On Wed, 29 Nov 2006 17:37:52 +0100, Sam Ruby <[EMAIL PROTECTED]>
wrote:
The bug would request that Wordpress doesn't try to output XML for the
text/html media type. That seems to be the problem here.
If the code for Wordpress fit on a page, that suggestion would be easy
to implement.
As it stands now, it appear that several hundred lines of code would
need to change. And in each case, the code would need to be aware of
the content type in effect. In some cases, that information may not be
available. In fact, that may not have been determined yet.
Why would the code need to be aware of that? Atom supports type="html".
One way cross-cutting concerns such as this one are often handled is to
simple capture the output and post-process it. Latchlan opted to do so
with the WHATWG Blog. The first pass for things like this generally
takes the form of simple pattern matching and regular expressions.
The solution used on the blog was a simple find (/>) and replace (>).
Often this evolves. What would be better is something that could take
that string and produce a DOM, from which a correct serialization can
take place.
Shouldn't this be done throughout the code instead of at the end?
Now, what type of parser would you use? HTML5's rules come
tantalizingly close to handling this situation, except for a few cases
involving tags that are self-closing...
Those cases are covered as well. You seem to think that "parse error"
means the same as it does in XML or something. It does not. It merely
means there's a syntax error and that you can continue parsing following
the specified rules.
--
Anne van Kesteren
<http://annevankesteren.nl/>
<http://www.opera.com/>
