Re: Bug in 'xsl:sort'. ( XT vs SAXON. )

Paul,

>There was no Sebastian's files. 

Sorry, I assumed from your Sebastian-baiting at the end of the original
question that they were his files that you were using.

><quote>
>lang specifies the language of the sort keys; 
>it has the same range of values as xml:lang [XML]; 
>if no lang value is specified, the language should be 
>determined from the system environment
></quote>
>
>Maybe they are 'determining' diffrent languages 
>from  the 'system environment'. Instant SAXON determines 
>one 'system environment' and XT  ( I was running 
>it on Sun's  JVM ) is determining another 
>'system environment'. 

That's possible.  You could try adding an lang attribute to the xsl:sort
element (which should override any system environment) and seeing whether
this makes a difference.

>I have a feeling that you understand something  that I
>don't understand. To me this is still crazy that 2 processors 
>are sorting this file differently and both are 'correct'. Anyway - 
>the usecase could be considered 'exotic and not critical'.

It's the difference between 'must' and 'may' or 'should' in the
Recommendation.  Conformant processors *must* do certain things in certain
ways, but there's some flexibility (usually constrained in some way) to how
they do others.  XSLT Processors "*should* sort text lexicographically..."
(my emphasis).  They are permitted to sort in other ways, and in any case
the 'culturally correct manner' is not specified within the XSLT
Recommendation - it's only *recommended* that implementers look at UNICODE
sorting.

As you say, it's not a big thing: in most cases, the sort will be
reasonable and consistent from one particular processor, and if it's not
exactly what you want, then you can always try putting in additional sorts
or nesting xsl:for-eaches.

Cheers,

Jeni


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list