: This approach works (I do a similar thing using solr), but you have to be : careful as BooleanQuery.TooManyClauses exception can be thrown depending where : you use the wild card. It should be fine in the case you described however.
You'll never get a TooManyClauses from a prefix or range query in Solr, it uses Filters instead of the default query classes. : > Why not just use the whole path as the unique identifying token for a given : > node on the hierarchy? That way, you don't need to map nodes to unique : > numbers, just use a prefix query. personally, i don't create a numeric->text mapping just to do this ... i already have the numeric mappings in my data, so it's *easier* for me to use the numericIds in Solr (one really key reason to do things this way is that i can change the name of a topic without reindexing every document mapped to that topic) Term queries should also be a little faster then prefix queries, but i won't swear to that. If you do go the prefix query route, make sure to leave on a trailing marker character so you don't run the risk of one Topic name being the prefix of a sister topic name... taxonomy:Place/NorthAmerica/USA/Washington* will match Place/NorthAmerica/USA/Washington/Seattle/ and Place/NorthAmerica/USA/WashingtonDC/ ... so use taxonomy:Place/NorthAmerica/USA/Washington/* -Hoss