[tei-council] report on rng:text attribute sweep

[Syd Bauman]

Here's the list of changes made in my recent sweep to root out the
remaining rng:text attributes. 
* Names of classes are delimited by "%" and ";" even though they are
  no longer expressed as parameter entities, just because I don't have
  a better way to make them stand out.
* Names of elements, attributes, and classes are as they were at the
  time; some of them have changed since.

--------- altered ---------
indexName=   of <index>               -> data.ident
width=       of <binaryObject>        -> data.htmlDimension /* also of <graphic> */
height=      of <binaryObject>        -> data.htmlDimension /* also of <graphic> */
scale=       of <binaryObject>        -> data.numeric
type=        of %att.entryLike;       -> data.enumerated
level=       of <sense>               -> data.numeric [1]
type=        of <form>                -> data.enumerated
extent=      of <orth>                -> data.enumerated
extent=      of <pron>                -> data.enumerated
type=        of <gram>                -> data.enumerated
type=        of <itype>               -> data.enumerated; also changed GI to <iType>
type=        of <colloc>              -> data.name /* EDW90 says enumerated [2] */
type=        of <usg>                 -> data.enumerated
type=        of <lbl>                 -> data.name /* EDW90 says enumerated [2] */
type=        of <re>                  -> data.name /* EDW90 says enumerated [2] */
type=        of <oRef>                -> data.enumerated
type=        of <oVar>                -> data.enumerated
placeAttrib= of <origPlace>           -> data.enumerated
evidence=    of <origPlace>           -> data.enumerated
unit=        of %att.measured;        -> data.enumerated
scope=       of %att.measured;        -> data.enumerated
certainty=   of %model.pPart.datable; -> data.enumerated
dateAttrib=  of %model.pPart.datable; -> data.enumerated
evidence=    of %model.pPart.datable; -> data.enumerated
notAfter=    of %model.pPart.datable; -> data.temporal
notBefore=   of %model.pPart.datable; -> data.temporal
interval=    of <when>                -> specialized; also interval= of <timeline> [3]
type=        of %att.pointing;        -> data.name  /* EDW90 says enumerated [2] */
lemma=       of <w>                   -> data.name  /* must become element, eh? */
baseForm=    of <m>                   -> data.name  /* must become element, eh? */
type=        of %att.textCritical;    -> data.enumerated
cause=       of %att.textCritical;    -> data.enumerated
key=         of %att.personal;        -> DELETED
key=         of %att.datePart;        -> DELETED [4]
type=        of %att.datePart;        -> data.name
mode=        of %att.identified;      -> data.enumerated
mode=        of <valDesc>             -> data.enumerated
mode=        of <valList>             -> data.enumerated
key=         of <attRef>              -> data.ident
type=        of <schemaSpec>          -> data.name /* EDW90 says enumerated [2] */
ns=          of <schemaSpec>          -> data.namespace /* also ns= of <attDef> & <elementSpec> */

--------- stet ---------
n=           of %att.global; /* data.names problematic */
rend=        of %att.global;
key=         of %att.naming;
pat=         of <fragmentPattern> /* now replacementPattern= of <cRefPattern> */
delim=       of <state>
value=       of <metSym>
mimetype=    of <equiv>
sortKey=     of <term>
mimeType=    of <graphic>
mimeType=    of <binaryObject>
encoding=    of <binaryObject>
key=         of %att.entryLike; -- changed to sortKey=
expand=      of %att.lexicographic;  /* must become element, eh? */
norm=        of %att.lexicographic;  /* must become element, eh? */
split=       of %att.lexicographic;  /* must become element, eh? */
value=       of %att.lexicographic;  /* must become element, eh? */
orig=        of %att.lexicographic;  /* must become element, eh? */
reg=         of <orgName> /* needs to be handled by different mechanism */
reg=         of <orgTitle>/* needs to be handled by different mechanism */
reg=         of <orgType> /* needs to be handled by different mechanism */
reg=         of <orgDivn> /* needs to be handled by different mechanism */
label=       of <graph>   /* slated to become child somehow; data.names problematic */
label=       of <node>    /* slated to become child somehow; data.names problematic */
label2=      of <node>    /* slated to become child somehow; data.names problematic */
label=       of <arc>     /* slated to become child somehow; data.names problematic */
label2=      of <arc>     /* slated to become child somehow; data.names problematic */
label=       of <tree>    /* slated to become child somehow; data.names problematic */
label=       of <root>    /* slated to become child somehow; data.names problematic */
label=       of <iNode>   /* slated to become child somehow; data.names problematic */
label=       of <leaf>    /* slated to become child somehow; data.names problematic */
label=       of <eTree>   /* slated to become child somehow; data.names problematic */
label=       of <triangle>/* slated to become child somehow; data.names problematic */
label=       of <eLeaf>   /* slated to become child somehow; data.names problematic */

Notes
-----
[1] I still think this is a bad idea. The data.numeric type permits
    negative values, fractions, and numbers way outside the bounds of
    possibility here. xsd:unsignedByte is much closer to what we want
    (and is a straight W3C type, to boot): an integer between 0 and
    255 (inclusive). Could also use xsd:nonNegativeInteger {
    maxInclusive="whatever-number-we-agree-on" }

[2] The problem here is that we have not yet really hammered down the
    relationship between the use of the data.enumerated type and the
    existence of a <valList>. It has been suggested (I think by
    several different people, myself & Lou included) that the
    relationship should be iff -- i.e., if there is a <valList>, then
    datatype must be data.enumerated; if the datatype is
    data.enumerated, then there must be a <valList>. This system has
    its appeal, but also has problems (which I hope to address
    later). For these attributes, I have recommended in EDW90 that
    the type be data.enumerated, but no one has come up with a
    suggested value list. Thus I have made the datatype data.name
    instead, at least for now. In retrospect, probably should have
    been data.ident, but I'm hoping we come up with value lists
    instead, anyway.

[3] The attribute type is essentially that which I suggested on this
    list almost a month ago (2005-09-20), and which I think Lou
    doesn't like. But no one has suggested anything better. It is
    currently 
       xsd:float { minExclusive = "0" } | "unknown"
    for <when>, and
       xsd:float { minExclusive = "0" } | "regular" | "irregular"
    for <timeline>. This mimics P4 pretty well. I, for one, would be
    happy if we came up with a better method altogether.

[4] Also fixed value= which had been a choice of lots of xsd, and is
    now data.temporal