W3C home > Mailing lists > Public > www-html@w3.org > April 2007

RE: [XHTML1.1] Error in Conformance Definition document?

From: Benjamin Hawkes-Lewis <bhawkeslewis@googlemail.com>
Date: Mon, 02 Apr 2007 22:37:16 +0100
To: "Paul Nelson (ATC)" <paulnel@winse.microsoft.com>
Cc: www-html@w3.org
Message-Id: <1175549836.28283.25.camel@galahad>

+1 for making serving XHTML 1.1 as application/xhtml+xml a MUST. In the
current XML-hostile climate, I would see no harm in W3C also defining an
text/html serialization of XHTML 1.1, especially in order to backport
accessibility features (HTML 4.02 would be an equally good name for
this). But please end the twilight zone non-transitional transition of
XHTML 1.0 as text/html.

Paul Nelson (ATC) wrote:
> If you are making a simple web page do you really want the user to be
> setting MIME type of "application/xhtml+xml"? It seems that "text/html"
> or creating a MIME type of "text/xhtml" for well formed requirement
> would be a better option. Most pages are not applications.

And yet most pages should be /produced/ by applications (since asking
ordinary people to learn freaky languages like HTML is unreasonable) and
those applications should obviously generate well-formed markup.

The only reason your argument has any traction is that most current
mass-market applications for producing HTML and XHTML are not fit for
purpose, not least because of assumptions by the programmers that
documents don't really need to be well-formed. But it's not the spec
that needs fixing there.

In any case, as Jukka adumbrated, there is more to the difference
between text/html and application/xhtml+xml than just processing
requirements relating to well-formedness.

And then a little later in the thread, Paul wrote:

> In the end there needs to be content creators who care about putting
> closing tags in the right place and generating well formed content.

I think this is exactly backwards. Assuming we are talking about human
beings, content creators probably shouldn't be worried about tags at
all, let alone closing them properly. (Meaning yes, angle-brackets no.)
Even if humans are unlucky enough to be entering tags by hand, their
software should mangle their data entry until it is well-formed. If it
is impossible to mangle their entry into well-formedness that is
probably a sign that it is ambiguous and further clarification from the
user is required. There is no justification for "tag soup out" systems.
They just transfer the problem of interpretation to an assortment of
idiosyncratic user agents.

Benjamin Hawkes-Lewis
Received on Monday, 2 April 2007 21:56:22 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 30 April 2020 16:21:02 UTC