Note that this is a "push" approach - all the input has been captured except for whitespace.
In the CML file I have made a best guess as to how to structure the file into CompCML. It has an init module, a final module and an intermediate one which contains the iteration steps. CML is capable of holding arbitrarily complex structures. However it requires some art to be able to write a generic parse which recreates the structure. I'll be trying all of Mark's outputs and intend this will act as a good Lensfield exercise for us.
We should also move this to the Wiki
-- Peter Murray-Rust Reader in Molecular Informatics Unilever Centre, Dep. Of Chemistry University of Cambridge CB2 1EW, UK +44-1223-763069