>Hi all,
>
>We have a successful application. Let us call it an applet that consumes both:
>- free and fee-based XML content (in nicely formatted XML trees),
>- local tabular input by the user.
>In the end we would have an XML document with:
>- all structural meta data in a nicely formatted XML header,
>- the CSV data in a blob with a CDATA section for performance and safety.
>
>Is it an acceptable frame of work? Possibly a beaten-to-death issue but we welcome any feedback!

What we do in one of our projects might be relevant. Our implementation does not involvce CSV data, only CDATA fields. Is there a reason you think that storing CSV data in a CDATA field could be a bad practice?

We save user-adjustable settings in cursors that are transformed into XML with CURSORTOXML(), and the XML output saved to a SQL Server field. Included in the settings is a cutomizable T-SQL code block. The code block is stored in a CDATA section.

Chris