FWIW, as bruce says.
Rich text - RTF or otherwise, just as long as it supports and responds to markup - is one of the oft-recurring pleas we've made to Sparx over the past few years. So far we've had a (very) few (very) limited improvements to overall notes visibility and formatting. Collectively, these slightly broaden diagram presentation options, but do little to address our problems, and nothing whatsoever to solve yours.
However...
You are certainly not alone in using XSLT to handle multiple text handling situations. While XSLT can be a lot of work - in almost any context, and certainly with EA and UML - the potential leverage of any solution makes it a compelling option.
That said, why not just store your XML in the Notes compartment? Pretty much everything in EA supports Notes, so you have a fairly universal way of storing and retrieving the XML. Just design your models so that you don't display Notes compartments - this is the default - and you won't clutter your diagrams. However, on processing you can direct the Notes text to your XSLT processor.
You'll have to handle things with care during transformations, since you are probably limited in how you can pass the Notes to an external routine. [You can do this via the EXEC_ADD_IN function macro.] With luck you won't have to do anything radical during transformation, since the sorts of things you identify won't require too much processing during a transform.
Now for the nifty part. EA stores its model in an SQL-accessible schema, regardless of back end. Many XSLT processors (or their supporting environments) support OLE DB connections to databases. You should be able to capture the Notes fields of elements (or sets of elements) and process them as you choose, singly or in batches.
So, while we wait for richer Notes fields (which, as bruce implies, is all we do) you can move ahead with your work.
HTH, David
