W3C home > Mailing lists > Public > www-tag@w3.org > December 2003

Re: Backwards compatibility (was Re: [errorHandling-20] CLOSED...)

From: <jon@hackcraft.net>
Date: Fri, 12 Dec 2003 10:08:39 +0000
Message-ID: <1071223719.3fd993a7722c8@82.195.128.192>
To: Rob Lanphier <robla@real.com>
Cc: Chris Lilley <chris@w3.org>, "www-tag@w3.org" <www-tag@w3.org>

> 2005 - Person B has a browser which fully complies with modern
> recommendations for handling the "text/html" document.  They encounter
> "profound.html", but their browser throws obscure error messages and
> refuses to render the HTML.  Person B assumes that the document is
> corrupted, and doesn't read the document.

Would the following be considered a reasonable error message?

"Error - unrecognised element <xmp>.
Other errors ommitted. Click for details:
The following is a best attempt at rendering, note that this is not a reliable 
rendering due to the above errors:

...
"

In other words we bring back "liberal in what you receive", but as part of the 
actual error message. This would provide a degree of backwards compatibility, 
and indeed some of the advantages claimed for liberal handling of errors, while 
remaining a clear error.

I used <xmp> in my example to raise the point that if you do attempt to have a 
best-attempt rendering as part of an error message then some errors will almost 
inevitably cascade into further errors. Given that 
<xmp><zxcv><asdf><qwert></xmp> had been valid, this would be an example of such.

--
Jon Hanna                   | Toys and books
<http://www.hackcraft.net/> | for hospitals:
                            | <http://santa.boards.ie>
Received on Friday, 12 December 2003 05:27:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:23 GMT