A low-memory streaming algorithm for XSLT processing implemented in Xord framework

We present an implementation of the Xord framework for streaming processing of XSLT transformations based on .Net technologies. Within the framework, we implement an efficient streaming algorithm capable to process a significant subset of top-down XSLT transformations. We exactly characterize this class of transformations. The algorithm uses a stack of the size proportional to the depth of the input XML document. Such memory usage is highly efficient in practice, since real-world XML documents are shallow. The evaluation of the algorithm supports our expectations and shows that a transformation of large but shallow XML documents is processed using a constant size of memory.