==================
* Fixes crash when HTML meta refresh URL is empty.
* Fixes crash when decoding a document that is malformed later in the document. These invalid documents are not searched for links.
* Reduces CPU usage when ``--debug`` logging is not enabled.
* Better support for detecting and differentiating XHTML and XML documents.
* Fixes converting XHTML documents where it did not write XHTML syntax.
* RSS/Atom feed ``link``, ``url``, ``icon`` elements are searched for links.
* API:
* ``document.detect_response_encoding()`` default peek argument is lowered to reduce hanging.
* ``document.BaseDocumentDetector`` is now a base class for document type detection.