Typechecking XML views of relational databases

Motivated by the need to export relational databases as XML data in the context of the World Wide Web, we investigate the type-checking problem for transformations of relational data into tree data (i.e. XML). The problem consists of statically verifying that the output of every transformation belongs to a given output tree language (specified for XML by a document type definition), for input databases satisfying given integrity constraints. The type-checking problem is parameterized by the class of formulas defining the transformation, the class of output tree languages and the class of integrity constraints. While undecidable in its most general formulation, the type-checking problem has many special cases of practical interest that turn out to be decidable. The main contribution of this paper is to trace a fairly tight boundary of decidability for type-checking in this framework. In the decidable cases, we examine the complexity and show lower and upper bounds. We also exhibit a practically appealing restriction for which type-checking is in PTIME