> increase the depth of tree. But if instead the tags are <br/> the depth 
> of the parse tree remains same.
> 
> The reason why I don't want to handle tag soup parsing in the parser is 

<br> in HTML is NOT tag soup.  Tag soup is things like:

<b>  xxx <i> yyy </b> zzz </i>      - elements don't nest
<b><p>......</p></b>                - content model violated (block in inline)

<br> in HTML has the same meaning as <br/> in XHTML, and results in the
same parse tree.

HTML will be required for decades to come, not just for legacy content.
IE doesn't support proper XHTML.

I believe the suggestion to have a tag soup and XML parser was trying to
say that a good tag soup parser will also parse valid HTML.  It wasn't
saying that valid HTML is tag soup.