W3C home > Mailing lists > Public > www-qa@w3.org > August 2002

Re: Usage of the word "specification" in the spec GL

From: Al Gilman <asgilman@iamdigex.net>
Date: Sun, 11 Aug 2002 08:51:45 -0400
Message-Id: <>
To: (wrong string) Žl-Massieux <dom@w3.org>
Cc: www-qa@w3.org

At 01:54 PM 2002-08-10, Lofton Henderson wrote:
>You're right, we have been careless in using "specification" to mean both 
>the document and the technology that it describes.
>>This is especially true for profiles and levels which are usually
>>defined in separate documents.

What you are stumbling over is whether it is the sign or the sense of the
specification which is the view in which the unit is perceived, and whether we
need at the level of a one-word term to make this distinction, or should
only make this distinction at the level of compound Terms of Art.

Using the term 'specification' generically in the way you have is correct
scoping in concept space, not a bug.  We need compound terms where we need
to distinguish.  The document fixation of our current habits of language use
is a legacy we do well to unlearn.

Two prominent bases for scoping or entifying 'a specification' are:

+ the collection of specification-stuff that is applied as a unit.  The unit
of sense.
-- this gets referred to as the 'governing' or 'effective' specification
-- unit in conformer's view
-- this refers to a fully-elaborated collection of constraints, forming an
-- the acceptor-pattern governing an aspect of a ?conforming? instance.

+ the collection of specification-stuff that is published as a unit.  The
unit of sign.
-- could be called a specification 'module' or 'document'
-- unit in author's view
-- this refers to utterance record in a form shareable across those creating
potentially conforming instances, and those evaluating these instances for
-- a writing inducing consistent acceptance patterns [observed in extenso]
in varied contexts where the operation of discriminating conforming
instances is desired.

The general use of the term 'specification' assumes both that there is a
published utterance record in some language denoting acceptor-pattern
clauses, and that the accptor-pattern one infers from any instance of the
controlled data [the writing in the document] is highly consistent by the
care with which the utterance is expressed.  In other words both of the
above.  Or either of the above, as a stem for a compound term.

The point is that the phrase "XHTML, as used in conformance with the
XHTML+MathML+SVG Profile," denotes a knowledge unit of class 'specification'
even 'though it is not separately published anywhere.  But the language of
constructing the reference is crisp enough so that this is a well-posed unit
of markup language specification.


>You wrote...
>At 06:21 PM 8/9/02 +0200, Dominique HazaŽl-Massieux wrote:
>>As of the current editor version of the spec GL [1], the word
>>"specification" seems to be used inconsistently. Sometimes it seems to
>>refer to a technical report unit (namely, an entry in the TR page [2]),
>>sometimes a technology unit. Especially, all the GL "Address the use of
>>[] to divide the specification" seems to use the word specification as
>>meaning technology (or assumes a pre-existing specification that would
>>then be divided which is often wrong for the said divisions).
>You're right, we have been careless in using "specification" to mean both 
>the document and the technology that it describes.
>>This is especially true for profiles and levels which are usually
>>defined in separate documents.
>>Anyway, I think we need:
>>- to clarify the meaning of the word specification in the introduction
>>- adjust the GL accordingly.
>For your second dash-item, how about "specification's functionality" in 
>place of the appropriate occurrences of "specification"?  Would something 
>that simple suffice?  (Feel free to propose specific wording for either 
>dash-item -- the editors would not be offended!).
>>1. http://www.w3.org/QA/WG/2002/08/qaframe-spec-0804.html
>>2. http://www.w3.org/TR/
>>3. http://www.w3.org/TR/XHTMLplusMathMLplusSVG/
>>Dominique HazaŽl-Massieux - http://www.w3.org/People/Dom/
Received on Sunday, 11 August 2002 08:52:02 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:40:29 UTC