I am trying to find how the Solr union compares with respect to Lucene compounds. In particular, if the Lucene connection uses any filter cache during the JOIN operation. I looked at the code and it seems that QParser has a cache link, but I'm not sure if this is a filter cache. If anyone has any experience with this, please share or tell me how I can find it.
Solr join wiki claims
Fields or other properties of documents that are joined "from" are not available for use in processing the result set "to" documents (that is: you cannot return fields to "from" documents as if they were a multi-valued field in "to" documents )
I find it difficult to understand the above limitation of the solr connection, does this mean that, unlike traditional RDMS connections, which can have columns from the TO and FROM fields, solr connections will only have fields from the TO documents? Do I understand correctly? If so, why is this a limitation?
In addition, there is some difference regarding scoring and what the wiki says
The Join request creates permanent points for all documents that match - the estimates calculated by the sub-request for "from" documents are not available for use in counting "to" documents
Does this mean that there is no main request in the subquery? If so, why was the solo touch taken in this way?
If there are any other differences to consider when switching from Lucene join to Solr, please share.