At the moment Solr Search Engine indexes Content field data in multivalued fields. This means values from different translations are stored in the same document, in multivalued fields. Because of this no sorting support on field data (Field sort clause) is implemented. Also, having all languages indexed in the same document means relevancy calculation won't be correct.
Using Solr's multicore/sharding capability to index Content field data per language would enable implementation of Field sort clause while keeping correct document/term statistics per language.
Search would have to target all/specific cores/shards, depending on the set of languages that need to be searched. Grouping should be used to resolve matches of the same Content (but different language) from the different cores/shards.
Implement multicore/sharded prototype of the Solr Search Engine, supporting:
- targeting specific languages (aka field filters parameter)
- Field sort clause