>>jdemle+(OP)
I guess these are quite slow (because no indexing) once you have a serious number of records? That in itself isn't a problem as long as you understand the scope of the project. I wonder why they didn't use (a well-defined subset of) CSV as the format however.