W3C home > Mailing lists > Public > public-html@w3.org > May 2007

Re: Rethinking HTML 5 (Was: Re: Semicolon after entities)

From: Jeff Cutsinger <jeff@cutsinger.org>
Date: Tue, 01 May 2007 12:16:10 -0500
Message-ID: <463775DA.3020907@cutsinger.org>
To: "Philip Taylor (Webmaster)" <P.Taylor@Rhul.Ac.Uk>
CC: public-html@w3.org
Philip Taylor (Webmaster) wrote:
> Andrew Sidwell wrote:
> 
>> If HTML5 is to be backwards-compatible
> 
> Is this a given ?  Is it even desirable ?

No, it isn't a given. Yes, it is absolutely desirable. If not to you,
then to a much larger class of developers.

> And what does it really mean ? That a document
> written in HTML5 will display  "correctly" in
> browsers that are HTML5-unaware ?

Yes.

> That's
> demonstrably impossible, unless HTML5 is
> a strict /subset/ of earlier incarnations
> of HTML (as opposed to superset, which an earlier
> correspondent proposed).

You are incorrect. The WHATWG specs as defined are (loosely speaking) a
superset of HTML 4 (in that they add useful features) and are also
backwards compatible.

> 
>> and have well-defined error
>> handling, then I'm not sure
>> how much that differs from defining error
>> handling for existing content.
> 
> I remain unconvinced that defining error handling
> for "existing content" is a worthwhile (or even
> achievable) exercise.  By all means define error
> handling for HTML5; even -- if you must -- allow
> a conformant HTML5 user agent to recover gracefully
> from /some/ (but clearly not all) syntactic errors;
> but please don't waste time defining how to process
> tag soup.

I don't know how many times this has been gone over. Your choices are:

1. Reject erroneous content. This is the path used by XML. If you desire
this, you can use XHTML 1.0, 1.1, 2.0, or 5. As to following this
course, look where it's gotten these standards: they are practically
non-existent on the web (in their true form).
2. Do what you suggest. Allow graceful recovery but don't specify how it
must be done. This is, of course, what is currently done and, of course,
an interoperability nightmare.
3. Do what the WHATWG has already done. Specify exactly what conformant
user agents must do when they encounter an error. This is what CSS does.
This is absolutely, clearly the best path to take. It allows for
interoperability while not entertaining any fantasies that developers
can be coerced into writing 100% bug-free code. Plus it makes it easy to
develop new code that handles existing content on the web.

Can we please move on? Really, if you're so stuck on this, the WHATWG
standard allows you to use XML! Or use XHTML 1.0 or 1.1 with the
application/xhtml+xml mimetype. Or try out XHTML 2.0.

Received on Tuesday, 1 May 2007 17:16:24 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:44 UTC