- From: James Graham <jgraham@opera.com>
- Date: Mon, 21 Sep 2009 17:50:20 +0200
- To: Mark Birbeck <mark.birbeck@webbackplane.com>
- CC: Henri Sivonen <hsivonen@iki.fi>, Manu Sporny <msporny@digitalbazaar.com>, HTML WG <public-html@w3.org>, RDFa Developers <public-rdf-in-xhtml-tf@w3.org>
Mark Birbeck wrote: > Hi James, > > I feel bad pointing this out after all of your hard work...but thanks > to Elias Torres there has been a Python RDFa parser since early 2006: > > <http://rdfa.info/2006/06/04/a-python-rdfa-parser/> I wasn't attempting to prove that it is impossible to implement a python-based RDFa parser. That would have been a silly thing to try and prove since one could, if it came down to it, implement an entire (X|HT)ML parser/tree API with whatever design is needed to process RDFa. Instead, I was attempting to show that the design of RDFa forces the developer to choose to either a) make a careful choice of tree APIs to match the assumptions of RDFa (but not of many tree-API implementors who are familiar with Namespaces-in-XML 1.0 but not RDFa) or b) treat XML and HTML input distinctly. Given this goal, the existence of a Python based RDFa parser that is based on DOM is not that interesting because DOM happens to be one of the tree APIs that meets criteria a). Indeed, given the relative quality and popularity of common Python DOM implementations compared to other tree-model implementations, I would regard a library chosing to use DOM today as somewhat suspicious in itself. > This is because the issue being discussed was not: > > Can an RDFa parser process xmlns-based attributes when running on > top of an XML stack? > > The answer to that is self-evidently 'yes'. The answer is more like "yes, if you choose an XML stack that allows you to hack around the issues caused by using xmlns attributes to declare prefixes that are used in content".
Received on Monday, 21 September 2009 15:50:41 UTC