Re: Querying the jcr:primaryType property [2]

[Marcel Reutegger]

hmm, interesting thought... and I think you're right. I guess that needs 
some clarification in the spec. it currently states that element node 
test and jcr:primaryType restrict nodes to the common type.

shouldn't the SQL equivalent be:
SELECT * FROM nt:base WHERE jcr:primaryType='X'
and
SELECT * FROM X
?
regards
 marcel
Peeter Piegaze wrote:
Well, actually this is how it was implemented and at the time 
we thought it made sense (or maybe we didn't think about it 
too much ;-). 

Now that I look at it again it does strike me as incorrect.
IMO the correct behavior should be:
//*[jcr:primaryType='X'] return all nodes with *declared* type of X.
//element(*,'X') returns all nodes of type X (declared and inherited)
The equivalents in SQL would be (respectively):
SELECT * FROM jcr:primaryType WHERE jcr:primaryType='X'
and
SELECT * FROM jcr:primaryType