W3C home > Mailing lists > Public > w3c-wai-hc@w3.org > October to December 1997

a TABLEs strawman

From: Al Gilman <asgilman@access.digex.net>
Date: Fri, 10 Oct 1997 14:53:57 -0400 (EDT)
Message-Id: <199710101853.OAA12614@access2.digex.net>
To: w3c-wai-hc@w3.org (HC team)
X-URL: http://www.access.digex.net/~asgilman/web-access/TABLE/straw1.html

                            Language Implementation
                                       
   This note outlines one possible approach to reforming HTML4 TABLEs to
   permit essential semantics to be conveyed with modest additional text.
   This is just an outline of the language provisions.
   
CLASS attribute

  on TABLE
  
   Reserve the following values
     * field -- There is one output dimension which is the same for cell
       values throughout the body of the table; both row and column
       indices are used to convey different input conditions.
     * record-list -- Each row is an entity, each column is an
       independent dimension unless grouped by COLGROUP. Row membership
       is stronger than column membership.
     * column-group -- default reading order is by columns. Newspaper
       format is an example of this class. Column membership is stronger
       than row membership.
     * array -- row and column identies have meaningful semantic
       connotations. This class is a superclass of field and record-list.
       The row and column membership are equally important in the absence
       of further information.
     * list -- table displays a semantic list in a graphically convenient
       folded form. Calendar pages are of this class. A week is a
       subsequence of the sequence of days, a column of week-rows is a
       folded display of the list of days.
     * set -- the tabulated items are a semantic entity as a set and lack
       other relational properties between them.
     * display -- the table structure is purely a visual framework.
     * other -- there is more semantic basis to the array locations of
       the various cells than in the display class, but it does not fit
       any of the stronger predefined patterns.
       
SUMMARY attribute

   In prose description, distinguish the cases of a table with a caption
   and a table without. If there is no caption, the summary describes the
   general plan or function of the table. If there is a caption, the
   SUMMARY is an optional expansion on the caption appropriate to be
   verbalized following the CAPTION.
   
ABBR attribute

   Extend to be applicable to TD as well as TH
   
AXES attribute

   Remove in favor of TYPE.
   
TYPE attribute

   Supercedes AXES. The heuristic expansion on the attribute name which
   conveys the attribute semantics is of-TYPE.
   
   Value is ID-LIST still. No formal constraints on what entity types may
   be in this list, or where they may be in the document. In the prose
   description, however say that the contents of the listed entities
   describe a general kind of value of which the contents of the current
   cell give an example.
   
   In the absence of an explicit TYPE attribute declaration on a TD or TH
   entity, one should assume the TH which is the closest TH spanning all
   columns that the current cell spans and appearing in the table above
   the current cell. If there is no such entity then this attribute is
   assumed empty.
   
CONDS attribute

   Supercedes KEY and COORDS discussed earlier.
   
   Add new attribute. Usable in TD entities only.
   
   The heuristic expansion on the attribute name which conveys the
   semantics of this attribute is under-CONDitionS.
   
   Value is ID-LIST. No formal constraint on the element types
   represented in this list. In the prose description, define this
   attribute to hold a list of IDs for entities which describe the
   specific conditions under which the contents of the current cell will
   be observed (are valid). In the absence of an explicit CONDS
   attribute, one should assume a value which is the IDs of all TD and TH
   elements in same row which have inherited THROUGHOUT=KEY.
   
THROUGHOUT attribute

   Supercedes and extends AXIS.
   
   Specializes and complements CLASS.
   
   Heuristic spelling is properties-applicable-THROUGHOUT-me.
   
   Usable at least on TABLE and sub-TABLE containers COLGROUP, TROW, etc.
   Consider extension to list and document hierarchy contexts such as
   DIV, too. Heredity from TH cells is defined below so that THROUGHOUT
   can be used on TH and TD as well. This may be considered to have a
   lower urgency that having this capability on the structural containers
   including TSPAN.
   
   Takes same type of values as CLASS (user defined terms with some
   predefined)
   
   THROUGHOUT values are implicitly inherited to all elements
   hierarchically within the declaring entity. Note that if TSPAN is
   accepted, that hierarchical inclusion and textual inclusion are not
   identical.
   
   Heredity from TH: If THROUGHOUT appears in a TH the heredity is
   defined below. This supercedes the SCOPE attribute. If THROUGHOUT
   appears on a TD cell it applies to the current cell only.
   
   Columm-wise: Column-wise heredity flows from the current TH to all TH
   and TD entities which are in later rows and entirely within the set of
   columns that the current TH spans. Column-wise heredity applies by
   default if the current TH has a [least] row index less than its
   [least]column index and no TD cells above it anywhere in the same
   column(s).
   
   Row-wise: Row-wise heredity flows from the current TH to all TH and TD
   entities which are in later columns and entirely within the set of
   rows spanned by the current TH. Row-wise heredity applies by default
   if the current TH has a row index more than its column index and no TD
   cells before it anywhere in the same row(s).
   
   Diagonal: Diagonal heredity flows from the current TH to all TH and TD
   entities which have a least row index greater than the greatest row
   index of the current TH and have a least column index greater than the
   greatest row index of the current TH. Diagonal heredity applies by
   default if the least row index of the current TH is equal to its least
   column index and the initial segment of the principal diagonal
   starting with the first row and column up to the least row and column
   of the current TH passes through no TD entities.
   
  Usage: on TH, TD, and all table containers
  
   Reserve the following values
     * KEY -- the TD elements that inherit this property are default
       CONDS values for all other TD elements in the same row.
     * STARTOVER -- used as the first list element, this functions as a
       directive which blocks heredity of THROUGHOUT items from enclosing
       scopes.
       
   In the prose, discuss the fact that on these elements it is
   appropriate to use as THROUGHOUT values
     * the names of physical dimensions such as LENTGH, PERMITTIVITY,
       EIRP, etc.
     * items from the journalist's agenda: WHO, WHAT, WHEN, WHERE, WHY,
       and HOW.
     * Type-names or Class-names defined in a data definition resource
       linked to the table but not in the current HTML document.
       
   If THROUGHOUT or some equivalent inheritable markup capability is not
   implemented, use CLASS in this way and assume the heredity for the
   purposes of heuristic or "guessing" algorithms.
   
AXIS attribute

   Remove from the language; use THROUGHOUT instead.
   
TSPAN container. Heuristic spelling Tabular-SPAN.

   Only usable in a TABLE.
   
   Start Tag requires ID.
   
   End Tag is required and requires ID matching start tag. IDs are unique
   per element, not per textual occurrence. All other end-tags are
   implicit because we can get away with textual inclusion to show
   structure for linear and tree structures. In tables that is not
   sufficient and we need to be able to have interdigitated start/stop
   tags tied by shared ID values. Is this going to break every SGML-based
   parser in the world? If so, we can do without, but this is the concept
   one wants in a TABLE context.
   
   Used to collect rectangular regions of convenience within a table.
   
   The elements semantically contained in the TSPAN entity is not the
   simple textual range as with SPAN. A "graphical" selection of a
   rectangle in row and column space is implied. The contents are all
   cells with row and column values not less than the [least] row and
   column of the first TD or TH following the <TSPAN ...> start tag and
   not greater than those [the greatest] of the last TD or TH element
   prior to the </TSPAN ...> end tag.
Received on Friday, 10 October 1997 14:54:17 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:56:11 UTC