Software maintainability can be difficult to plan for when strategies for software development present tradeoffs between popular support and expressiveness. These tradeoffs were explored in the context of a recent software project, where two schema transformation applications were developed in two ways: a first that used the traditional XSLT 1.0 plus pull processing approach to application development, and a second using XSLT 2.0 and push processing. The improvements obtained with the second strategy, which took 1/4 of the time to implement while substantially reducing the size of both applications and the complexity of one, suggest that the benefits of transitioning to XSLT 2.0 and push processing far outweigh the benefits of the older approaches.