>Hi Antonio, indeed your model has the advantage of saving the metadata,
>solves the property name case problem and position - we're in sync !-
>
>I took that approach in other companion component called
>nsonXs = no-fox simple object schema notation (you can take a look at it on-line
>presentation I made about it: https://youtu.be/BBC3Ly6Im10?t=947 ) that , as xmlschema
>and json schema, stores that information about the objects, and is able to
>set it using a function called xsApply().
>
>Your code could serve as a function that I was willling to develop for VFP objects
>, called getSchemaFromObject() like http://jsonschema.net/#/ but for XML .
>
>so maybe the right place is not the object itself, but the object schema.
>
>Give me some time and I'll share with you the nson specs and source for you to test.

I still think that the object must carry all the required information to return data back to XML, and that means even more than what you have in the current version. My test document is getting really weird but I'm putting in it everything that a) I can remember that could be an issue but b) I have an idea on how to handle.

Although the next days will be a bit more rough with work and travels, I'll be glad to have a look into nson when possible.