Presenting mathematical formulae is a fundamental part of its communication. Authors, and applications, may choose to encode formulae in their documents using presentation markup, as is done in TeX or MathML-presentation. An alternative practice is to encode formulae by their semantics, using OpenMath or Content-MathML and letting an automated process convert the formulae to presentation code.

This last way supports the many value-added services of semantic mathematical encodings, since it exposes the formulae semantics. We present an approach where the rendering process is configurable by author-defined notations that can be interleaved with user-defined notations. This process can be dynamically enriched by new notations of (new) symbols that were just found since it bases on notations documents which are declarative.

Aside of serving of food for the rendering process, the notations documents appear to provide also the right information that can be presented to authors which shop for their symbol.