RE: [xsl] Constructing Simple Content and Built-in Template Rules

[Peter Gerstbach <peter@xxxxxxxxxxxx>]

Quoting Michael Kay <mike@xxxxxxxxxxxx>:
> The effect is the same, but by coincidence rather than by definition. I
> think the effect will always be the same if the content is untyped.
>
> In 2.0 in the presence of a schema, however, doing the recursive descent
> with apply-templates will give you the concatenation of the text nodes,
> whereas value-of takes the typed value and converts it to a string, which
> may be subtly different. In the worst case, if <root> is described in the
> schema as an element with element-only content, value-of will fail, because
> such an element has no typed value.

Thanks for your quick answer.

But this worst case is in fact a problem fore me.
Why does atomization of an element-only node raises an error? Wouldn't it be
better to concatenate the typed value of all node descendants, as it is done
with the string value?

Peter