To summarize, our ideas so far on how to store and export content with ML when there's an existing precedent (e.g. a standard schema) for the sort of data you're working with:
Alternative strategies: (1) Transform the standard XML as part of the loading process so that it is convenient and efficient to query and search. For compatibility with external systems, have a transformation/export process. This can be done in different ways: a. At load time, move the document's content into a top-level container node, <standard-xml>, to store the original structure. Add to this a sibling container node, <internal-xml>, to store the structure that has been transformed for convenience. On export, simply remove <internal-xml> and restore the standard structure. b. At load time, store only the transformed document. On export, apply a reverse transformation back to the standard. (2) Work around the difficulties, performing multiple searches if necessary and dealing with extra complexities in indexing. Evidently, the choice (between these two main alternatives, anyway) depends on how many difficulties (2) would entail. I suppose it boils down to a comparison of two costs: the cost of the import/export transformation process vs. the cost associated with handling a particular standard structure in ML. Thanks to everyone for examples and ideas! Karl _______________________________________________ General mailing list General@developer.marklogic.com http://xqzone.com/mailman/listinfo/general