teic / tei-simple Goto Github PK

I could understand Simple not having any of these three elements, but to have two and not the third seems strange; is there a reason for it? I typically use all three when documenting encoding practices.

Should you be able to place a <modelSequence> inside a <modelGrp>?

Currently there is an example of this:

https://github.com/TEIC/TEI-Simple/blob/master/tei-pm.odd#L656

However the specification of a <modelSequence> says it should only contain a number of <model>s:

https://github.com/TEIC/TEI-Simple/blob/master/tei-pm.odd#L553

The current ODD file modifies att.global to remove @Style and @rend and customize @rendition. Those attributes will be in att.global.rendition rather than att.global in the next release. Just a warning that things will break when that happens.

Indented first lines in paras and other blocks are common in all early modern texts. The current simplePrint spec doesn't provide anything for this. I propose simple:indentFirst, or failing that, from CSS, simple:textIndent.

There's no specification on how the behaviour attribute's value should parsed. How should strings, URIs and XPath expressions should be quoted.

Unless this is a technical term from typography that I don't know, simple:letterspace is a bit confusing; it doesn't say what the size of the letter spacing is (wider than normal, narrower than normal?). Isn't it a bit like saying "font-size" without providing a size? If it's analogous to CSS letter-spacing, it needs a value of some kind, surely?

An implementation question:

As I was reviewing @wolfgangmm's tei-simple-pm app, I noticed the documentation.xml file used the <code> element, which the TEI-Simple docs say is...:

not allowed in TEI SImple, but should be transformed to another element

... namely, <hi>. I found an appropriate @rendition value: simple:typewriter, and could wrap blocks in <ab> (possibly with a CSS @rend override of white-space: pre to ensure preservation of line breaks? or is @rend not allowed?). We'd lose the possibility of language- or vocabulary-specific syntax highlighting that we had with the old <code lang="xquery">, etc., but at least we'd have standard TEI-Simple. Then I noticed that the TEI-Simple-PM docs do use <code>, so I started to wonder if <code> really isn't intended to be allowed in TEI-Simple?

While project documentation may not be the primary case for TEI-simple, it occurred to me that the question might be worth raising, because it gets at a question I had: what's the best route for a project that wants to extend or customize TEI Simple?

Functions often take an xpath expression and in the examples it's
often the current node, but what should happen if the xpath expression
returns multiple nodes?

<model behaviour="block(.//)"/>

Here we're returning all nodes below the current node as asking them to be formatted as block(wrapped with a <div>)

currently behaviour="note" needs a label param explicitly specified in the <model>, otherwise invalid output xslt is produced, should there be a default and, if so, what the default should be?

This in teisimple.odd is a bit puzzling:

valItem ident="display"
desc: formatted like a quotation

What does "formatted like a quotation" mean, and why would a value of "display" map to this?

Unless I'm missing something obvious, this is subtly cruel.

to specify persistent and human/designer-friendly class names to be passed through to the output

How should a function behave when its XPath expression doesn't return anything. If it's a function that outputs some content (block(../element)) should the output container still be written? I'm guessing not, but it's not explicit in the specification.

Sorry if this is a silly question, but is there a TEI Simple DTD file somewhere? I notice it's referenced in tests/testsimple.xml, but I don't see it in the parent directory.

simple:left, simple:right and simple:centre are included, but there's no simple:justify or simple:justified. This does exist -- see, for instance, the italicized block at the bottom of the first column on this page:

http://internetshakespeare.uvic.ca/Library/facsimile/book/SLNSW_F1/567/?zoom=850

so I think it should be included.

Your expanded list of values in att.placement is very helpful: this was one of the things I found particularly unclear when I was starting out with TEI.

One thing I don’t see there is an equivalent to EpiDoc’s ‘overstrike’, indicating that something has been written or carved over whatever was there before (which may or may not be decipherable). I assume that ‘inline’ is not referring to this, which I would take as indicating, for example, that a character had been squeezed between two others. I'm not sure what this should be called; 'superimposed' is one example in the Guidelines that isn't specific to a particular medium.

It could also be helpful to have a 'centre' or 'textblock' value (though there might be a better solution that I am missing here). For instance, if one is using this list to roughly record the position of an annotation on a page, what would one do if it was placed somewhere in the middle of the page (e.g. on a page with relatively little text, such as a title page or verso)?

I also wonder about the definition of 'margin' as 'in the outer margin', whereas the Guidelines give it as 'in the margin (left, right, or both)'. The TEI Simple additions pinpoint the ambiguity precisely, but on the other hand I can see them leading to inconsistency, with some people using 'margin-left' and 'margin-right' and others 'margin' – and then what if you wanted to say that something was in the inner margin?

For a long time, we have had the notion that e.g. <orgName> is syntactic sugar for <name type="org"> and we even have an example of @type="org" in the Guidelines:

http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-name.html

Why does Simple adopt the unconventional <name type="organization">?

<name type="placeGeog"> is even weirder, alongside <name type="place">. Aren't all geographical names places? Why not <name type="geog">?

Currently the specification refers to XPath expressions but doesn't elaborate as to which version of XPath should be used.

The constraintSpec below is located in the classSpec which modifies att.global.rendition:

<constraintSpec ident="corresppointer" scheme="isoschematron">
            <constraint>
              <rule xmlns="http://purl.oclc.org/dsdl/schematron" context="tei:*[@corresp]">
                <let name="results"
                  value="for $t in        tokenize(normalize-space(@corresp),'\s+') return starts-with($t,'#') and not(id(substring($t,2)))"/>
                <report test="some $x in $results  satisfies $x"> Error: Every local pointer in "<value-of
                  select="@corresp"/>" must point to an ID in this document (<value-of select="$results"
                  />)</report>
              </rule>
            </constraint>

However, @corresp is defined in att.global.linking, not att.global.rendition. Unless I'm misunderstanding something, it should be moved.

This is in the file teisimple.odd.

Hanging indents are not uncommon in Early Modern texts -- for instance, all the chapter headings in Dekker's Guls Horne-Booke are laid out with hanging indents:

http://eebo.chadwyck.com/fetchimage?vid=5530&page=8&width=1200

I'd suggest simple:hanging as an appropriate ident.

teisimple.odd now includes a file called headerelements.xml, but this is missing somehow.

xmllint --valid teisimple.dtd tests/testsimple2.xml # works fine
xmllint --valid --relaxng teisimple.rng tests/testsimple2.xml # sits there like a pudding
jing teisimple.rng tests/testsimple2.xml # works fine
rnv teisimple.rnc tests/testsimple2.xml # works fine

The <rendition> tag can have a scope attached to it, but there is no example of this.

The specification says that a <modelGrp> can have its own <rendition>:

https://github.com/TEIC/TEI-Simple/blob/master/tei-pm.odd#L602

but there is no example or elaboration about how this should be handled.

The <rendition> tag can have a scheme attribute:

http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-rendition.html
http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-att.styleDef.html

Do implementors of the specification need to worry about this attribute?

It is possible in HTML to specify @type for <ol> to specify the numbering type, as follows:

<ol type="1">
<li>Test</li>
<li>Test</li>
<li>Test</li>
</ol>

I'm assuming that one would need to convert this into TEI using @rend on <list>. For consistency, it might be useful to have preset attributes corresponding to these in TEI Simple.

@behaviour="glyph":

"glyph (content) show a character by looking up reference"

Wouldn't this function need a URI for looking up the reference? Or if the content is the URI, I think the param name is a bit misleading.

(from Martin Holmes)

In the regular Guidelines, we've settled on -ize, but the word "visualisations" appears in the Simple ODD. If the same standards are intended to apply, this should be "visualizations".