Ha, no that's right: 7.0-2.2. Then that makes a lot of sense now. It would be nice if it threw an exception, or at least logged a warning.
-Will On Apr 25, 2014, at 4:27 PM, Michael Blakeley <[email protected]> wrote: > Will, did you mean 7.0-2.2? If not, can I borrow your time machine? > > It would be nice if the server told you this directly, but > https://docs.marklogic.com/guide/admin/range_index#id_63906 has a mention: > > You cannot have a path span across a fragment root. > Paths should be scoped within fragment roots > > I suppose it reads "should" instead of "must" because the server doesn't > enforce that rule actively, but behaves unpredictably if it is not followed. > > -- Mike > > On 24 Apr 2014, at 09:21 , Will Thompson <[email protected]> wrote: > >> We’re noticing some odd reindexing and querying behavior, and I’m curious if >> there’s anything fundamentally wrong about inserting documents like: >> >> <things> >> <thing-type> >> <thing>value</thing> >> … >> >> with a fragment root set on <thing>, and a path range index for >> things/thing-type/thing. >> >> A cts:values query on that path reference returns empty. Remove the fragment >> root on <thing>, and it returns the expected values. Also (and this is >> really strange) if instead of removing the fragment root, I add an element >> range index for <thing>, reindex, and remove that index, then (after >> reindexing) the query returns expected values. But if following that I force >> a manual reindex, it will return empty again when the reindexing completes. >> This is on ML 7-2.2. >> >> -Will >> >> >> _______________________________________________ >> 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
