To me it seems easier to fix the code problem than to create a new index. Per http://docs.marklogic.com/cts:element-values simply specify the correct collation with the cts:element-values call(s). It defaults to the default collation in your evaluation context, whatever that is.
-- Mike On 1 Apr 2014, at 06:52 , Gary Larsen <[email protected]> wrote: > Hi, > > After changing the collation to codepoint on the XDBC server existing range > indexes were failing: > > "No string element range index for > fn:QName("http://developer.aaa.com/xmlns/aaa/bbb/";, "contentObjectClass") > http://marklogic.com/collation/codepoint"; > > It seemed to happen when using cts:element-values() but did not research > this further. This is why I ended up creating a range index with the > codepoint collation. > > Thanks, > Gary > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of Jason Hunter > Sent: Saturday, March 29, 2014 12:59 AM > To: MarkLogic Developer Discussion > Subject: Re: [MarkLogic Dev General] false match on cts:element-value-query > > Yes, that will get the job done. If you could get it to work following my > suggestion, and if you aren't using the range index for other purposes, then > you could free up the memory used by that range index. > > -jh- > > On Mar 28, 2014, at 7:01 AM, Gary Larsen <[email protected]> wrote: > >> Hi Jason, >> >> Was able to avoid using 'filtered' query by: >> >> - creating element-range-index with codepoint collation >> - changing element-value-query() to element-range-query() using codepoint >> collection >> >> Didn't need to use 'exact' in my case. >> >> Thanks! >> >> Gary >> >> -----Original Message----- >> From: [email protected] >> [mailto:[email protected]] On Behalf Of Jason Hunter >> Sent: Thursday, March 27, 2014 12:28 PM >> To: MarkLogic Developer Discussion >> Subject: Re: [MarkLogic Dev General] false match on > cts:element-value-query >> >> It will be more efficient if you can make it work unfiltered, which is >> actually possible. In the value query you need to specify the "exact" >> option (which is a shorthand for everything sensitive), and make sure that >> your collation context is right. I believe your collation needs to be >> codepoint. MarkLogic maintains a term list for "exact" matches using the >> codepoint collation, so it can resolve fully sensitive out of indexes. > You >> won't see it normally used because the default collation is root. >> >> -jh- >> >> >> _______________________________________________ >> General mailing list >> [email protected] >> http://developer.marklogic.com/mailman/listinfo/general > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general > _______________________________________________ General mailing list [email protected] http://developer.marklogic.com/mailman/listinfo/general
