David Morris (
Sat, 8 Jul 95 01:09:24 EDT

On Thu, 6 Jul 1995, Eric Bina wrote:

> > Example 3:
> >
> > <!doctype html PUBLIC "-//IETF//DTD HTML Experimental//">
> > <html>
> > <head>
> > <title>blah</title>
> > <style>blah...</style>
> > <!-- <STYLE> unknown; don't infer </HEAD><BODY>; ignore content -->
> > </head>
> > <body>
> > <newel>blah</newel>
> > <!-- <NEWEL> unknown, but <BODY> has been seen; include content -->
> If I understand this proposal, it fails for new non-content head tags.
> In the above example you assume that even though the browser doesn't know
> <STYLE>, it knows to match it to </STYLE> and ignore content. Suppose
> you have:
> <!doctype html PUBLIC "-//IETF//DTD HTML Experimental//">
> <html>
> <head>
> <title>blah</title>
> <newtag>blah...
> <!-- <newtag> unknown; how do you know how far ahead to look for
> </newtag> before giving up and inferring </HEAD><BODY>
> to make "blah..." the start of the body of the document? -->

Well, we have a new (or erroneous input) tag used which doesn't happen
to require content. By definition such a tag would post date this
error recovery implementation so it is reasonable (IMHO) to expect
the author who uses the unknown new <head> tag to include something
which will definitively start the <body>.

Existing 'poorly' (or lazily) formed documents will generally not fail,
the unfortunate exception being those which use a non-universally
implemented tag before the start of the body can be recognized.

<newtag> is closed by </newtag> or a recognized tag which starts the

Dave Morris