Searching is now powered by solr on the blog.
I'm pretty happy with how this came out and I do think solr could be useful for more things.
I'm curious how the highlighting fragments stuff will work with PDFs. I haven't really tested other documents types, currently it's just raw markdown blog posts that have been exported to json.
You can now search for:
purely basic
"purely basic"
"purely basic"~1
These use the solr syntax for searching. The coolest feature is the fuzzy searching and being able to control the edit distance.
I think this could be useful for data entry screens but there is a worry that if solr is down then the application could hang. It's almost like I need a way to make sure solr is up before querying it.
While adding the search on the blog, it required fixing my JSON.QUERY routine to handle quoted keys. I also had to remember that my JSON.PARSE routine requires the raw json to be in one line. I really need to document these things as it was a learning experience to figure out what was wrong.