> There is a massive gotcha here (or a bug in the spec, 
> depending on your point of view).

A feature, I think!
> Note the template in 'a.xsl' specifies priority 10.  Even 
> with this, the
> output is 'B' from 'b.xsl'.  Because the first rule in 
> template conflict
> resolution is import precedence, and one of the rules of import
> precedence is 'the last import wins', so the processor ignores all the
> usual ways to control priority.  
> 
> However in this case, the usual "ambiguous rule match" warning isn't
> output, making it very hard to track down.  I have asked Mike on the
> Saxon list a while back if it's possible to give a warning for this
> specific case, so hopefully it will make it into 8.5 :)  

This test would be a very expensive one, and it would give a lot of spurious
messages for perfectly valid stylesheets. The test would therefore have to
be switched off by default; but 99% of the people who would be helped by the
test wouldn't know about the switch. So it's not clear that this is a useful
thing to do.

Michael Kay
http://www.saxonica.com/