Attribute doubling in XHTML root element

[technosophos]

Under some circumstances, reading an XHTML document and then writing it out results in the doubling of certain special-use attributes on the root element -- namely, xmlns and xml:lang.

This can be reproduced in plain DOM with something like this:

$doc = new DomDocument("1.0");
$doc->loadHTML($_html);
print $doc->saveXML();

It is unclear whether this has any truly negative side-effects, but it is incorrect nonetheless. Since it does appear to be a DOM bug, for QueryPath to fix it, it would have to be done as a post-processing step.

[technosophos]

That is interesting indeed. I may have to look at the libxml source to see what might be the cause of this.

[hakre]

There is a different mode in XML or HTML. As HTML does not have any of those XML namespace prefix declarations (and technically in XML those namespace prefix declarations are not attributes), the doubling can be explained by having attributes which look the same like namespace declarations.

So this might be a technical explanation why/how this can happen, however it's not the solution. So probably better load X(HT)ML documents with loadXML instead of loadHTML?