Clucene was way too slow for body text, more than 1GB. I had my own header parser in C++ (though you can do that in Python easily). I'm trying again on that 80

Clucene was way too slow for body text, more than 1GB. I had my own header parser in C++ (though you can do that in Python easily). I'm trying again on that 800GB with KISS DB (append-only hashtable), and Elasticsearch. Doesn't matter if GPL because it's a website.