Re: [xsl] Confusing trace from Saxon HE9

[Jeroen Hellingman <jeroen@xxxxxxxx>]

Hi Ken,

Thanks for your look, I'll try drilling down where the $node assignment 
happens, but I am pretty sure I am not intentionally putting the things
in there that my code exposes.

G. Ken Holman wrote:
> Your stylesheet doesn't compile standalone because of missing template 
> rules, which indicates to me there are other fragments at play.
That is correct, since it is part of a whole, which includes about 15 
XSL files. I'll try to isolate the issue in a smaller single-file XSL 
sheet for this list.
> > Furthermore, the failing template receives two nodes in a variable 
> > that I hardly expect to be grouped together. In the source document 
> > they are not even siblings.
Expose your $node variable after you assign it and see where it gets 
more than one node in it.

The $node variable is not used at all in the way I am using the 
templates at the moment. What I try to do is use some slightly complex 
way of splitting
a TEI into manageable sized chunks while converting it to XHTML for ePub 
consumption, and a range of related ePub support files. I need to do the 
same complex splitting in four different
cases, so instead of writing the complex splitting code several times, I 
wrote it once, giving an additional parameter $action through the entire 
recursion, to select what
to do when I've reached the bottom of the recursion. In some cases, I 
need an additional $node parameter, but in the case that puzzles me, no 
$node parameter
is needed. I now explicitly set it to the empty string at calling time, 
but the issues still appears.

The current question is, how can a trace show an error in a stylesheet 
that is not even matched or called in the stylesheet higher on in the trace.

Jeroen.