Roger Wallin writes:
 > Hi,
 > this is probably a newbie-question but we do have difficulties to 
 > understand the contextsets. Hope you don't mind. 

Not at all.  All context sets do is provide a context in which to
interpret index names (and relation names, relation modifiers, boolean
modifiers and sort-index modifiers).  They are strongly analogous to
XML namespaces, which of course provide a context in which to
interpret XML element names.  If you understand that
	<a:title xmlns:a="http://example.com/1"/>
and
	<a:title xmlns:a="http://example.com/2"/>
are different XML elements, and that
	<a:title/>
may be the same as either or different from both depending on its
context, then you'll be fine with CQL context sets.

(At the meeting in the Hauge a couple of years ago, I suggested that
it might be helpful to rename CQL context sets as CQL namespaces.  It
didn't get changed, but the analogy is still helpful.)

 > <index id="4">
 >   <title>author</title>
 >   <map>
 >     <name>author</name> 
 >   </map> 
 >   <map>
 >     <name set="dc">creator</name> 
 >   </map>
 > </index>
 > 
 >
 > If you want to show by the explain that the server accepts indexes without 
 > a contextset, is it possible to map the index as above, ie. author without 
 > contextset, allowing queries:
 > query=dc.creator="lindgren%"
 > query=author="lindgren%"

Yes.

 > or should there always be a contextset?

There always is one, but the client doesn't always need to specify it:
the server may have a default context set that it uses when none is
explicitly specified.

 _/|_	 ___________________________________________________________________
/o ) \/  Mike Taylor    <[log in to unmask]>    http://www.miketaylor.org.uk
)_v__/\  "Art isn't easy, any way you look at it" -- Steven Sondheim,
	 "Putting it Together"