Most XML attributes are not unescaped by the XML parser. After checking the source code it looks like only <ereg> is.

While it has no practical consequences most of the time, it makes it impossible to e.g. use <strcmp> to do something like this:

  <strcmp
      assign_to="result"
      variable="strvar"
      value="&#34;Some User&#34; &lt;sip:[email protected]&gt;"
      />

I suppose solving #31 would fix the problem globally, but in the meantime unescaping at least attributes that are meant to contain strings would be a nice improvement.