Re: 9. WYSIWYG editor (enforcing the signature)

Karl Dubost writes:

> I don't believe having a metaname for wysiwyg editor is solving
> anything at all.

I think it's:

* We want that an author of a valid HTML5 document isn't permitted to
  use certain 'presentational' things, instead putting semantic mark-up
  in the HTML and using CSS to achieve the desired presentation.

* Some HTML is generated now by users directly, but by applications,
  such as word processors with a 'save as HTML' option.  Some of those
  applications permit users to specify presentational output without
  giving a reason for doing so (for example, a user may consistently
  highlight all defining instances of terms in blue, or use a different
  typeface for headings, but there's no way the application is aware of
  this).

  So when generating the HTML the application cannot follow the rules
  for authors writing HTML by hand; if the user has only told the
  application to make something blue, all the app can do is generate
  HTML which specifies its blueness.

  So _some_ HTML-generating software can't meet all the author
  requirements.

* That software isn't going to go away.  Whatever this group decides,
  such apps will still be used.

* Given such an app's existence, it's better if it meets all the _other_
  author requirements in the spec, the things that it does have enough
  information to do.  So the spec acknowledges the few things that such
  apps cannot meet and declares documents they generate still to be
  valid HTML5 if they meet everything else.

* The meta tag enables documents to be distinguished, so as to know
  which leve of conformance they were aiming for.

> Saying that only wysiwyg editors need/put font tag is ignoring the
> fact that document move from one tool to the other. A document is not
> edited by one tool, but can be edited by multiple type of tools at
> different times.

True, but that doesn't prevent such binary tagging from being used: the
tag indicates the document contains some content which was generated
purely from presentational direction by a user and so may be doing
things that hand-written HTML5 shouldn't and be missing relevant
semantic information.

So if any of the document has been generated by such software, the tag
still applies (until such a time that all such-generated HTML has been
hand edited and it's no longer relevant).

But ... we could still achieve the distinction in validation levels
without documents self-identifying.  Given that the full requirements
are a superset of those for non-semantic editors, it's possible for a
validator simply to check a document against all the rules and then give
one of three responses:

* This document conforms to valid HTML5.  Congratulations, here's a
  logo.

* This document is acceptable HTML5 if it were generated by software
  which couldn't know any better.  If that's the case, then here's a
  (different) logo.  But if you wrote this by hand then it isn't good
  enough, here's some things to fix ...

* This isn't HTML5.  Here's why not ...

> This is a shorcut, I have not made in the proposal. :)
> 	<meta name="conform" content="html5-bp"> 
> is not for assessing the quality of the document. That would not be  
> very effective and easily spoofed.

Indeed.  That's why if a meta tag is to be used to distinguish the two
levels of documents it needs to be the 'lessor' level that is labelled,
such that the label is 'I know I'm substandard; please be nice to me',
rather than 'I'm claiming to be really good'.

Smylers

Received on Wednesday, 8 August 2007 13:03:04 UTC