Technique for Attribute Access

Per our discussion today, here's a possible checkpoint and technique.

Checkpoint: Allow the user to learn of attributename="value" pairs for any
element. 
Priority: 3

Technique: 
The meta-information about an element contained in attribute values can 
support the understanding of the containing element, or its descendents.
It should be exposable by user request.

Details: 

Many attributes are implicitly present, so the presence in an element of an 
attribute=value pair cannot be depended upon for information. It is up to the 
semantic description of each attribute name and its allowed values for an 
element type to indicate both the explicit meaning of the attribute, and what 
is the defaulting meaning of a missing attribute. Often this meaning can be 
determined from the value of the same-named attribute on an ancestor element.
Other times the semantics indicate no inheritance (e.g., an id, a document-
unique identifier for referencing a particular element). 

Some attributes are generally applicable to all or most element types. For 
these, common understanding of their use is reasonable to expect for authors 
and users. 

Search on attributename and possibly value is important. Search may be on 
the next attributename, or on attribute value, possibly inherited, possibly 
qualified by element name. 

Each search returns the next occurrence of the containing element. An 
implementation possibility is to list all matches, and return them in 
order to the user.

Another useful result is to learn the different attributename="value" pairs 
or a subset of them for any particular element.

Another useful search is to find the next element that can have a particular 
attributename, and for it seeking its default value, possibly from the 
nearest ancestor.

Details for HTML 4.0: These common attributes are associated with almost 
all HTML 4.0 element types, and can be learned: 

% attrs
  "%coreattrs; %i18n; %events;"

including:
 
% coreattrs
 "id          ID             #IMPLIED -- document-wide unique id --
  class       CDATA          #IMPLIED -- space separated list of classes --
  style       %StyleSheet;   #IMPLIED -- associated style info --

    ELEMENT TYPES WITHOUT %attrs; but with %coreattrs
        BDR  -- bi-directional override--
        BR   -- forced line break --
        HR   -- horizontal rule --

% i18n
  lang        %LanguageCode; #IMPLIED -- language code --
  dir         (ltr|rtl)      #IMPLIED -- direction for weak/neutral text --"

    ELEMENT TYPES WITHOUT %attrs; but with %i18n; 
        HTML, HEAD   -- lang, dir defaults for document --
        META, TITLE  -- lang, dir, for use with content --
        STYLE        -- lang, dir, for use with title in HEAD --
    
% events;
 "onclick     %Script;       #IMPLIED -- a pointer button was clicked --
  ondblclick  %Script;       #IMPLIED -- a pointer button was double clicked--
  onmousedown %Script;       #IMPLIED -- a pointer button was pressed down --
  onmouseup   %Script;       #IMPLIED -- a pointer button was released --
  onmouseover %Script;       #IMPLIED -- a pointer was moved onto --
  onmousemove %Script;       #IMPLIED -- a pointer was moved within --
  onmouseout  %Script;       #IMPLIED -- a pointer was moved away --
  onkeypress  %Script;       #IMPLIED -- a key was pressed and released --
  onkeydown   %Script;       #IMPLIED -- a key was pressed down --
  onkeyup     %Script;       #IMPLIED -- a key was released --"

    ELEMENT type without %attrs; but with %events; (and %coreatts;)
        HR -- horizontal rule, takes space, has no text --

    ELEMENT types with none of %attrs;, %coreatts;, %i18n;, or %events;
        PARAM
        BASE
        SCRIPT

Some attribute names are used in consistent ways, as reflected by their comments. A few seem
to have different semantics. The latter should be avoided, particularly where inheritance
of attribute value is presumed.

For XML applications, a similar set of general attributes may be useful. A common occurrence
is to specialize a general element name to indicate a local variation on the general
element type. For example, in electronic books, different books have different names for
the hierarchic sectioning. Some use chapter, some use part, some use section, some use "El Parto".

Regards/Harvey Bingham

Received on Wednesday, 5 May 1999 18:25:18 UTC