Re: ISSUE-61: conformance-language - suggest closing on 2009-09-03

Henri Sivonen <>, 2009-08-21 17:05 +0300:

>  On Aug 21, 2009, at 16:56, Dan Connolly wrote:
> > "Please change the draft so that document conformance does not
> > depend on author's intent at all."
> >  --
>  Obviously, validators cannot check for conformance criteria that depends on 
>  the author's intent.

It's not just an issue for validators -- it's an issue for anybody
who has a document in isolation from the author who created and
who wants to determine whether it's a conformant document or not.

So for any given document, if we want to make it possible for
somebody other than the author who created it to be able to make
an assessment about it whether it's conformant or not, then we
need document-conformance criteria to be restricted to only what
can be determined without knowing the author's intent, and without
knowing what process was used to author or generated the document.

An example of conformance criteria that can't be evaluated given
just a document alone (without knowledge of how it was authored or
generated) is the following statement:

  The DOCTYPE legacy string should not be used unless the document
  is generated from a system that cannot output the shorter string.

>  However, the very concept of semantic markup is useless unless markup is 
>  used according to the intended meaning. Therefore, I think it is even 
>  interop-sensitive (in the sense that receivers process the messages in ways 
>  compatible with sender expectations) that authors use markup according to 
>  the specified semantics.

The question is whether those criteria that depend on knowing the
author's intent should properly be part of the definition of what
a conformant document is.

Others on the list have questioned whether a spec of this kind
should contain authoring conformance criteria at all, or anything
else that relates to document conformance but that could be
classified as something that is just a judgment call being made by
the editor but that another editor might make a different
judgement call on.

Some of the criteria could in fact be seen more as best-
practices-like material that's appropriate for an authoring guide
but that it's not typical to see in technical specifications.

And even for authoring-conformance criteria that are interop-
sensitive, it's not a given that this spec is the appropriate
place for them, nor that it is the optimal place for them as far
as making the intended audience of authors aware of them.


Michael(tm) Smith

Received on Thursday, 27 August 2009 19:54:53 UTC