Currently XML export is tied to lookups. When we implemented the KEW lookups for doc type, rules, etc. in Rice 1.0 we had to tie them into custom query execution instead of using the KNS's built in auto-search generation query features. As a result of this, we aren't bound by the KNS's default cap on results returned from lookups.
It turns out this is actually a good thing for the XML export feature because it is often required to export more than 200 results (especially in the case of rules). However, this results in inconsistencies in our lookup implementations.
What we need to look into is designing functionality which will provide better support for both XML ingestion and XML export. This would likely include:
1) A dedicated screen for handling export of multiple types of business object data
2) Move the XML ingester out of kew and into the core (or some other module?). Improve hooks into this feature so that even client apps can take advantage of it if they want to.
3) Improvements to the XML ingester. Ability to create better "packaged" exports and import them instead of requiring them to handle a bunch of individual files.
4) Once these capabilities are avaiable, retrofit the KEW lookups to properly throttle search results to be consistent with the rest of the lookups
These are just some ideas, brainstorming and design work will be required here to flesh this out.