On Sat, Jan 18, 2014 at 11:25 PM, <d...@geschan.de> wrote: > Hello, > ... > So, the new block-joing feature seems to solve some of these problems > (sadly not all – as far as I see, I can’t get the parent document and the > articles where the search term was found in a nested result). >
it should be something like https://issues.apache.org/jira/browse/SOLR-5285Please, raise new issue or describe your case there. > So, my question now: can I change my existing index in just adding a > is_parent and a _root_ field and saving the journal id there like I did > with j-id or do I have to reindex all my documents? > Absolutely, to use block-join you need to index nested documents as blocks, as it's described at http://blog.griddynamics.com/2013/09/solr-block-join-support.html eg https://gist.github.com/mkhludnev/6406734#file-t-shirts-xml > I made some test in adding the id of the parent journal in the _root_ > field of the articles and trying to make a query like q={!parent > which='is_parent:true'}+description:test but it didn’t seem to work. I > only got an error message: > > java.lang.IllegalArgumentException: docID must be >= 0 and < > maxDoc=1418849 (got docID=-1)\r\n\tat org.apache.lucene.index. > BaseCompositeReader.readerIndex(BaseCompositeReader.java:182)\r\n\tat > org.apache.lucene.index.BaseCo This is definitely 'a courtesy bug'. Feel free to raise an issue. mpositeReader.document(BaseCompositeReader.java:109)\r\n\tat > org.apache.lucene.index.IndexReader.document(IndexReader.java:436)\r\n\tat > > > Do you have any advice how to fix this or to use block-join properly? > Gesh, Block-join is also called the index time join, until you index docs in blocks, it doesn't work. > > Thanks, > Gesh > > -- Sincerely yours Mikhail Khludnev Principal Engineer, Grid Dynamics <http://www.griddynamics.com> <mkhlud...@griddynamics.com>
