in it (not the stripped down version that you might see at the very start)

JC:

spec is meant to exemplify one particular element, examples in guidelines are meant to exemplify how you might encode a particular phenomenon. (so would involve good examples of using several different needed elements and attributes)

MH:

danger of included examples falling out of sync with prose

SR:

when you look at the reference page for p, pull in the relevant examples from chapter docs. This is what is done now, but we need a way to discriminate among the examples

LB:

how to decide whether to put an example in the spec or the chapter? Examples might appear in element spec, chapter text, and “list of all examples”

JC, MH: we want the list of all examples to be more useful—prioritized or sorted in some way

JC:

for many people, the element reference is their only view/use of the Guidelines

LB:

we don’t currently exemplify attributes as opposed to elements—we could/should do this

MH:

would this cause the attribute examples to show up in “list all examples” of whatever element the attribute is on?

SR:

could we have a third set of larger, more complex examples. Instead of canonical, think of significance: to look at all examples of p, is useless;

PS:

noisy examples can actually be useful in revealing what you can do if you need to

LB:

if an example is bad, we should get rid of it, not rank it lower or try to hide it in some way

RW:

is there some way to “reduce the noise” of examples by not displaying many examples of exactly the same usage? Problem is: this would be very difficult to do

MH:

if there’s a single example that excellently exemplifies more than one element, we only want to have to maintain this one in place, not repeat it in each element spec

MH:

one complaint: “I hit show all examples and there’s too much there to make sense of it”

JC:

people get here by just jumping right into the element reference, and scroll down to the examples, where there’s either one, overly simple example, or they click show all, there’s too much there to make sense of.

MH:

dense English prose is difficult for many people, so even if they click to the chapter (which they may not), they won’t read it

SR:

put the examples from the chapter at the top of the “see all examples” list—this might get us closer to relevant examples at the top. Consensus was this was worth doing.

JC:

should we increase the number of examples in the element spec

SR:

as more and more people are writing element specs, much of the “intelligence” goes into the element spec, and not into the text of the guidelines.

MH:

We don’t provide many examples for attribute classes. att.Ascribed has no examples, and where would he go to look? said has examples, but none of them has a @who on them.

LB:

every element that is a member of att.ascribed

: MH look at existing pages for attribute classes and see if there are any examples, try adding an example to att.ascribed class spec to see if it works

: Sebastian will tweak “show all example” so that at the top it lists examples from the appropriate chapter, and links to that chapter.

: MH to create a wiki page to coordinate adding examples to attribute classes. Since most attribute classes do not have examples, all of us should generate examples to be added to attribute class pages. Someone should coordinate this work? KH: create a wiki page where each class can be listed that is missing examples (or that needs more) and people can sign up

: change build process so that you can’t build without examples on attribute classes (this would have to be implemented later)

2.1.5 Group E: PS, SR, PB.

Reporting on tickets from grouping H

  • 3434990

    JC/PB: suggested that we change the prose to read “a lg contains “one or more L” (LB has already changed)

    PS: it could be the case that in a damaged source, lines are missing even if there’s an lg there

    : MH will write a Schematron rule requiring that lg contains l, lg, or gap

  • 3305016

    can you have two graphics?

    Kevin proposed a content model in Sourceforge, SR and group felt it was fine.

    : SR will implement the content model proposed by KH.

  • 3416130

    putting content inside empty elements might break processors? SR doesn’t think so

    LB:

    not persuaded by argument to change content model of : concern that allowing any content will open up a can of worms

    GB:

    adding certainty allows for using @match to use relative XPaths to point to a parent element

    SR:

    this would bring milestoneLike elements in line with gap, which does allow certainty

    What about allowing desc? GB sees this as different: a way of adding content to an empty element. SR says no, it’s a way of adding documentation to an empty element

    BB:

    what does this mean for editing the guidelines? Onus on us to describe how this might be used. This would be in the attribute class.

    MH:

    Would have to find every instance of, e.g., “ is an empty element’ and change in (and do this for every element in milestoneLike)

    SR:

    certainty is an empty element—so even adding it to an empty element, the element remains “morally” empty: it has no human readable content

    PB it is an overgeneralization to say that something milestoneLike must be an empty element. Rather, it is a *point*—not a span, division, or section. PS adds that it contains metadata about the text or document (not anything from the text stream). Also distinctions to be made between human-readable meaning of “empty” and schema requirement that there are no child elements. That is, to a reader, pb remains an empty element, even though it may have a child.

    GB:

    perceives adding an element like certainty as adding an “enhanced” attribute

    : GB will survey the guidelines to see what kind of changes this would involve. He will report back via email, unless there is great disagreement the FR should be approved and implement. If there is disagreement, address at our next teleconference

    : JC will update the ticket in sourceforge to assign to GB and add a note about our decided action

  • macro.paraContent out of place in gram etc [cf u discussion]

    link to ticket is broken, so group E did not discuss;

    : JC to find out what happened to this ticket.

2.1.6 TEI Bugs and FR

Group F: PS, KH, JC, GB

  • 3500566

    Didn’t like b because we can’t come up with a short definitive list of @type values.

    : KH to implement proposal ‘a’ on the ticket: add new example to idno elementSpec

  • 3497079

    GB uncomfortable with not using biblScope in a bibliography (as opposed to in a footnote)

    Proposal seems not to deal with difference between part and whole of article (analytic item) and part and whole of thing that article is in (monographic item).

    : GB to write a comment on the ticket giving examples of citations that would be problematic or confusing according to this proposal. Paul can also provide example citations from his U-M CV project. Kevin will discuss with Gabby and incorporate marked-up examples into proposal and revise proposal if appropriate. Leave ticket open for now

  • 3440977
  • 2493417

    : Agreed. Already agreed and implemented, KH to add more examples/prose

  • 3480057

    : KH to close.

  • 3439587

    : MH will investigate adding distributor and publisher to model.respLike, noting that last time he tried to do something like this, it got messy.

  • 3513147

    : we found this recently opened and yet related ticket but didn’t have time to look at it

2.2 Afternoon

2.2.1 TEI Bugs and FR

Group G: LB, RW, PB

  • 3496494

    Proposal is to create a class for from and to so that these are defined and used consistently in all cases where they are defined as data.pointer. Currently they are described inconsistently. arc should not be a member because the semantics are slightly different. biblScope should not be a member because @from and @to are used totally differently—not as pointers, but to express start and end points of a citation. The explanation of the usage of @from on span seems problematic (for example, there is reference to “the end of the passage,” without explaining what is meant by “passage”. However, the usage of @to and @from is slightly different for each of these elements. So ticket will be closed once language is tidied up by RW.

    : RW will adapt defintion of span to take better wording from 17.3

  • 3495987

    MH’s case for what we think is described here: You have an orgList and want to transclude the name of each org elsewhere in the document. You could do this using @decls. But JC said that you would instead do this using @ref or a similar from att.canonical.

    : LB will ask Laurent for examples in case we are assuming incorrectly and will look at TEI-L archives.

  • 2834511

    If this was implemented, you would need both a way of rendering whatever the span wants to do only on the text/content, or also on the markup (e.g., if a paragraph ends and another begins within the span, is the rendering applied to

    as well as the content? Call for reassessment of this ticket. Nothing has happened to it for three years, and meanwhile there has been other development of genetic editing work.

    : LB will add proposed Schematron rules but not add any other elements without clear use cases.

  • 3291540

    According to ticket, because att.editLike gives you att.dimensions and att.ranging, lots of elements end up getting unneeded attributes. Ticket proposes removing certain attributes from certain elements.

    We all agree with the proposal.

    : SR will take the ticket. We’ll leave it to SR to decide which of the second list of elements should be included as well. Sebastian will send a list to TEI-L so people could comment on any elements where it would be removed. If, in the future, we might decide we “got it wrong” for some of these, so we could change our mind if needed, but better if we don’t need to.

  • 3284816

    While some elements in the list already have the attributes requested, others don’t. We accepted. SR will do any that don’t have them already.

    : SR to implement.

Group J: MH, BB, SR

  • 3415801

    This a valid thing to do, and we should implement it. We need to go through the Guidelines and find places where a) an attribute is defined locally simply because it has a valList or its own desc, and could go back to the att class, or b) the constraint is expressed only in prose.

    : SR will implement.

  • 3118435

    We cannot do this now, because it would stop us using DTDs. It is desirable, but is a P6/ ODD3 feature. it should be recorded on P6 wiki page, and discussed at the ODD3 meeting in Hamburg. Noted that it would mean revisiting a lot of content models to see whether interleave was what was intended. Also not clear how interleave in XSD is supported.

    : LB will close ticket.

  • 3426828

    : SR to close this ticket noting that requestor hasn’t responded.

  • 1650195

    We can recommend use of http://www.w3.org/TR/xml-model/ in order to reference a schema. Separately, if you want to embed a schema, use the new schemaSpec (which will be added to model.encodingDescPart)

    : MH to implement adding schemaSpec to model.encodingDescPart.

2.2.2 TEI FR: Floating Div

A wide ranging discussion took place concerning the floating div proposal which council had reviewed. PS presented many examples of cases where it might be useful.

Do we want to:

  • Loosen definition floatingText to allow anything besides a full text?
  • Create new element for floatingDiv for anything besides a full text?
  • Tell them that what they think is a “floating div” is actually a floatingText, spGrp, or a div that hasn’t been properly identified?
  • Remove the word ‘whole’ from element description of body?

Consensus emerged that many of the PS proposed use cases were solved with spGrp; and the rest with floatingText as more loosely defined. It was difficult to categorise a way to distinguish floatingText from proposed floatingDiv; more information and examples are needed.

: JC will reply to the message on TEI-L: In light of the current text and examples in the Guidelines on floatingText and spGrp, we’d like examples of proposed “floating divs” and how they would differ from floatingTexts.

2.2.3 TEI ODD Meeting

Will be a panel at DH2012 discussing problems with ODD and future of ODD. TEI-C is funding a post-conference meeting (July 21-22) for room and some travel expenses to continue the discussion (about how to develop ODD3, which might happen before or with P6).

JC invited a number of people to attend and encouraged them to suggest others to invite. Approximately 15 attendees who will have expenses covered. JC will send invitaiton to TEI-L to allow others to attend at their own expense.

JC asked us what we would like the people at the meeting to address. Should they theorize about what ODD3 would look like if starting from scratch? Or offer concrete proposals for improvements to ODD?

We’d like to avoid just a list of complaints about how things currently work.

SR:

do we want only things that are backwards-compatible? Are we open to requests for things like tools for processing ODDs that don’t require changing ODD itself?

JC:

anything that is backwards-compatible could become a feature request in SourceForge now.

PB:

perhaps we can ask them to propose solutions to particular problems we’ve identified?

MH:

solicit topics in advance?

JC pointed us to a list in the wiki which lists problems in ODD.

LB said he has started but not finished a proposal for resolving the Durand Conundrum.

BB:

we might be asked to produce training and support materials for ODDs.; What other systems do people use for maintaining XML standards?

KH:

NLM/JATS maintains a homegrown system developed by Mulberry Technologies.

A list of other people, not yet invited, was suggested by Council to make sure were targetted with the public announcement. [JC has note of all individuals, but seems sensible not to minute them here.]

SR:

worries that everyone at the meeting will agree with each other. We need a skeptic [SR suggested some].

JC:

We could ask others (Balisage folks, literate programming folks) after the meeting to comment on any proposals.

LB:

Kevin Hawkins should invite Kevin Reiss.

SR:

what we’re really looking for is prose to put into a funding application for what we want to develop. Need to list problems that need to be solved and the likely direction for the solution.

JC:

please add additional suggestions to http://wiki.tei-c.org/index.php/ODD-dev .

2.2.4 TEI Release Process JC:

We need to make sure that all the steps are included and everything is explained for GB who is next release technician.

MH:

There are problems in the stylesheets release process that depend on SR, but JC, MH and SR are fixing that process.

SR:

GB doesn’t have access to a Linux machine, so that will be a problem/benefit in testing.

MH:

Also need to make sure the process will work with Ubuntu Precise.

SR:

A lot of documentation needs to be written about the testing process. Are there tests that we should be running but aren’t?

Actually, our priority should be rewrite of Roma. If a pending Oxford JISC grant doesn’t come through, we should shop our application around to other granting agencies.

Wednesday April 18th 2012

3.1 Morning

3.1.1 TEI Outreach and Education JC:

It has been suggested that the Technical Council should do more education/outreach activities, such as a newletter to go to sponsors reporting on technical and other updates

JC suggests that the job of the Council is primarily the development and maintenance of the guidelines. Outreach is good, we’re all in favor, and while it is part of our mission, it’s not the main focus/mission of the Technical Council.

SR:

we have a “moral” responsibility to step up if asked to, say, write about the current state of ODD; Consensus was that we would participating as individuals is a good idea.

LB:

why did we fail to deliver on the tutorial last time we were asked? lack of resources, etc.

GB:

the way to get the council to do something like this would be to make it a council duty. The question is, do we think this is high enough on our priorities to make it a duty? (The way to get it on the agenda is to put it on the agenda)

SR:

this would be at the expense of something else.

BW and JC:

this is a good thing to do, but many people can do it. The development work done by the Council can’t be done by someone else.

SR:

points out that there is some gray area: Council *does* spend time trying to make the examples and other reference materials more sensible and accessible

MH:

maybe Council should be focusing on developing the relationship with with the intermediate user

SR:

we’re responsive to tickets, but not proactively reaching out

JC:

sees marketing and outreach as primarily board activity; the kind of education/outreach that is being suggested seems closer to a Board activity. A report from Council and report from the Board could be included in each issue of the TEI Journal, but I wonder if that is really appropriate. KH thinks this is possible. BB is interested participating in outreach activity

SR:

No one can make the guidelines better without the involvement of the Technical Council, so there’s a level on which we must engage in some level of outreach/engagement

PS:

suggests training is most useful in the context of a specific project, and that abstract training is not so useful or desirable; MH and JC disagree

JC:

there’s a gap in knowledge/training at the intermediate level—that’s where it gets hard because of the specificity of any given project. It’s straightforward to teach intro level, and harder things like Schematron and ODD.

KH reviewed bylaws: duties do include development of training materials and tutorials, or to make recommendations to the board about such things. JC would be happy to make recommendations Council developed.

SR:

we develop and maintain educational tools like Roma which can be used as a training tool.

General consensus: The technical council is entirely open to some degree of involvement in outreach and education activities but it generally views these as a board-led activities. We believe such activities are worthwhile, but don’t have the resources to drive any such effort and don’t feel it is or should be the core activity of the council. We are more than happy to contribute to such activities to some degree, especially on technical aspects. Individual members of the council may also, of course, participate separately as members of the TEI community if they have time available.

3.1.2 Creating small working groups:

  • The dreaded signed/salute/closer/trailer/sigBlock ‘Div-Liminal’:

    JC proposes an ad-hoc council working group for this. LB suggests the remit be a “div-liminal working group” (covering divTop and divBottom). Lou will convene, Paul asked to participate, will work on email and produce a report for the Council before next FTF

    KH suggests working on this this afternoon to produce a new report; JC could the proposed group have an initial conversation about remit this afternoon;

    LB:

    remit is to review all relevant elements and classes, determine whether they are defined properly and whether their placement in the content model is correct

    PS:

    has a sense of a proposal that will preserve much ambiguity where necessary

    Working group will be: PS, LB, BB, KH (with advice from rest of council where necessary).

    : LB to chair working group will report back by next teleconference, concrete proposal by next face to face meeting with status reports at teleconferences.

  • Audio/Video Facsimiles:

    SR will head up the council’s contribution to the working group. LB and MH also interested

    PS:

    Need to know more about existing practices and competing standards from outside experts/users

    LB:

    Thomas Schmidt has looked most at how to use TEI for encoding speech

    PB:

    suggests Andreas Witt and people from Edinburgh

    LB:

    Linguists tend to be interested in this problem. Sociologists, folklorists, local historians, may also may be good sources

    KH:

    Faith Lawrence also suggested by GB

    PB:

    start with a firmly known problem (e.g., what linguists need), and expand from there.

    : Working group to be chaired by SR, with LB, MH, and possibly PB; to approach TEI-L and others doing work in this area to get outside expertise; an initial proposal or detailed report of progress at next FTF with status reports at teleconferences.

  • TEI and URIs @target/@cRef/@ref/@key: MH:

    Some way to declare what you mean by your private URI in the header. There should be a documented way to do this, that is machine-readable; Two separate issues here: deprecation of @key and documentation of private URI scheme. We might agree on the latter even though there is likely disagreement on the former. JC would like to see both handled together.

    MH:

    if we are suggesting that people use private URIs instead of magic tokens, because they can be validated, we ought to offer a place to explain the URI scheme; cRefPattern is the model of what MH would like to do here

    SR suggests that GB has knowledge of the classical world where this is particularly of interest; KH proposes that MH write up what he has in mind, then shop it around for reactions/suggestions

    : MH will chair this going to be working group, bring in GB, SY, and outside people suggested by GB. They will develop concrete proposals for next FTF meeting with status reports at teleconferences.

3.1.3 Any Other Technical Council Business

  • Autumn meeting will likely be in Oxford, perhaps before or after the EEBO-TCP conference (September 17-18)
  • SR:

    what else is lurking that we need to prepare to address/work on for Autumn? Is there something we should be aware of that has work being done on it?

    GB:

    there was a working group to re-write critical apparatus chapter but no word on progress here; JC: a mailing list was set up with some conversation, but didn’t progress further; Particularly hard to fix—some suggestions are on a wiki page

    GB thinks the chapter may be OK as it is; LB: we could go through the proposals on the wiki page ourselves. LB: proposes asking Marjorie for a report on the status of this work

    : James will prod the critical apparatus working group before next teleconference.

  • BB:

    Has there ever been a TEI SIG for periodicals (capturing existing)? MH says there is a scholarly publishing SIG, but this is not the same as a working group for capturing newspapers, for example

    JC:

    if such a group produced a best practices document or similar, that would be welcome; MH is working on something with 19th century journals. MH and BB could start a SIG.

  • Why does the stylesheet build keep breaking? MH:

    want it to be the case that the stylesheets are built first and checked, and then used in the rest of the build (this is not the case now—the reverse); want it to be the case that a change in the stylesheet build will trigger a build of everything

    SR:

    this is all part of the process making the build and release process more transparent and better documented. Aim is to get away from the fact that Sebastian is the only one who can make stylesheet changes take effect in the whole guidelines

  • Council Meeting preparations:

    PB noted that in Paris, JC assigned an action to email SIG lists for a report before Council meetings, but this didn’t happen; JC apologised and will rectify this before next teleconference.

    PB suggests a checklist or similar for Council meetings preparations to make sure everything is done, including asking SIG conveners for report; JC agrees.

    : JC to produce a Council meeting/telco checklist on TEI-C Wiki

  • Report from LingSIG (by PB):

    developing bibliography in Zotero (tei-lingsig). Meant to engage the linguistics community even outside of TEI. Currently 153 items.

    Special issue of jTEI guest-edited by conveners (including article about the bibliography)

    PB will talk at ILIT (Eastern Michigan University) on TEI for Linguists; Opened a project on SourceForge (beta)—hopes to make it easy for people to move around and experiment in this space, create a space/sense of belonging for members, that is more cohesive than just being on mailing list

  • Report on TEI and linguistic data analysis tools (such as UIMA) (by SY)

    SY wants to bridge the gap between the output of UIMA and GATE and TEI; UIMA throws out everything that isn’t a string of words in the document—metadata, comments, etc. They use XML, but rely on multiple levels of stand-off/byte-offset markup—good for applying multiple kinds of POS tagging and comparing differences; not so easy to roundtrip back to TEI; SY has plans to develop a way to do this and bring it back to council for review; LB: recommends existing literature on stand-off markup—how does this project fit in? PB recommends papers he wrote with a colleague, recommending almost purely TEI conformant approach for stand-off analysis of text, that handles precisely this issue of multiple kinds of linguistic tagging. e.g. See LingSig bibliography

    : SY will follow up offline w/ LB, PB

    : PB will put SY in touch with others working in UIMA and interested in converting to TEI

    : PB and SYreport back to Council on progress by next FTF

3.1.4 Bugs and FRs: any remaining

  • 3454814

    There are huge numbers of comments sprinkled through the Guidelines, dating back to 1992 or before. While many of these are historically interesting, most are irrelevant. Cleaning them out would be beneficial, but this must be done carefully. In particular:

    • Comments inside egXML must be preserved (they are displayed).
    • Commented biblio items can be removed after checking that there is a bibl or biblStruct for them.
    • Informative/useful comments should be turned into notes.
    • Some comments (e.g. “<!– Better example needed –>”) are useful and may remain.

    When new comments are added (and especially when we commenting out blocks of existing code because we’re nervous about deleting them), they should be initialled and dated. Old code should be deleted after a week.

    : LB will write up guidelines for purging comments and distribute to TEI-Council list, including describing first passes to deal with low-hanging fruit and suggested format for how we should comment going forward

    : from there, work of reviewing and purging comments or turning them into notes will be divided among council members

    : RW will incorporate language from LB’s guidelines into the style guide when appropriate

  • 3454803

    : MH will compare Chapter CH against the latest version of BCP 47 for inconsistencies

  • 3432216

    There should be a list of translators and contact information for them so we can reach them with translations need to be reviewed (KH reports that this does exist in FM1, but it’s not up to date) GB: as a practice, he has been going ahead and updating glosses in any language that he knows, as well as English. SR: if you make a confident revision in another language, update the date of the translation?

    SR:

    the person who did the translation once, may not have ongoing responsibility for the translation

    LB:

    if the translator is credited, others should not edit the translation without that person’s involvement? Disagreement about whether our credit of translators constitutes acknowledgment of a one-time translation task our ongoing editorial responsibility for translations

    Agreed Policy: if you feel very comfortable about updating an element spec translation, proceed with caution. Do not change the date in gloss version=””. This way, the translator will see the change when they review again.

    : SR: double check FM1 to see if list of translators is up to date

    : KH will add this policy to the “how to edit the guidelines” working paper

  • 3223636 MH:

    when investigating how xml:space is supposed to work, it became clear that there are contradictory explanations in the spec

    MH:

    because processors handle xml:space differently, it is difficult to describe how this should be used;

    SR:

    Ask Oxygen and Saxon teams to make sure xml:space is being treated consistently with (their interpretation of) the W3C spec?

    MH:

    people keep losing white space where they need to keep it. xml:space should be a solution for this, but it isn’t.

    : Expand our explanation of xml:space in the Guidelines:

    • in note: expand to acknowledge that many parsers may not handle xml:space properly
    • clarify definition of xml:space’s “default” value to match the XML spec (that is, “whatever the processor would normally do)
    • remove sentence in prose of guidelines in chapter 22 “the global @xml:space is available to indicate that white space should be preserved within the content of these elements”
  • 3393781 SR:

    @version was intended to link updates to translations to releases of the English guidelines. Indicates the date of the state of the English guidelines when this translation was updated.

    Leave @version as it is; change the attribute that we are using to indicate revisions to translations to @versionDate

: JC to add TEI Sourceforge site to the TEI category so all TEI sourceforge projects can be located at once

3.2 Afternoon

3.2.1 Hacking for those working on things who don’t need to catch flights right away.

Div-liminal group (minus BB, plus SR) began brainstorming:

  • LB defined the scope of the group to include assessment of elements in model.divWrapper, model.divTopPart, and model.divBottomPart (and their children).
  • In looking at the content models of opener and closer, he suggested that new model classes be created for these: model.openerPart and model.closerPart.
  • PS said he had been pondering the problem and think it’s best solved by the following:

    • Create a new signatory element for use of only designations of signatories (names and phrases like “the late sergeant” that go with names). It would have the content model that signed currently has.
    • The current signed element would have an expanded content model including salute and list.
    • The salute element would be left just for “polite phrases” that tend to occur before a signatory.
    • For example, this example in the Guidelines: I have the honour to remain, dear sir, your obedient servant RICHARD CUFF (late sergeant in the Detective Force, Scotland Yard, London). would become: I have the honour to remain, dear sir, your obedient servant RICHARD CUFF (late sergeant in the Detective Force, Scotland Yard, London).
    • Another way to summarize his proposal is that he proposes to create a new “super signed” or “signed block” (the new signed) and turn the old signed into a signatory.
  • Implications of PS’s proposal:
    • Everybody who has a signed containing only text or phrase-level elements could (but need not) change this signed to signatory.
    • Optionally, they could (but need not) choose to wrap the new signatory and any salute in signed.
  • KH proposed an addendum to add dateline to the new signed as well. PS didn’t like this because then there would be essentially no difference between signed and closer. KH said he thinks of signed as being equivalent to closer type=”signed”.
  • Lou noted that under KH’s addendum we would have to remove “as a kind of heading or trailer” from the definition of dateline in order to turn this into a purely semantic element.
  • KH and SR debated whether dateline is structural or semantic. SR said it’s easier to write stylesheets if it can be assumed to be structural and block-level, but KH said he thought of it as semantic and phrase-level.

JC Thanked all those remaining Council members for participating.