[xsl] Re: C14N and XSLT

["Flynn, Peter pflynn@xxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

On 2018-04-10 14:13:07+01:00 Geert Bormans geert@xxxxxxxxxxxxxxxxxxx wrote:
I believe that by the book, I can not guarantee an XML serialisation from an
XSLT renders a canonical XML, can I?
To me things like <foo></foo> instead of <foo/> are impossible to force, or
did I missed something?
(I am aware of some undocumented tricks that help me force this in some
processors, but that is no guarantee for a long lasting solution)

That particular one failed to make the grade, for reasons I was never able to
understand. The argument I heard was similar to the one for broken white-space
handling: that it depends on the schema/dtd for the processor to evaluate if
<foo/> is declared EMPTY or if <foo></foo> is allowed content but just happens
not to have any in this instance, or if there was no schema/dtd in use (even
if one existed).

But this would have meant the output being different depending on whether or
not the schema/dtd had been used, and this was apparently too much for some
participants to accept, even though that is already the case in respect of
default and fixed attribute values.

I have only ever come across one instance where it really matters, however, so
for my own purposes it's not really important. But it seems to depend on
whether or not the community can live with it.

///Peter